H. Russell Bernard, Clarence C. Gravlee-handbook Of Methods In Cultural Anthropology-rowman & Littlefield Publishers (2014).pdf

Handbook of Methods in Cultural Anthropology

14_107-Bernard.indb i

6/9/14 10:02 AM

14_107-Bernard.indb ii

6/9/14 10:02 AM

Handbook of Methods in Cultural Anthropology Second Edition

Edited by H. Russell Bernard and Clarence C. Gravlee

ROWMAN & LITTLEFIELD Lanham • Boulder • New York • London

14_107-Bernard.indb iii

6/9/14 10:02 AM

Published by Rowman & Littlefield A wholly owned subsidiary of The Rowman & Littlefield Publishing Group, Inc. 4501 Forbes Boulevard, Suite 200, Lanham, Maryland 20706 www.rowman.com 16 Carlisle Street, London W1D 3 BT, United Kingdom Copyright © 2015 by Rowman & Littlefield First edition copyright © 1998 by AltaMira Press All rights reserved. No part of this book may be reproduced in any form or by any electronic or mechanical means, including information storage and retrieval systems, without written permission from the publisher, except by a reviewer who may quote passages in a review. British Library Cataloguing in Publication Information Available Library of Congress Cataloging-in-Publication Data Handbook of methods in cultural anthropology / edited by H. Russell Bernard and Clarence C. Gravlee. — Second edition. pages cm Includes bibliographical references and index. ISBN 978-0-7591-2070-9 (cloth : alk. paper) — ISBN 978-0-7591-2071-6 (pbk. : alk. paper) — ISBN 978-0-7591-2072-3 (electronic) 1. Ethnology—Methodology. I. Bernard, H. Russell (Harvey Russell), 1940– GN345.H37 2015 305.8001—dc23 2014007881

™ The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, ANSI/NISO Z39.48-1992. Printed in the United States of America

14_107-Bernard.indb iv

6/9/14 10:02 AM

Contents

Preface Introduction: On Method and Methods in Anthropology H. Russell Bernard and Clarence C. Gravlee

vii 1

PART I. PERSPECTIVES

1

Epistemology: The Nature and Validation of Knowledge

21

Michael Schnegg

2

In Search of Meaningful Methods

55

James W. Fernandez and Michael Herzfeld

3

Research Design and Research Strategies

97

Jeffrey C. Johnson and Daniel J. Hruschka

4

Ethics Carolyn Fluehr-Lobban

131

5

Feminist Methods Christine Ward Gailey

151

Participatory Methods and Community-Based Collaborations

185

6

Stephen L. Schensul, Jean J. Schensul, Merrill Singer, Margaret Weeks, and Marie Brault PART II. ACQUIRING INFORMATION

7

Sampling and Selecting Participants in Field Research Greg Guest

215

8

Participant Observation Kathleen Musante (DeWalt)

251

Behavioral Observation

293

9

Raymond Hames and Michael Paolisso

10

Person-Centered Interviewing and Observation

313

Robert I. Levy and Douglas W. Hollan v

14_107-Bernard.indb v

6/9/14 10:02 AM

vi

Contents

11 12

Structured Interviewing and Questionnaire Construction Susan C. Weller

343

Discourse-Centered Methods

391

Brenda Farnell and Laura R. Graham

13

Visual Anthropology

439

Fadwa El Guindi

14

Ethnography of Online Cultures

465

Jeffrey G. Snodgrass

15

Social Survey Methods

497

William W. Dressler and Kathryn S. Oths PART III. INTERPRETING INFORMATION

16

Reasoning with Numbers

519

W. Penn Handwerker and Stephen P. Borgatti

17

Text Analysis

533

Amber Wutich, Gery Ryan, and H. Russell Bernard

18

Cross-Cultural Research

561

Carol R. Ember, Melvin Ember, and Peter N. Peregrine

19

Geospatial Analysis

601

Eduardo S. Brondizio and Tracy Van Holt

20

Social Network Analysis

631

Christopher McCarty and José Luis Molina PART IV. APPLYING AND PRESENTING INFORMATION

21

Theories and Methods in Applied Anthropology

661

Robert T. Trotter, II, Jean J. Schensul, and Kristin M. Kostick

22

Presenting Anthropology to Diverse Audiences

695

Conrad Phillip Kottak

23

Public Anthropology

719

Thomas Hylland Eriksen

14_107-Bernard.indb vi

Author Index

735

Subject Index

763

6/9/14 10:02 AM

Preface

A lot has changed in anthropology since the earlier edition of this handbook was published in 1998. (We discuss those changes in our introduction to the book, which follows.) But one thing has not changed: In this second edition, as in the first, our goal has been a handbook that is useful to academic anthropologists and practicing anthropologists; to interpretivists and positivists; to idealists and materialists. The great epistemology debate that has so long pervaded the social sciences remains as strong as ever. Each side claims support from an indisputable observation: On the one hand, people construct their own realities and the process is dynamic, ever-changing; on the other, there are regularities in human behavior and human thought. While rhetorical energy is spent arguing that (1) the first fact renders impossible the pursuit of the second or that (2) the second fact renders irrelevant our worrying about the first, working scholars of all persuasions are out there doing empirical research and drawing on the same awesomely large kit of tools. As the toolkit has grown, so, too, has the handbook. Nine chapters are completely new— commissioned for this edition—and the remaining 14 have been thoroughly revised and updated to reflect changes in the discipline. We hope the second edition, like the first, will prove to be a valuable resource for anthropologists (and other social scientists) at all levels. The chapters provide a thorough introduction to the diversity of methods available to today’s fieldworker and are written to be accessible to students and experienced researchers alike. We expect that this edition of the handbook will find a place in many graduate courses on research methods in anthropology, as the first edition did. At the same time, the 23 chapters provide the depth and nuance that make the book a true handbook—a regularly thumbed resource on the shelf of veteran anthropologists. We asked contributors to place the methods in historical context and to serve as honest brokers of the advantages and disadvantages of complementary (and sometimes competing) approaches, even if that meant pointing out the limitations of approaches the authors favor. The result is a handbook that can be read at multiple levels and that will yield new insights as readers—whether beginning graduate students or seasoned ethnographers—return to it over and over again. Projects like this one involve a lot of people, a lot of time—and a lot of patience. We are grateful to Leanne Silverman at AltaMira Press for her care and attention in bringing this book to print. Our thanks also to the authors of the chapters and to the following reviewers: Frances K. Barg, Tom Boellstorff, Robert Borofsky, Ivan Brady, vii

14_107-Bernard.indb vii

6/9/14 10:02 AM

viii

Preface

Caroline Brettell, Alexandra Brewis Slade, Garry E. Chick, Colleen Ballerino Cohen, Alan Page Fiske, Linda C. Garro, Edward Gonzalez-Tennant, Carla Guerrón Montero, Craig Hadley, Daniel Hruschka, Eric C. Jones, Jeremy Koster, Rebecca Lester, Carole McGranahan, Miriam Meyerhoff, Karen Nakamura, Anthony Onwuegbuzie, John van Willigen, and Linda M. Whiteford. Finally, with 23 chapters and 41 authors, the task of managing and copy editing would seem Sisyphean to most of us, but not to Carole Bernard. Her good humor, patience, and competence are simply inspiring.

14_107-Bernard.indb viii

6/9/14 10:02 AM

Introduction On Method and Methods in Anthropology H. Russell Bernard and Clarence C. Gravlee

Just as when the first edition of this handbook was published in 1998, method is still about choice—the choice of taking a Verstehen or a positivist approach; the choice of collecting data by participant observation or in the archives, by direct observation or by interviewing; the choice of making quantitative measurements or collecting oral, written, or visual text. The authors in this handbook deal with all these choices, and more. At the same time, the passage of years since the first edition has brought along several major developments, including the rise of “mixed methods,” the emergence of new technologies, and an evolution in our thinking about ethnographic writing. The contributors to this second edition tackle these changes as well. We invite you to explore the history and scope of methods in cultural anthropology, as well as the changes in research methods (and this book), in the pages that follow. ON METHODS IN ANTHROPOLOGY

New methods develop within particular disciplines, but any method that seems useful will get picked up and tried out, sooner or later, across the sciences. Anthropologists have been prodigious consumers and adapters of research methods, but have made important contributions to the big social science toolkit as well. Cluster analysis, for example, was developed first by anthropologists (Driver and Kroeber 1932). It was quickly advanced by psychologists (Tryon 1939; Zubin 1938) and is now used across all the sciences. The consensus model of culture was developed in the 1980s by two anthropologists and a psychologist (Romney et al. 1986) and is used today in nursing (Shiu et al. 2010), water management (Mathevet et al. 2011), civil engineering (Kim et al. 2008), and psychology (Lynch and Holmes 2011; Torres 2009).

The False Qual–Quant Divide There has always been a certain tension between those who would make anthropology a quantitative science and those whose goal it is to produce documents that convey the richness—indeed, the uniqueness—of human thought and experience. Alfred Kroeber’s son, Karl, recalls his father characterizing anthropology as the most humanistic of the sciences, while also the most scientific of the humanities (Kroeber 2003, 144)—a theme that Eric Wolf echoed in his wonderful overview of the field (1964, 88). For generations, students of cultural anthropology were asked early in their training to take a stand for interpretivism or positivism, humanism or science, qualitative or quantitative research. The mixed methods movement across the social sciences is this generation’s response to this false choice. 1

14_107-Bernard.indb 1

6/9/14 10:02 AM

2

H. Russell Bernard and Clarence C. Gravlee

Indeed, readers of this handbook will find no support for a polarized vision of method. Instead, they will find scholars laying out the methods they use in practicing their craft—a craft rooted, for every author in this book, in one of the most essentially empirical traditions in all of science: participant observation fieldwork. Some authors are identified with interpretivist methods, some with quantitative methods for the collection and analysis of data, but none dismiss humanism or science and none ask their readers to choose once and for all between expressing their findings in words or in numbers.

Romancing the Methods John Whiting and some of his fellow graduate students at Yale during the 1930s asked about having a seminar on methods. “Leslie Spier informed us disdainfully,” recalled Whiting, “that this was a subject to discuss casually at breakfast and was not worthy subject matter for a seminar” (Whiting 1982, 156). Try quoting Whiting at a convention of anthropologists. Lots of people will chime in with a similar story or a complaint about having to learn methods in the field. It’s all well and good for anthropologists to romanticize fieldwork—vulcanologists and oceanographers do it, too—particularly for fieldwork in places that take several days to get to, where the local language has no literary tradition, and where the chances of coming down with a serious illness are real. Social research really is harder to do in some places than in others. But that doesn’t mean our methods should remain mysterious. There is, in fact, a long, noble tradition of concern with research methods in anthropology—quantitative and nonquantitative, humanistic and scientific. Kathleen Musante (aka DeWalt) quotes at length in Chapter 9 what is surely one of the most-cited early discussions of methods in anthropology: Malinowski’s introduction to Argonauts of the Western Pacific (1922). It is justly famous because, as Musante says, it established the importance of long-term participant observation as a strategic method for field research on other cultures. No peering off the veranda at the natives for Malinowski. Participant observation is an important method in anthropology but, as Musante points out, it is one of many methods used in fieldwork. By the time Malinowski went to the Trobriands, Notes and Queries on Anthropology—the fieldwork manual produced by the Royal Anthropological Institute (RAI) of Great Britain and Ireland— was in its fourth edition (the first came out in 1874). The sixth (and last) edition was published in 1951 and was reprinted five times until 1971. That final edition was edited by Brenda Seligmann and “a committee of the Royal Anthropological Institute” that included the most illustrious British anthropologists of the day including E. E. Evans-Pritchard, Daryl Forde, Raymond Firth, Meyer Fortes, and W. E. Le Gros Clark. Strip away the quaint language and the vestiges of colonialism—“a sporting rifle and a shotgun are . . . of great assistance in many districts where the natives may welcome extra meat in the shape of game killed by their visitor” (RAI 1951, 29)—and the book is still full of useful, late-model advice about how to conduct a census and what questions to ask about sexual orientation, infanticide, food production, warfare, art. The book is just a treasure—must reading for anyone interested in learning about field methods.

14_107-Bernard.indb 2

6/9/14 10:02 AM

Introduction

3

Anthropology—The Original Research Methods Discipline In the 1920s, leading sociologists were concerned with moving their discipline away from an emphasis on social reform—away from the study of what ought to be and toward the study of what is. If the public were ever to trust social science, said Carl Taylor, then the emphasis had to be on “exact and quantitative expressions and measurements” (Taylor 1920, 735). This, he said, required “technologies which will reduce observations to a comparative basis” (p. 753). The technology of choice, said Taylor, was the social survey, a method dating at least to John Howard’s monumental, comparative study of prisons (1973 [1792]). Taylor’s idea of what a survey should be was much broader than just questionnaires. “The survey method,” he said, “is nothing whatever but the recognized and accepted comparative method of all science” and he concluded that “what . . . surveys can do and have done in the field of anthropology and ethnology, they can do and probably are destined to do for any body of knowledge or field of research to which they are applied (1920, 752–53; emphasis added). Taylor singled out the systematic study of vision, hearing, and pain that Charles and Brenda Seligmann and others (1911) had done on the Veddas of Sri Lanka. (Charles Seligmann was an ethnologist and physician.) Their 422-page ethnographic account covered family life, religion, the arts, property, and inheritance—and an 18-page report of the results of some psychological tests that they had used in their study of Vedda senses. Some of those tests had been devised by W. H. R. Rivers, an experimental psychologist who became interested in anthropology in 1899 when he was invited to join the Torres Straits expedition and saw the opportunity to do comparative studies of non-Western people (Tooker 1997, xiv). Rivers, of course, developed the genealogical method—highly detailed, ego-centered graphs for organizing kinship data. The genealogical tables he produced in his study of the Murray Islanders were singled out by Taylor as an example of “as perfect a scientific compilation as could well be imagined” (Taylor 1920, 753. See Rivers’s work in Volume VI of Haddon 1901–1935). These works by anthropologists, said Taylor, were examples of research to which all social science could aspire. Anthropologists today extend this tradition using methods from network analysis to visualize complex kinship structures (Alvard 2011; White and Jorion 1992). Rivers continued his work in anthropology and his development of the genealogical method with his research on the Todas of India (Rivers 1906). He developed what he called the “method of indirect corroboration.” This method involves “obtaining the same information first in an abstract form and then by means of a number of concrete instances” (p. 11). Here is Rivers explaining the method with reference to his study of the laws of property inheritance: I first obtained an account of what was done in the abstract—of the laws governing the inheritance of houses, the division of the buffaloes and other property among the children, &c. Next I gave a number of hypothetical concrete instances; I took cases of men with so many children and so many buffaloes, and repeating the cases I found that my informant gave answers which were consistent not only with one another but also with the abstract regulations previously given. Finally I took real persons and inquired into

14_107-Bernard.indb 3

6/9/14 10:02 AM

4

H. Russell Bernard and Clarence C. Gravlee what had actually happened when A or B died, and again obtained a body of information consistent in itself and agreeing with that already obtained. (p. 11)

Rivers discussed his selection of informants, how he came to know that one of his informants had lied to him, the pros and cons of paying informants for their time, the need for getting information from many informants rather than just from a trusted few, and the importance of using the native language in field research. Taylor must also have known about Lewis Henry Morgan’s study, Systems of Consanguinity and Affinity of the Human Family (1870). It was a massive, cross-cultural survey of kinship systems. Morgan collected a lot of the data on various Indian tribes himself, but he also sent questionnaires to missionaries and Indian agents. And Taylor surely also knew about Sir Edward Burnett Tylor’s key contribution to the literature on cross-cultural surveys (see Tylor 1889). Contrary to popular wisdom, then, anthropologists have been keen methodologists from the earliest days of the discipline. Unlike sociologists, however, anthropologists studied small, remote groups of people. “These groups,” observed Robert Lynd in 1939, “were ‘primitive,’ according to Western European standards, and therefore the older social sciences did not much care what anthropology did with them” (p. 14). The point is, by the time the first Handbook of Method in Cultural Anthropology was published in 1970, the concern for methods in anthropology, including survey methods, was already a hundred years old. That volume, edited by Raoul Naroll and Ronald Cohen, was an enormous compilation—1,000 pages and 49 chapters by 46 authors (Naroll and Cohen wrote six of the chapters and participated in several others), including five that were reprinted from journal articles (Naroll and Cohen 1970). The chapters in that handbook, as well as all the chapters on methods in Anthropology Today (Kroeber 1953), and in the Handbook of Social and Cultural Anthropology (Honigmann 1973), are as useful today as they were when they first appeared. The pioneering textbooks by Pertti Pelto (1970; and see Pelto and Pelto 1978), by John Brim and David Spain (1974), by Allen Johnson (1978), by James Spradley (1979, 1980; Spradley and McCurdy 1972), and by Michael Agar (1996 [1980]) contain a wealth of information and wisdom. In writing the first edition of his own textbook, Bernard (1988) depended heavily on the work of all these predecessors and on the work of colleagues across the social sciences. Since then, the wealth of published wisdom about the real how-to of ethnography has grown and grown (Emerson et al. 2011; LeCompte and Schensul 2013; McCurdy et al. 2005; Werner and Schoeple 1987; Wolcott 1994, 2005, 2010). Methods, in fact, are us.

Methods Belong to All of Us And yet, they aren’t ours: A nice thing about methods is that disciplines cannot own them. Just as it was never true that only sociologists did surveys, it was never true that only anthropologists did participant observation fieldwork. We’ve heard talk at anthropology conventions about how the discipline somehow “lost” the method of participant observation. We could not disagree more. Anthropologists continue to make consistent use of participant observation fieldwork, but

14_107-Bernard.indb 4

6/9/14 10:02 AM

Introduction

5

we did not invent that method by ourselves, and it was never ours to lose. Sociologist Beatrice Webb was doing participant observation—complete with real note taking and informant interviewing—in the 1880s, and she wrote trenchantly about the method in her 1926 memoir (Webb 1926). Just about then, the long tradition in sociology of urban ethnography began at the University of Chicago under the direction of Robert Park (one of the many participant observers trained by Park was his son-in-law, Robert Redfield). That tradition continues today in the pages of the journal Contemporary Ethnography, which began in 1972 under the title Urban Life and Culture. Participant observation today is absolutely ubiquitous in the social sciences. It has been used in recent years by political scientists (Schatz 2009; Wedeen 2008, 2010), social psychologists (Chan 2013; Ezzell 2012), students of consumer behavior (Cronin and McCarthy 2011) and management (Griesbach and Grand 2013; Kenny 2012), researchers in nursing (Higham and Davies 2013; Jessri et al. 2013), education (Davidson and Hawe 2012; Riggs and Due 2011), social work (Cleaveland 2010), and water management (Birkenholtz 2013) as well as by legions of sociologists. Among the salutary results of this disciplinary free-for-all is a continually growing body of literature, including a lot by anthropologists, about participant observation itself. There are highly focused studies, full of practical advice, and there are poignant discussions of the overall experience of fieldwork (Agar 1996 [1980]; Smith and Kornblum 1996; Wolcott 2005). The boundaries between the disciplines remain strong, but those boundaries are no longer about methods—if they ever were—or even about content. Anthropologists today are more likely to study investment bankers on Wall Street (Ho 2009) or openair markets in Italy (Black 2012) or urban gangs (Phillips 2012) than they are to study isolated tribal peoples (Marcus 2008). Today, the differences within anthropology and sociology with regard to methods are more important than the differences between the social sciences. There is an irreducible difference, for example, between those of us for whom the first principle of inquiry is that reality is constructed uniquely by each person and those of us who start from the principle that external reality awaits our discovery through a series of approximations. There is also an important (though not incompatible) difference between those of us who seek to understand human phenomena in relation to differences in beliefs and values and those of us who seek to explain human thought and behavior as the consequence of external forces. Whatever our epistemological differences, however, the actual methods by which we collect and analyze our data belong to everyone across the social sciences.

Anthropology: The Humanistic, Positivistic, Interpretive Science Anthropology was built by empiricists, including many who understood that reducing people to words was no better than reducing them to numbers and who did both with aplomb. Franz Boas took his Ph.D. in physics from the University of Kiel, with mathematics and geography as his minors. When he turned his attention to anthropology, he advocated a historical, particularizing approach, but he did not abandon numbers. Nor did most of his students. Alfred Kroeber, for example, analyzed measurements on 300 years of data on skirt length, waist height, and depth of décolletage to look for long-term patterns in style (Richardson and Kroeber 1940).

14_107-Bernard.indb 5

6/9/14 10:02 AM

6

H. Russell Bernard and Clarence C. Gravlee

Robert Lowie, Clark Wissler, Edward Sapir, Margaret Mead, Ruth Benedict, and Elsie Clews Parsons considered themselves scientists. One of Boas’s students, Paul Radin, famously rejected his mentor’s scientific bent. He accused Boas of being naturwissenschaftlich eingestellt, or science minded, of treating ethnology as a branch of natural science and named his cofrêres—Sapir, Kroeber, Mead—as examples of the bad things that happen to cultural anthropologists who follow the path of quantification (Radin 1933, 10). But even Radin was an arch empiricist. His passion against quantification was surpassed by his commitment to the continual collection of texts. It was Radin who made it clear for all time that, while theories come and go, original texts, in our informants’ own languages, are available for every generation to analyze anew. Inspired by Boas’s efforts with George Hunt (a Kwakiutl), Radin handed Sam Blowsnake (a Winnebago) a pad and a pencil—an act that produced one of anthropology’s most famous texts, Crashing Thunder: The Autobiography of a Winnebago Indian—and established a model for emically oriented ethnography (Blowsnake 1920).1 The distinction between quantitative and qualitative is often used as cover for talking about the difference between science and humanism, but that’s a mistake. Lots of scientists do their work without numbers, and many scientists whose work is highly quantitative consider themselves to be humanists. Neither quantitative nor qualitative researchers have the exclusive right to strive for objectivity; neither humanists nor scientists have a patent on compassion; and empiricism is as much the legacy of interpretivists and idealists as it is of positivists and materialists. Humanism is often used as a synonym for humanitarian or compassionate values and a commitment to the amelioration of suffering. The myth that science is the absence of these values is truly pernicious. We must reject a culture that equates objectivity with being cold. Counting the dead accurately in Iraq is one way—not the only way—to preserve outrage. We need more, not less, science—lots and lots more—and more humanistically informed science, if we are to contribute more to the amelioration of suffering and the weakening of false ideologies—racism, sexism, ethnic nationalism—in the world. Humanism sometimes means a commitment to subjectivity—that is, to using our own feelings, values, and beliefs to achieve insight into the nature of human experience. This kind of humanism is, of course, the foundation of many clinical disciplines as well as the foundation of participant observation ethnography (see Berg and Smith [1985] for a review of clinical methods in social research). It is not something apart from social science. It is a method used by idealists and materialists alike. Humanism sometimes means an abiding appreciation of and search for the unique— the unique in human experience and the unique in culture. If the rest of social science was the search for regularities, then anthropology was the search for exceptions. It is a truism that one cannot fully know someone else’s life without living that person’s life. Giving birth, surviving hand-to-hand combat, living with AIDS . . . in some way, all experience is surely unique. But just as surely, there are commonalities of experience. To write a story about the thrill or the pain of a successful or failed border crossing, a successful or failed job hunt, a winning or losing struggle with illness—or to write someone else’s story for them, as an ethnographer might do—these are not activities

14_107-Bernard.indb 6

6/9/14 10:02 AM

Introduction

7

opposed to a natural science of experience. They are the activities of a natural science of experience, whether or not they are done in service to explanation or understanding, to nomothetic or idiographic goals. Humanism is sometimes pitted against positivism, which is said to be linked to support for whatever power relations happen to be in place. Historically, though, positivism was linked to the most critical stance. The Subjection of Women (1869), by John Stuart Mill was a radical work in its day, advocating full equality for women. Adolphe Quételet, the Belgian astronomer whose study of demography and criminology carried the audacious title Social Physics (1969 [1835]), was a committed social reformer. The legacy of positivism as a vehicle for social activism is clear in Jane Addams’s work with destitute immigrants at Chicago’s Hull House (1926), in Sidney and Beatrice Webb’s attack on the British medical system (Webb and Webb 1910), in Charles Booth’s account of the conditions under which the poor lived in London (1902), and in Florence Nightingale’s (1871) assessment of death rates in maternity hospitals (see McDonald [1993, 1994] on Nightingale’s long-ignored work). The central position of positivism as a philosophy of knowledge is that experience is the foundation of knowledge. We record what we experience visually, auditorily, and emotionally. The quality of the recording, then, becomes the key to knowledge. Can we, in fact, record what others experience? Yes, of course we can. Are there pitfalls in doing so? Yes, of course there are. For some social researchers, these pitfalls are evidence of natural limits to social science; for others, they are a challenge to extend the current limits by improving measurement. Whether they rally to the idealist or the materialist flag, all social scientists can be humanists by any or all of these definitions. When insight and understanding are achieved, we may call the result knowledge. The fact that knowledge is tentative is something we all learn to live with. The practice that we associate today with the positivist perspective in social science, however, is not the positivism of Auguste Comte, of Adolphe Quételet, of John Stuart Mill; nor is it even the logical positivism of the Vienna Circle. None of these is the tradition that so many today love to hate. That honor belongs to what Christopher Bryant (1985, 137) called “instrumental positivism.” In his 1929 presidential address to the American Sociological Society, William F. Ogburn laid out the rules. In turning sociology into a science, he said, “it will be necessary to crush out emotion.” Further, “it will be desirable to taboo ethics and values (except in choosing problems); and it will be inevitable that we shall have to spend most of our time doing hard, dull, tedious, and routine tasks” (Ogburn 1930, 10). Eventually, he said, there would be no need for a separate field of statistics because “all sociologists will be statisticians” (p. 6). This kind of rhetoric just begged to be reviled. There were challenges to Ogburn’s prescription, but, as Oberschall (1972, 244) concluded, it was Ogburn’s commitment to value-free science and to statistics that won the day. “It is certainly desirable to be precise,” said Robert Redfield (1948, 148), “but it is quite as needful to be precise about something worth knowing.” We are all free to identify ourselves as humanists or as positivists, but it’s much more fun to be both. The scientific component of anthropology demands that we ask whether our measurements are meaningful, but the humanistic component forces us to ask if we are pursuing worthwhile ends and doing so with worthwhile means.

14_107-Bernard.indb 7

6/9/14 10:02 AM

8

H. Russell Bernard and Clarence C. Gravlee

In the end, the tension between science and humanism is wrought by the need to answer practical questions with evidence and the need to understand ourselves—that is, the need to measure carefully and the need to listen hard.

Permanent Methodological Eclecticism Anthropology has always been an eclectic discipline with regard to methods, but anthropologists increasingly adopted the full range of social research methods after 1970. This was not an accident. In 1970, most Ph.D.-level anthropologists went into academic jobs; by 1975, most did not, and nonacademic employment has been the norm ever since. A generation earlier, in 1950, just 22 Ph.D. degrees were awarded by U.S. universities in anthropology. In 1957, the Soviet Union launched the first orbiting satellite— Sputnik. The arms race that followed, with its MAD (mutually assured destruction) strategy, produced billions of dollars for expansion of U.S. universities. And this wasn’t just for math and the natural sciences. All those engineers and physicists needed courses in English, philosophy, history, and the social sciences. Combined departments of sociology and anthropology separated, and Ph.D.-granting anthropology programs proliferated. For example, when Bernard began graduate study at the University of Illinois in 1961, anthropology had recently separated from sociology. Four years later, the department awarded its first Ph.D. degree. The expansion was irrepressible. By 1970, the number of anthropology Ph.D.s had jumped to a heady 195, but that was just the beginning. By 1974, all those new Ph.D.-granting departments were filled with assistant professors training thousands of graduate students—and awarding over 400 Ph.D. degrees per year. With the expansion complete, all those assistant professors were in place and there were no new jobs in the academy (D’Andrade et al. 1975; Givens and Jablonski 1996). The attraction of anthropology for new students, however, remained undiminished. The number of Ph.D. degrees in anthropology awarded annually in the United States averaged about 400 between 1974 and 1992. Since 2009, the number has been over 500 per year. At a rough count, about 17,000 Ph.D. anthropologists were produced from 1970 to 2013 (Fiske et al. 2010; Rudd et al. 2008). Demography being more-or-less destiny, those who had participated in the great 1960s expansion began to retire in the mid-1980s. From 1985 to 1994, an average of 331 academic jobs were listed in the Anthropology Newsletter (Givens and Jablonski 1996). But with all those anthropologists from previous years still applying for academic jobs, the openings were (and remain) far fewer than the number of available candidates. By 2008, the online Career Center of the American Anthropological Association listed just 199 jobs (Terry-Sharp 2009). The thousands of Ph.D.s in anthropology who were not in academe were not out of work. On the contrary. During the early, tough years of the 1970s and 1980s, young Ph.D. anthropologists from all four fields knocked on, and eventually opened, many doors. They rewrote their résumés and competed successfully for jobs in medical schools, business schools, schools of agriculture, and other parts of the university. They also competed successfully for jobs in industry and government. They joined consulting firms and opened their own firms, specializing in international market research,

14_107-Bernard.indb 8

6/9/14 10:02 AM

Introduction

9

in training executives for overseas assignments, in rapid rural assessment for health care delivery programs, in developing ecotourism—the list is as big as the hunger and imagination of the pioneers themselves. By 1986, for the first time in the history of the discipline, more anthropologists, including M.A.s and Ph.D.s, were employed in nonacademic jobs than in academe (Fluehr-Lobban 1991, 5), and in 1993, the jobless rate for anthropologists was 1.6% (Givens and Jablonski 1996). In the 1960s, a colleague in chemistry told Bernard that it was hard to keep good doctoral students in academe. The best students, he said, were being lured away by better pay in industry. At that time, it was rare to see anything but an academic affiliation on the name tags of those attending meetings of the American Anthropological Association. Today, many anthropology students, especially those in cultural anthropology, enter graduate school planning for a nonacademic career, and the name tags at the AAA meetings show affiliations from government and industry. The U.S. Bureau of Labor Statistics estimates that there were 6,100 jobs for anthropologists and archaeologists in 2010, including jobs that require only a master’s degree, and that the number of those jobs would grow by 21% by 2020. On a percentage basis, this makes anthropology one of the fastest-growing job sectors in the United States. Given the production of anthropologists in the academy, however, the competition for those jobs will be strong (Bureau of Labor Statistics 2013). Anthropologists who are skilled in research methods will, as they have in the past, continue to compete successfully for the thousands of jobs in government and industry that are not advertised specifically for anthropologists. The 1970–2000 drought in academic jobs was as severe in sociology as it was in anthropology, but in sociology, the effects were quite different. As anthropologists competed for nonacademic jobs, they were forced to be more explicit about their methods. Program administrators were accustomed to hiring social scientists whose methods and products were known commodities. It was the anthropologists who had to make the adjustment, not the sociologists, psychologists, and economists. By and large, the result was a synthesis of qualitative and quantitative approaches—a synthesis that we see accelerating today under the rubric of “mixed methods” (Alise and Teddlie 2010). In the absence of, say, a few thousand new academic jobs for anthropologists, the continued move toward sophisticated methodological eclecticism is permanent. OVERVIEW OF THE HANDBOOK

This edition of the handbook, like the first, is organized into four parts that span the chain of research. Part I, Perspectives, includes chapters on epistemology (Schnegg), ethics (Fluehr-Lobban), and a range of approaches for defining the goals of anthropological research and designing a project to meet those goals. The breadth of this range reflects the enormous range of the field: interpretive approaches (Fernandez and Herzfeld); feminist approaches (Gailey); community-based participatory research (Schensul et al.); and systematic research designs in the positivist tradition (Johnson and Hruschka). All of these chapters also reflect the major theme of this handbook: All the methods belong to all of us. Part II, Acquiring Information, focuses on methods for data collection. Here we are reminded that anthropologists draw on—indeed, have always drawn on—the

14_107-Bernard.indb 9

6/9/14 10:02 AM

10

H. Russell Bernard and Clarence C. Gravlee

whole toolkit of social science. There’s a chapter on participant observation, to be sure (Musante, aka DeWalt), but also chapters on sampling (Guest), unstructured interviewing methods (Levy and Hollan), structured interviewing methods (Weller), discourse-centered methods (Farnell and Graham), observational methods for studying behavior (Hames and Paolisso), visual anthropology (El Guindi), ethnography of online communities (Snodgrass), and social survey research (Dressler and Oths). The chapter on sampling is new to this edition, as are the chapters on ethnography of online communities and on social survey research. This breadth of approaches carries over to Part III, Interpreting Information, with a focus on data analysis. The five chapters in this part cover the hows and whys of analyzing both numeric (Handwerker and Borgatti) and textual (Wutich et al.) data, testing comparative hypotheses on cross-cultural data sets (Ember and Peregrine), analyzing and visualizing data on social networks (McCarty and Molina), and integrating geospatial analysis into ethnographic research (Brondizio and Van Holt). (The chapters on network analysis and on geospatial analysis are new to this edition.) In some ways, this part may be more challenging but also more necessary: We anthropologists have always been consummate data collectors, but our training historically gives short shrift to data analysis. That’s changing, though, and the chapters in this part reflect that change. An important consequence of this new appreciation of the range of analytic approaches available to us is that it can spark creativity about the kinds of questions we can pursue in the first place. Finally, Part IV, Applying and Presenting Anthropology, considers what happens after all the data have been collected and analyzed—what can we do and how can we bring our work to diverse audiences? The three chapters in this part (Trotter et al., Kottak, and Eriksen) draw on themes from earlier parts of the book—all the methods are relevant here—and lead us to think about what’s at stake. What motivates our work? Who is our audience, and how do we reach them? What impact do we want to have? The answers, of course, are tied up with the ethical and epistemological issues addressed in the first part of the book, so this part is perhaps less a conclusion than an invitation to revisit the beginning.

New to the Second Edition All four parts include new chapters that were commissioned for the second edition, and the remaining chapters have been updated and revised. The additions and changes reflect three major developments since the first edition: (1) the rise of a “mixed methods” movement across the social sciences; (2) the emergence of new technologies for gathering, analyzing, and sharing data; and (3) evolving views on the forms and purpose of ethnographic writing. One of the most striking developments since the first edition was published is the explicit emphasis on mixed methods in the social sciences. Before the year 2000, there were only 21 references to mixed methods in the Social Science Citation Index. Since then, there have been more than 3,500. This meteoric rise is not just a fad; it’s a real movement. There is a Journal of Mixed Methods Research, several textbooks on how to do this kind of research (e.g., Creswell and Plano Clark 2011; Greene 2007; Teddlie and Tashakkori 2009), a handbook (Tashakkori and Teddlie 2010), university courses,

14_107-Bernard.indb 10

6/9/14 10:02 AM

Introduction

11

conferences—even a guide to best practices in mixed methods research from the U.S. National Institutes of Health (http://obssr.od.nih.gov/mixed_methods_research/). One might scoff at the idea of “mixed” methods—it implies that methods are otherwise pure and distinct—but we welcome the premise that dividing the toolkit of social science into qualitative and quantitative methods is misleading and counterproductive (Bernard 1996; Gravlee 2011). You can use some of the methods, or you can use all the methods, and the choice should be driven by the research questions in any particular project, not by an ideological commitment to “qualitative” or “quantitative” research. The first edition of the handbook embraced this approach, but it’s even more apparent in the chapters that are new to this edition. Network analysis (McCarty and Molina, Chapter 20), for example, can be applied to words or numbers, and in either case the raw data are usually turned into pictures. Geospatial analysis (Brondizio and Van Holt, Chapter 19) accommodates text, photographs, satellite images, audio recordings, demographic data, ecological information, and more. Both network analysis and geospatial analysis involve the qualitative search for patterns, and both are enriched by statistical techniques to quantify associations or test hypotheses. The integration of all this qualitative and quantitative data and analysis is a reminder that the methods have always been mixed. Social network analysis and geospatial analysis also point to the second development reflected in this new edition: the proliferation of technologies that are changing how anthropologists collect, analyze, and disseminate data and results. As McCarty and Molina (Chapter 20) recount, social network analysis has deep roots in anthropology, stretching back to the work of Manchester anthropologists in the 1940s and 1950s. In subsequent decades, however, anthropologists were slow to adopt and develop network methods, a pattern that McCarty and Molina attribute to technological constraints: “the lack of methods (particularly, the lack of computers) for dealing with the large amount of network data collected during fieldwork stopped its development.” This technological barrier is now gone, and anthropologists have been at the forefront of developing tools, including software, to facilitate the collection, analysis, and visualization of network data in the field. The impact of technology is evident in other areas as well. Computer-assisted qualitative data analysis software (CAQDAS) facilitates the management of text and multimedia data and makes accessible a wider range of qualitative and quantitative analytic methods (Wutich et al., Chapter 17). Mobile computers, smartphones, digital recorders, and voice recognition technology offer new possibilities for collecting data during participant observation (Musante, Chapter 8) or for direct, systematic observation of behavior (Hames and Paolisso, Chapter 9). The declining costs and increasing capabilities of multimedia technology continue to drive the development of visual anthropology, including interactive modes of ethnographic representation, the integration of photography and video in advocacy and applied anthropology, and participatory methods of collecting data in visual form (El Guindi, Chapter 13). And rapid proliferation of the Internet has opened the door to ethnography of online worlds (Snodgrass, Chapter 14). The third development evident in this new edition of the handbook is a debate over the purpose and forms of ethnographic writing. The watershed moment in this

14_107-Bernard.indb 11

6/9/14 10:02 AM

12

H. Russell Bernard and Clarence C. Gravlee

debate—the publication of Writing Culture (Clifford and Marcus 1986)—occurred well before the first edition, but its implications have become clearer with time (see the August 2012 special issue of Cultural Anthropology, marking the volume’s 25th anniversary). One key implication is that the methodological (including epistemological and ethical) dimensions of writing have moved to the center of the discipline. The most extensive treatment of ethnographic writing in this edition of the handbook is Schnegg’s new chapter on epistemology (Chapter 1), which builds on the framework established by the late Thomas Schweizer in the first edition. Schnegg traces the historical roots of two venerable eighteenth-century epistemological traditions—positivism and hermeneutics—and shows that our discipline is firmly rooted in both. He then describes the more recent emergence of radical constructivism and its manifestation as postmodernism in anthropology. For each of these traditions, Schnegg draws out the implications for fieldwork and examines four distinct approaches to ethnographic writing. The methodological dimensions of ethnographic writing also appear in Musante’s chapter on participant observation (Chapter 8), Fernandez and Herzfeld’s chapter on meaningful methods (Chapter 2), and Gailey’s chapter on feminist methods (Chapter 5). In his chapter on presenting anthropology to diverse audiences, Kottak (Chapter 22) addresses writing from a strategic and stylistic point of view, and Eriksen (Chapter 23) discusses writing as a crucial methodological challenge of doing public anthropology. Thus, although there is no separate chapter on ethnographic writing, this theme ties together several others. These three themes—mixed methods, technology, and ethnographic writing—also intersect in interesting ways. One of the implications of the Writing Culture moment, for example, is a shift in our understanding of ethnographic authority and representation. This shift coincides with the growth of participatory approaches (Schensul et al., Chapter 6), which elevate the importance of participants’ voices and encourage active collaboration between researchers and participants in all phases of research, from framing of questions to dissemination and development of interventions (see also Gailey, this volume). Technology potentially advances these goals by increasing access to tools for collecting, interpreting, and sharing qualitative and quantitative information, as in participatory photo and mapping approaches (El Guindi, Chapter 13; Brondizio and Van Holt, Chapter 19; see also Gubrium and Harper 2013). Schnegg (Chapter 1) further suggests that Internet-based technologies have eroded the “gap between field and desk,” as “more and more of our informants are connected to Internet devices, providing new opportunities to contextualize the ethnographer.” These changes, Schnegg says, “make it relatively easy and inexpensive to draw a larger public into debates about the validity of anthropological knowledge.” Thus, as technology opens new possibilities for the collection and analysis of anthropological data— qualitative and quantitative alike—it also refocuses attention on the epistemological and ethical questions that motivated a concern with ethnographic writing. CONCLUSION

The field of cultural anthropology has changed dramatically since the last edition of this handbook. Anthropologists today are as likely to be studying urban street gangs

14_107-Bernard.indb 12

6/9/14 10:02 AM

Introduction

13

or organic farmers as they are to be studying isolated communities. This increase in the scope of our subject matter brings with it the need for training in an increasingly broad range of methods. New career opportunities have opened up, too. We often hear the lament that there are too few jobs for anthropologists, especially for those with the Ph.D. The lament is true if by jobs we mean only tenure-track academic jobs, but this view is too narrow. As Fluehr-Lobban notes in her chapter on ethics, there have been more anthropologists in applied work than in academe since 1986. Some 30 years later, many anthropologists enter graduate school aiming for a career in international development or in public health or in product design rather than in academe. Those students need to be trained broadly in the range of methods reflected in this handbook. We see growing demand for methodological training among academic anthropologists, too. Consider the range of workshops on methods being offered at the annual meeting of the American Anthropological Association today, as compared to 1998 when the first edition of the handbook appeared. Or consider the array of conversations about methods in Anthropology News, in our journals, or in the anthropological blogosphere. These conversations are not only more common but also less polarizing than they were a generation ago. More and more, methods really do belong to all of us. NOTE

1. The first edition of Blowsnake’s autobiography was published in 1920, as part of the University of California’s series on archaeology and ethnology. The book, titled Crashing Thunder: Autobiography of a Winnebago Indian, was published in 1926 by D. Appleton and Co. and was reprinted in 1983 by the University of Nebraska Press. REFERENCES

Addams, J. 1926. Twenty years at Hull-House. New York: Macmillan. Agar, M. 1996 [1980]. The professional stranger: An informal introduction to ethnography, 2nd ed. San Diego: Academic Press. Alise, M. A., and C. Teddlie. 2010. A continuation of the paradigm wars? Prevalence rates of methodological approaches across the social/behavioral sciences. Journal of Mixed Methods Research 4: 103–26. Alvard, M. 2011. Genetic and cultural kinship among the Lamaleran whale hunters. Human Nature 22: 89–107. Berg, D. N., and K. K. Smith, eds. 1985. Exploring clinical methods for social research. Beverly Hills, CA: Sage. Bernard, H. R. 1988. Research methods in cultural anthropology. Newbury Park, CA: Sage. Bernard, H. R. 1996. Qualitative data, quantitative analysis. Cultural Anthropology Methods Journal 8: 9–11. Birkenholtz, T. 2013. “On the network, off the map”: Developing intervillage and intragender differentiation in rural water supply. Environment and Planning D: Society and Space 31: 354–71. Black, R. E. 2012. Porta Palazzo: The anthropology of an Italian market. Philadelphia: University of Pennsylvania Press. Blowsnake, S. 1920. The autobiography of a Winnebago Indian. P. Radin, ed. and tr. Berkeley: University of California Press. (University of California Publications in American Archaeology and Ethnology 1).

14_107-Bernard.indb 13

6/9/14 10:02 AM

14

H. Russell Bernard and Clarence C. Gravlee

Booth, C, ed. 1902. Life and labor of the people of London. New York: Macmillan. Brim, J. A., and D. H. Spain. 1974. Research design in anthropology: Paradigms and pragmatics in the testing of hypotheses. New York: Holt, Rinehart and Winston. Bryant, C. 1985. Positivism in social theory and research. New York: St. Martin’s Press. Bureau of Labor Statistics. 2013. U.S. Department of Labor, Occupational outlook handbook, 2012–13 edition, Anthropologists and Archeologists. http://www.bls.gov/ooh/life-physical -and-social-science/anthropologists-and-archeologists.htm (accessed June 7, 2013). Chan, C. S. C. 2013. Doing ideology amid a crisis: Collective actions and discourses of the Chinese Falun Gong movement. Social Psychology Quarterly 76: 1–24. Cleaveland, C. 2010. “We are not criminals”: Advocacy and unauthorized migrants. Social Work 55: 74–81. Clifford, J., and G. E. Marcus, eds. 1986. Writing culture: The poetics and politics of ethnography. Berkeley: University of California Press. Creswell, J. W., and V. L. Plano-Clark. 2011. Designing and conducting mixed methods research, 2nd ed. Los Angeles: Sage. Cronin, J. M., and M. B. McCarthy. 2011. Fast food and fast games. An ethnographic exploration of food consumption complexity among the videogames subculture. British Food Journal 113: 720–43. D’Andrade, R. G., E. A. Hammel, D. L. Adkins, and C. K. McDaniel. 1975. Academic opportunity in anthropology 1974–90. American Anthropologist 77: 753–73. Davidson, C. M., and P. Hawe. 2012. School engagement among Aboriginal students in northern Canada: Perspectives from activity settings theory. Journal of School Health 82: 65–74. Driver, H. E., and A. L. Kroeber. 1932. Quantitative expression of cultural relationships. University of California Publications in American Archaeology and Ethnology 31: 211–56. Emerson, R. M., R. I. Fretz, and L. L. Shaw. 2011. Writing ethnographic fieldnotes. Chicago: University of Chicago Press. Ezzell, M. B. 2012. “I’m in control”: Compensatory manhood in a therapeutic community. Gender and Society 26: 190–215. Fiske, S. J., L. A. Bennett, P. Ensworth, T. Redding, and K. Brondo. 2010. The changing face of anthropology: Anthropology masters reflect on education, careers, and professional organizations. AAA/CoPAPIA 2009 anthropology career survey. Arlington, VA: American Anthropological Association. Fluehr-Lobban, C. 1991. Professional ethics and anthropology: Tensions between its academic and applied branches. Business and Professional Ethics 10: 57–68. Givens, D., and T. Jablonski. 1996. 1996 AAA Survey of Departments. In The AAA Guide, 304–15. Arlington, VA: American Anthropological Association. Gravlee, C. C. 2011. Research design and methods in medical anthropology. In A companion to medical anthropology, ed. M. Singer and P. Erickson, 69–91. Malden, MA: Blackwell. Greene, J. C. 2007. Mixed methods in social inquiry. San Francisco: Jossey-Bass. Griesbach, D., and S. Grand. 2013. Managing as transcending: An ethnography. Scandinavian Journal of Management 29: 63–77. Gubrium, A., and K. Harper. 2013. Participatory visual and digital methods. Walnut Creek, CA: Left Coast Press, Inc. Haddon, A. C., ed. 1901–1935. Reports of the Cambridge Anthropological Expedition to Torres Straits. Cambridge, U.K.: The University Press. Higham, S., and R. Davies. 2013. Protecting, providing, and participating: Fathers’ roles during their child’s unplanned hospital stay, an ethnographic study. Journal of Advanced Nursing 69: 1390–99. Ho, K. 2009. Liquidated: An ethnography of Wall Street. Durham, NC: Duke University Press.

14_107-Bernard.indb 14

6/9/14 10:02 AM

Introduction

15

Honigmann, J., ed. 1973. Handbook of social and cultural anthropology. Chicago: Rand McNally. Howard, J. 1973 [1792]. Prisons and lazarettos. Vol. 1. The state of the prisons in England and Wales: With preliminary observations, and an account of some foreign prisons and hospitals. Montclair, NJ: Patterson Smith. Jessri, M., A. P. Farmer, and K. Olson. 2013. Exploring Middle-Eastern mothers’ perceptions and experiences of breastfeeding in Canada: An ethnographic study. Maternal Child and Nutrition 9: 41–56. Johnson, A. W. 1978. Quantification in cultural anthropology: An introduction to research design. Stanford, CA: Stanford University Press. Kenny, K. 2012. “Someone big and important”: Identification and affect in an international development organization. Organization Studies 33: 1175–93. Kim, T.-G., E. T. Donnell, and D. Lee. 2008. Use of cultural consensus analysis to evaluate expert feedback of median safety. Accident Analysis and Prevention 40: 1458–67. Kroeber, A., ed. 1953. Anthropology today. Chicago: University of Chicago Press. Kroeber, K. 2003. Curious profession: Alfred Kroeber and anthropological history. boundary 2 30: 144–55. LeCompte, M. D., and J. J. Schensul. 2013. Analysis and interpretation of ethnographic data: A mixed methods approach, 2nd ed. Lanham, MD: AltaMira. Lynch, E. B., and S. Holmes. 2011. Food group categories of low-income African American women. Journal of Nutrition Education and Behavior 43: 157–64. Lynd, R. 1939. Knowledge for what? The place of social science in American culture. Princeton, NJ: Princeton University Press. Malinowski, B. 1922. Argonauts of the western Pacific: An account of native enterprise and adventure in the archipelagoes of Melanesian New Guinea. New York: E.P. Dutton. Marcus, G. E. 2008. The end(s) of ethnography: Social/cultural anthropology’s signature form of producing knowledge in transition. Cultural Anthropology 23: 1–14. doi:10.111 1/j.1548-1360.2008.00001. Mathevet, R., M. Etienne, T. Lynam, and C. Calvet. 2011. Water management in the Camargue Biosphere Reserve: Insights from comparative mental models analysis. Ecology and Society 16: 43. http://www.ecologyandsociety.org/vol16/iss1/art43/ (accessed December 3, 2013). McCurdy, D. W., J. P. Spradley, and D. Shandy. 2005. The cultural experience. Ethnography in complex society, 2nd ed. Long Grove, IL: Waveland. McDonald, L. 1993. The early origins of the social sciences. Montreal: McGill-Queen’s University Press. McDonald, L. 1994. The women founders of the social sciences. Ottawa: Carleton University Press. Mill, J. S. 1869. The subjection of women. New York: D. Appleton. Morgan, L. H. 1870. Systems of consanguinity and affinity of the human family. Washington, DC: Smithsonian Institution. Naroll, R., and R. Cohen, eds. 1970. A handbook of method in cultural anthropology. New York: Natural History Press. Nightingale, F. 1871. Introductory notes on lying-in institutions. Together with a proposal for organising an institution for training midwives and midwifery nurses. London: Longmans, Green. Oberschall, A. 1972. The institutionalization of American sociology. In The establishment of empirical sociology, ed. A. Oberschall, 187–251. New York: Harper and Row. Ogburn, W. F. 1930. The folk-ways of a scientific sociology. Publication of the American Sociological Society 25: 1–10. Pelto, P. J. 1970. Anthropological research: The structure of inquiry. New York: Harper & Row.

14_107-Bernard.indb 15

6/9/14 10:02 AM

16

H. Russell Bernard and Clarence C. Gravlee

Pelto, P. J., and G. Pelto. 1973. Ethnography: The fieldwork enterprise. In Handbook of social and cultural anthropology, ed. J. Honigmann, 241–88. Chicago: Rand McNally. Phillips, S. A. 2012. Operation Fly Trap: L.A. gangs, drugs, and the law. Chicago: University of Chicago Press. Quételet, A. 1969 [1835]. Physique sociale, ou, essai sur le développement des facultés de l’homme. Paris: J.-B. Bailliere et fils. Originally published in 1835. Reprinted in translation in 1969 from the 1842 edition as: A treatise on man and the development of his faculties. Gainesville, FL: Scholars’ Facsimiles and Reprints. Radin, P. 1933. The method and theory of ethnology. New York: McGraw-Hill. Redfield, R. 1948. The art of social science. American Journal of Sociology 54: 181–90. Richardson, J., and A. L. Kroeber 1940. Three centuries of women’s dress fashions: A quantitative analysis. Anthropological Records 5: 111–53. Riggs, D., and C. Due. 2011. (Un)common ground?: English language acquisition and experiences of exclusion amongst new arrival students in South Australian primary schools. Identities: Global Studies in Culture and Power 18: 273–90. Rivers, W. H. R. 1906. The Todas. New York: Macmillan. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus. American Anthropologist 88: 313–38. Royal Anthropological Institute of Great Britain and Ireland (RAI). 1951. Notes and queries on anthropology, 6th ed. London: Routledge and Kegan Paul. Rudd, E., E. Morrison, J. Picciano, and M. Nerad. 2008. Social science PhDs five+ years out. Anthropology report. CIRGE Report 2008-01. Seattle: CIRGE. www.cirge.washington.edu. http://depts.washington.edu/cirgeweb/social-science-phds-five-years-out-anthropologyreport/ (accessed May 23, 2013). Schatz, E., ed. 2009. Political ethnography: What immersion contributes to the study of power. Chicago: University of Chicago Press. Seligmann, C. G., B. Z. S. Seligmann, with C. S. Myers, and A. M. Gunesekara. 1911. The Veddas. Cambridge: The University Press. Shiu, A. T-Y, S. F. Twinn, G. C-T Lee, C. W-H. Chan, and K. K-F Kwong. 2010. Strategies to maximise cervical cancer screening rates among older Hong Kong Chinese women—A cultural consensus analysis study. Journal of Clinical Nursing 19: 3244–46. Smith, C. D., and W. Kornblum. 1996. In the field: Readings on the field research experience, 2nd ed. Westport, CT: Praeger. Spradley, J. P. 1979. The ethnographic interview. New York: Holt, Rinehart, and Winston. Spradley, J. P. 1980. Participant observation. New York: Holt, Rinehart, and Winston. Spradley, J. P., and D. W. McCurdy. 1972. The cultural experience. Ethnography in complex society. Chicago: Science Research Associates. Tashakkori, A., and C. Teddlie, eds. 2010. Handbook of mixed methods in social and behavioral research, 2nd ed. Los Angeles: Sage. Taylor, C. C. 1920. The social survey and the science of sociology. American Journal of Sociology 25: 731–56. Teddlie, C., and A. Tashakkori. 2009. Foundations of mixed methods research. Newbury Park, CA: Sage. Terry-Sharp, K. 2009. 2009. Anthropology faculty jobs market report. Anthropology News 50: 34. Tooker, E. 1997. Introduction. Lewis Henry Morgan. Systems of consanguinity and affinity of the human family. Lincoln: University of Nebraska Press. Torres, L. 2009. Latino definitions of success: A cultural model of intercultural competence. Hispanic Journal of Behavioral Sciences 31: 576–93.

14_107-Bernard.indb 16

6/9/14 10:02 AM

Introduction

17

Tryon, R. C. 1939. Cluster analysis. Ann Arbor, MI: Edwards Brothers. Tylor, E. B. 1889. On a method of investigating the development of institutions applied to the laws of marriage and descent. Journal of the Royal Anthropological Institute of Great Britain and Ireland 18: 245–72. Webb, B. P. 1926. My apprenticeship. London: Longmans, Green. Webb, S., and B. P. Webb. 1910. The state and the doctor. New York: Longmans, Green. Wedeen L. 2008. Peripheral visions: Publics, power, and performance in Yemen. Chicago: University of Chicago Press. Werner, O., and G. M. Schoepfle. 1987. Systematic fieldwork (2 vols.). Newbury Park, CA: Sage. White, D. R., and P. Jorion. 1992. Representing and computing kinship: A new approach. Current Anthropology 33: 454–53. Whiting, J. W. M. 1982. Standards for psychocultural research. In Crisis in anthropology. View from Spring Hill, 1980, ed. E. A. Hoebel, R. Currier, and S. Kaiser, 155–64. New York: Garland. Wolcott, H. F. 1994. Transforming qualitative data: Description, analysis, and interpretation. Thousand Oaks, CA: Sage. Wolcott, H. F. 2005. The art of fieldwork. Walnut Creek, CA: AltaMira. Wolcott, H. F. 2010. Ethnography lessons: A primer. Walnut Creek, CA: Left Coast Press, Inc. Wolf, E. R. 1964. Anthropology. Englewood Cliffs, NJ: Prentice Hall. Zubin, J. A. 1938. A technique for measuring likemindedness. Journal of Abnormal and Social Psychology 33: 508–16.

14_107-Bernard.indb 17

6/9/14 10:02 AM

14_107-Bernard.indb 18

6/9/14 10:02 AM

Part I

PERSPECTIVES

14_107-Bernard.indb 19

6/9/14 10:02 AM

14_107-Bernard.indb 20

6/9/14 10:02 AM

CHAPTER 1

Epistemology The Nature and Validation of Knowledge Michael Schnegg

This chapter examines the nature of knowledge in social and cultural anthropology and the ways it can be obtained. The philosophical discipline concerned with the nature and validation of knowledge in general is called epistemology and closely relates to ontology, which deals with the nature of existence. Since the anthropological frameworks discussed here share similar ontological positions, I focus on the epistemological foundation of knowledge, which seeks to answer the following three questions: 1. What is knowledge (i.e., nature)? With reference to Plato’s Theaetetus, knowledge is often defined as justified true belief. Not all true beliefs are knowledge. Consider a football game: You may believe that the Steelers will beat the Giants 28–14. Even if this turns out to be true, you did not know the result; you guessed the final score correctly. You may, however, actually know the results of a game that has already been played— perhaps because you observed it or read about it in the newspaper. This information justifies your belief. In anthropology, the central question about the nature of knowledge is not whether a true belief is justified but how we justify that a belief is true. This brings up the second question. 2. How can we obtain knowledge (i.e., sources)? Most epistemologists agree that the quality of the reason given to justify a belief is essential. Not only is it difficult to judge, but different philosophical traditions favor different justifications of knowledge claims: claims based on reason (logics and mathematics), on the senses (experiences), or on introspection. 3. What are the limits of knowledge (i.e., its validity)? Since the beginning of formal Western thought, epistemologists (like Pyrrho of Elis [c. 360–c. 270 BC]) have asked about the limits of knowing and knowledge and whether we can even obtain knowledge about the world at all.

The most basic distinction in the epistemology of the social and behavioral sciences and the humanities is the split between scientific and humanistic approaches (Bohman 1991; Carson 2001; D’Andrade 1986; Franklin 1998; Hollis 1994; Kuznar 1997; Lang 1994). Hermeneutics and postmodernism/radical constructivism are the main contemporary currents of the humanistic approaches; the scientific approach is commonly associated with different branches of positivism (e.g., logical, analytical). Positivism proposes a unified methodology for different branches of the sciences and the humanities. It aims to objectively explain the causes for and relationships among the discrete phenomena of nature. Positivism proposes the discovery of theories and general 21

14_107-Bernard.indb 21

6/9/14 10:02 AM

22

Michael Schnegg

laws as the ultimate goal of scientific inquiry. The cumulative growth of knowledge in the natural sciences is the model for this framework. Hermeneutics (the more general term, or interpretivism, a more narrow term), the rival framework to positivism, is rooted in the humanities. It tries to establish a special methodology for understanding (Verstehen) phenomena in the human and social sciences. The model case of hermeneutics is the interpretation of texts, and its objective is the exploration of common understandings in historically based cultural traditions that are present in their textual products. In the interpretation of texts, understanding is the result of an ongoing dialectical or dialogic relationship between texts and the interpreter. Understanding is not objective or universal; rather, it is subjective and highly contextual. Different interpreters can come up with very different readings of the same empirical material. Positivist and hermeneutic approaches are creations of the nineteenth century; postmodernism and radical constructivism are more recent developments. Some consider them radical versions of hermeneutics; others claim that these approaches are true alternatives. Postmodernism questions systematic approaches to the production of cumulative knowledge (which radical constructivism does not reject), and postmodernism and radical constructivism both adopt a relativistic stance that stresses the creative power of scientists to construct reality. In the next three sections, I present the developments, main tenets, and some strengths and weaknesses of these frameworks. Their answers are not completely different, and the frameworks are internally more heterogeneous than the classification implies. In the discussion, I will move from the philosophical positions to anthropological appropriations and show how they lead to specific methodologies and ways of ethnographic writing. The questions introduced above will guide the comparison. The final section aims to connect positions with arguments and proposes some thoughts for a unifying epistemology of anthropological research. POSITIVISM AND THE ANALYTIC TRADITION

Today, the term “positivism” is often negatively associated with narrow-minded data collection, number crunching, and acceptance of the status quo. This was not its meaning when it was established by European social scientists and social reformers. Positivism developed out of Empiricism, which emerged in England in the seventeenth and eighteenth centuries through the works of John Locke (1632–1704) and David Hume (1711–1776). Both opposed the then-dominant rationalist position that knowledge comes from metaphysical agents (e.g., God). On the one hand, they had experienced how the state and the church abused this logic to enforce their interests. On the other hand, they questioned the dominant logic by arguing that, if all knowledge came from reason, a newborn should also know about language and geometry. Since this was not the case, they argued that the brain of a newborn is like an empty container—Locke’s tabula rasa, or blank slate—that fills through our experiences and guides our interpretation of the world. Empiricists argue that this logic not only applies to everyday learning but to science as well and that scientific knowledge must be obtained through our senses. Auguste Comte transferred these ideas to the study of society and called this sociology. Comte argued that society was more complex than nature and human thought

14_107-Bernard.indb 22

6/9/14 10:02 AM

1: Epistemology

23

and thus needed its own discipline, one that would build on the data and knowledge obtained by other sciences through observations, experiments, and comparisons. His mission was twofold: (1) like Hume and Locke, he tried to get rid of the then-dominant metaphysical explanations for social facts and wanted to establish social laws based on observations and empirical data; and (2) he aimed to spread his knowledge to form a new worldview of humankind to replace religions (Bunnin and Tsui-James 1996; Dancy et al. 2010; Papineau 1996). After this first wave of positivism in the nineteenth century in philosophy and some related disciplines, positivistic epistemology experienced its main boost in Vienna during the 1920s and 1930s (Achinstein and Barker 1969; Carnap 1963; Geier 1993; Janik and Toulmin 1973). In this context, the self-denomination also changed from positivism to logical empiricism, sometimes accompanied by the establishment of a more general analytic tradition in the various disciplines (Soames 2003). From now on, we will label this analytic anthropology. As with earlier positivists, the so-called Vienna Circle wanted to free philosophy and the empirical sciences of any speculative assumptions not founded on observation or confirmed by empirical tests. At the same time, they turned away from Comte and argued that he had violated his own paradigm by proclaiming that his ideas constituted a new worldview, thus adding just another metaphysical explanation (Outhwaite 1876, 6). The newly invented methods of formal logic were applied to philosophical and other scientific statements to clarify language, detect and refute metaphysical claims, and create better science. The ideas of the logical positivists spread to other parts of Europe (Berlin, Prague, Oxford, Cambridge) and the Americas before and after World War II and strongly influenced social science in the first half of the twentieth century. Positivism starts with the assumption that we can describe reality through our senses. For logical positivists, statements make sense only if they can be inferred inductively from statements that have been tested empirically. This principle is known as verification. Here are three examples: 1. Earth’s core does not consist of chocolate. 2. The risk of getting infected with HIV is higher among women than men. 3. The Virgin of Guadalupe is more powerful than Saint Michael.

The first statement can be tested through drilling deep into the earth. The result may be a statement like the following: Person A observed at time t after drilling with instrument X 15 kilometers into the earth that the material obtained was not chocolate. Other observations may lead to similar statements. In sum, they allow us to infer a proposition like the one made above. The second statement is an assertion of probability rather than of simple observation, but the same logic of verification can be applied to the second statement as to the first. The question about whether the Virgin of Guadalupe or Saint Michael is more powerful, however, cannot be verified. It is thus outside this realm of science. To come up with valid statements about the world, science needs criteria to judge whether a statement is true or not (Dancy et al. 2010). Positivistic thinking is based on the correspondence theory of truth. This theory states that we can determine the truth value of a statement by comparing it with facts.

14_107-Bernard.indb 23

6/9/14 10:02 AM

24

Michael Schnegg

If the statement reflects the relationships in reality properly, it is said to be true. This raises the problem of how to determine whether a statement refers to facts and marks the border between epistemology and methodology. Within the positivistic framework, conceptual statements are operationalized (e.g., connected to empirical indicators) and finally measured (e.g., through an observational instrument that allows the collection of empirical indicators) (Bernard 2011; Fiske 1986). Within this framework, theory building is an ultimate goal of science and induction is a process of making general knowledge claims on the basis of verified statements. Wittgenstein (1922, 25) argued that the world can be understood as the totality of verified statements. “The world is everything that is the case,” as he states in the first sentence of the Tractatus Logico-Philosophicus. A knowledge claim is considered true so long as the collection of verified statements is consistent with the knowledge claim. However, should verified statements emerge that are inconsistent with the knowledge claim, the inductive researcher must either revise the knowledge claim or question the accuracy of the verified statement itself. The only way to do the latter is to collect additional observations following the same protocol that produced those verified statements that were inconsistent with the original knowledge claim. If this yields additional anomalous verified statements, then the original knowledge claim must be revised to accommodate the new information. Apart from induction, there is a second way to build theory from data. Viennese philosopher Karl Popper (1969a, 1969b [1934], 1972) regarded himself as a critic of (logical) positivism. Nevertheless, he developed critical rationalism in close connection to logical positivism. Popper concluded that positive evidence (confirmation) and the inductive method (the search for rules that leads from limited observations to the establishment of valid generalizations) are not at the heart of science. Rather, negative evidence (falsification) and deduction are at the core. Since our tests are limited, we can never be sure that a (universal) theoretical statement is true. Even if all present evidence is confirmatory, there could always be a refutation in the future. In the spirit of critical rationalism, when choosing between rival hypotheses, we should always select the one that has a higher information content by being more general and thus more challenging due to its wider range of application. And we should keep the one that has survived serious attempts at falsification and therefore has proven less false than its rival. Against this view of falsification, some philosophers of science, most notably Willard Quine (1961, Ch. 2), pointed out that the test situation is always more complex: A wrong conclusion could imply that the hypothesis at hand is wrong, but it could also falsify any other statements in the premise, including measurement theories and observational statements. Hence, the uses and tests of hypotheses are always based on a holistic understanding of the whole test situation and are guided by inductive (probabilistic) judgments of the development potential of theoretical statements in future applications (Papineau 1996). So, the one-sided critical view of falsificationism has to be supplemented by a more positive heuristic of hypothesis construction and theory development. Today, deduction and induction are rarely seen as true alternatives. All inductions are based on some theoretical assumptions, including assumptions about how to describe and measure phenomena, and deduction is confronted with empirical

14_107-Bernard.indb 24

6/9/14 10:02 AM

1: Epistemology

25

realities that modify theories. Deduction and induction are thus more appropriately conceived of as ideal–typical poles of a continuum, and any specific research project should be positioned as being on the continuum rather than at one or the other pole (Beckermann 1997, 32). POSITIVISM IN ANTHROPOLOGY

Epistemological Positions To understand the impact of positivism on anthropology, we consult the work of one of the discipline’s founders: Bronislaw Malinowski (1884–1942). Anthropology was deeply rooted in colonialism and evolutionism when Malinowski got interested in the subject after finishing his dissertation in philosophy at Cracow University in Poland. Like other anthropologists of the time (e.g., Boas, Rivers), Malinowski had trained in the natural sciences and was educated in an intellectual environment that upheld the unity of all sciences. Malinowski was influenced by positivism when he started engaging with anthropology (Flis 1988, 125). At that time, compared to psychology or sociology, anthropology didn’t have a well-developed tradition of empirical research or well-developed standards for data collection, induction, and theory building. The works of prominent anthropologists, including Sir James Frazer and Sir Edward Tylor, were still written in the spirit of evolutionary thinking and based on secondary sources. While anthropologists sat in their home institutions (museums or universities) and wrote their scientific works, travelers collected the data. The problems associated with this approach are discussed in the first edition of Notes and Queries on Anthropology in 1874. Tellingly, the subtitle reads: For the Use of Travelers and Residents in Uncivilized Lands (Notes 1874). Malinowski’s Argonauts is one of the early examples of ethnographic writing that breaks with this separation between data collection and authorships and declares the combination of the two efforts as a research program. In the introduction, setting the empirical tone of his work, Malinowski (1922, 3) begins by criticizing his teachers: “It would be easy to quote works of high repute, with a scientific hall-mark on them, in which wholesale generalizations are laid down before us, and we are not informed at all by what actual experiences the writers have reached their conclusion.” Analyzing the works of the past and reflecting on his own ethnography in the Trobriands, he concludes that “success can only be obtained by patient and systematic application of a number of rules of common sense and well-known scientific principles” (Malinowski 1922, 6). The student of anthropology, said Malinowski, must “possess real scientific aims . . . put himself in good conditions of work, that is, in the main to live without other white men, right among the natives . . . [and] apply a number of special methods for “collecting, manipulating and fixing his evidence” (Malinowski 1922, 5). For Malinowski and most of his peers, culture could be described as an objective reality. The aim of anthropology was to use the best methods available to collect ample evidence about it and to draw generalizations from there. His plea for a “concrete statistical documentation” of observed events for the collection of behavioral observations and for the documentation of “the native’s point of view” mentality all reflect this attitude (Malinowski 1922, 6). His aim to grasp that native point of view was less an attempt to understand the world from the perspective of his informants than an attempt

14_107-Bernard.indb 25

6/9/14 10:02 AM

26

Michael Schnegg

to document their behavior and attitudes most precisely and reliably through the use of scientific methods. The prevalence of this approach is also seen in Rivers’s call for vital statistics as a starting point of ethnography (Rivers 1910). How does Malinowski—the ethnographer—establish the validity of his statements (i.e., how does he justify that they are true)? He does so through being there and living with the people, through his subjective experience. Malinowski deliberately separates himself from travelers, missionaries, and traders whose involvement “makes a real, unbiassed, impartial observation impossible” (Malinowski 1922, 18). While writing about the Kula, he concludes that “Not even the most intelligent native has any clear idea of the Kula as a big, organized social construction, still less of its sociological function and implications” (Malinowski 1922, 83). The objectification of ethnographic facts is thus accomplished through being there (Clifford 1986b, 26; Fuchs and Berg 1995, 29). Malinowski’s work is deliberately inductive (Malinowski 1964, 11). In his early analysis and criticism of the work of Frazer, he concludes that “A precise concept of totemism, like all empirical concepts, is acquired only through induction and comparing phenomena: carefully investigating the characteristics in each case and taking what is common in all these cases as a general notion of totemism” (Malinowski 1993, 127). At the same time, he recognized that “every description, out of necessity, has to be based on a theoretical foundation” (Malinowski 1993, 127). This separated Malinowski from radical empiricists like Ernst Mach (Flis 1988, 126), a prominent physicist and a philosopher of science. For a time, Mach notoriously denied the existence of atoms whose very existence his colleagues Max Planck and Niels Bohr had proven (Dancy et al. 2010, 540). On the other hand, Mach equally argued that explanations belong to the realm of metaphysics and should be replaced by functional (mathematical) statements that describe the covariance of elements without making a decisive statement of what explains what. Malinowski followed that lead and made it the core of his theoretical model: All parts of culture are related and fulfill a function with respect to the whole, which should be the subject of inquiry. This was to become the kernel of functionalism (Flis 1988, 119). Like many of his contemporary colleagues, he aimed to describe and document culture as the totality of religion, policy, art, economy, and so on. In sum, much in line with positivistic thinkers, most anthropologists of the first half of the twentieth century, like Franz Boas (1961 [1940], 260, 269), Robert Lowie (1959 [1937], 279, 291), Alfred Kroeber (1948, Sect.1, 7), George Peter Murdock (1949), and A. R. Radcliffe-Brown (1957 [1948], 1968 [1952]), claimed that the sources of anthropological knowledge are our senses (question 2, sources of knowledge). Starting with Malinowski, a strong and lasting tradition developed that may be considered empiricist. In line with the scientific paradigm, improving methods and the transparent communication of data, analysis and results became the tenets of ethnography. Early positivists turned anthropology into a truly empirical science that aimed to explain social and cultural phenomena and its variations through laws or law-like statements. Today, many agree that it was unrealistic to measure anthropologists and other social scientists against the standards put up for physics. The social world is too complex to find laws of similar scale. As D’Andrade (1986) points out, a search for generalizations is a more appropriate goal for an analytic anthropology, as most of the

14_107-Bernard.indb 26

6/9/14 10:02 AM

1: Epistemology

27

anthropological work of the time, done within this epistemological framework, was based on the principle of induction and verification (question 2, sources of knowledge). There was relatively little concern about the limits of knowledge (question 3, limits of knowledge), and many of the first generation of anthropologists did not question that cultures were real and existed independently of the anthropologists whose task it had become to document them.

Methodological Implications Ethnography in the Malinowskian style follows a certain logic. Malinowski himself identified three steps: 1. The organisation of the tribe, and the anatomy of its culture must be recorded in firm, clear outline. The method of concrete, statistical documentation is the means through which such an outline has to be given. 2. Within this frame, the imponderabilia of actual life, and the type of behaviour have to be filled in. They have to be collected through minute, detailed observations, in the form of some sort of ethnographic diary, made possible by close contact with native life. 3. A collection of ethnographic statements, characteristic narratives, typical utterances, items of folk-lore and magical formulae has to be given as a corpus inscriptionum, as documents of native mentality. (Malinowski 1922, 24)

The three steps clarify Malinowski’s understanding of the ethnographic endeavor and the way to achieve knowledge. Statistical documentation, detailed observations, and qualitative statements allow “recording” the culture to be studied. The ethnographer stays in the field for long periods of time and learns the language. For Malinowski, the application of scientific methods including the presentation of detailed notes, tables, and pictures is at the core of his approach. Stocking (1983, 109) argued that being cut off from the rest of the world was a second major strategy to justify the validity of Malinowski’s knowledge claims. Most of the work done in the analytic tradition is based on the correspondence theory of truth. Empirical research defines indicators for the statements being made and proposes ways to examine and measure those indicators. The epistemological challenge to justify that a belief is true is met through the control of data collection and analysis. In the scientific paradigm, a statement is said to be a justified true belief—and, thus, knowledge—if it has high reliability and validity. Validity refers to the quality of the data as an image of reality. We distinguish between different types of validity, most importantly construct validity, content validity, and criterion validity. Construct validity refers to the extent to which the conceptualization and operationalization of a theoretical construct are really measured by the instrument—for example, the extent to which an IQ test actually measures intelligence. Construct validity is tested by using different instruments or methods to measure the same concept. If the results are strongly correlated, construct validity is said to be high (Campbell and Fiske 1959). While the idea was originally developed for quantitative data and psychological tests, Campbell later extended it in an early appeal for the use of mixed methods to explore one construct (Campbell 1970, 67). Content validity

14_107-Bernard.indb 27

6/9/14 10:02 AM

28

Michael Schnegg

determines whether all relevant dimensions of a theoretical construct are really captured and measured—that is, whether the complexity of constructs like religiosity or environmental awareness is adequately reflected in the instrument designed to measure it. Finally, criterion validity examines whether a measurement is correlated with other measurements that are known to measure the construct adequately. Reliability—the second measure—refers to the probability that another round of data collection and analysis will reveal similar or the same results (Bernard 2011). In anthropology, this may refer to a restudy of the same ethnographic situation by a different ethnographer or a second application of a data collection method by the same researcher. The former problem was famously raised in anthropology with the restudy of Tepoztlán, Mexico, by Oscar Lewis in the 1940s. Lewis found the people of Tepoztlán to be unhappy and unfriendly, individualistic and uncooperative, conflict ridden and not peaceful—practically the opposite of the way Robert Redfield had described them 20 years earlier (Lewis 1951; Naroll 1970, 928; Redfield 1930). In response, even before the postmodern critique, anthropologists began developing ways to contextualize ethnographic information, taking account of things like the funding agency, the graduate professor, the field team composition, and so on (Naroll 1970, 935). Usually, the reliability of systematic (and thus narrower) data collection methods, like questionnaires and experiments, is higher, making the choice about methods also a choice between validity and reliability.

Writing Ethnography In his critical analysis of ethnographic writing, van Maanen (2011, 46) identified four characteristics of this realistic style of writing in anthropology. Even though they draw a rather one-sided picture, they help us understand some of the characteristics of the classical ethnographies that emerged as a new literary genre with Malinowski and others around the turn of the twentieth century: • Experiential author(ity): Perhaps the most distinct characteristic of early ethnographic writing is the exclusion of the fieldworker from the report. Having stayed in the field and collected the data, he or she describes how the people feel and behave. By taking the “I” (the observer) out of the report, worries about the subjectivity of the fieldworker become moot. • Typical forms: The second characteristic element of style is the particularized description of details of everyday lives, including events that are at the center of interest for the ethnographer. • The native’s point of view is usually transmitted through detailed description or through extensive quotes. • The interpretative omnipotence of the author refers to the strategy of creating a single line of interpretation of the collected information that often links ethnographic details to the concepts of respected heroes of the discipline.

A fifth point can be added. The Argonauts and related works introduce a new way of ethnographic writing by overcoming the separation between the object (people/ culture) and the writer. Malinowski argues that, while travelers had only superficial experiences and were not able to overcome lay people’s biases, he, as the scientist who

14_107-Bernard.indb 28

6/9/14 10:02 AM

1: Epistemology

29

had lived there, could. His detailed, realistic, and objectified description became possible through his subjective relationship with the informants: being there, living with the people, and sharing daily routines. In sum, the early ethnographic monographs written in the scientific tradition share the conviction that social reality exists independently of our perception and can be recorded as such in the ethnographic report—in a monograph. It soon became evident that perceptions of ethnographers differ and that we can never assess reality in a fully unbiased way. For empiricist ethnographers, the extent to which bias exists can be assessed by applying the concepts of validity and reliability and improving validity and reliability remains a key concern of modern ethnographies written in this tradition (Cancian 1992; Cohen 1999; Hirsch 2003; McCabe 2004; Rao 1998). To practitioners of this school, the transparency of data collection and analysis provides intersubjectivity that distinguishes science from art and other attempts to grasp the world around us (Gingrich 2009, 187). HERMENEUTICS

Hermeneutics (from Greek hermeneutike, meaning the art of interpretation) has its roots in the classical tradition of Greek antiquity. It originally flourished in the humanities, which were preoccupied with the exegesis of historical texts—like the Bible, philosophical and jurisdictional treatises, historical documents, and literature of the past—often written in foreign languages and stemming from distant epochs (Geldsetzer 1989). A major problem for philologists in Europe was to understand and unravel the meaning of specific terms and to grasp the meaning of the text at large. This became especially evident when translating texts. Martin Luther’s translation of the Bible was so controversial because some of his interpretations were far from consensual at the time. Thus, the translator faces a problem: To know the meaning of the whole (a paragraph, a parable), he or she needs to understand the meaning of its parts (words, sentences). But to understand the meaning of the parts, we often need some understanding of the text (the whole), which provides us with background knowledge. Hermeneutic work on texts or discourses shifts between these two sides of the interpretive undertaking, with the object of grasping the meaning of both the parts and whole. This process is called the “hermeneutic circle” and is one of the fundamental tools in hermeneutic text interpretation. Beginning in the late nineteenth and early twentieth centuries, some (mostly German) philosophers and associated humanistic methodologists developed the larger vision that hermeneutic rules of text interpretation could be extended to become general guidelines for research in the humanities, or even all disciplines that dealt with human experiences. In contrast to the positivistic paradigm that emerged with Comte and others, Friedrich Schleiermacher and especially Wilhelm Dilthey argued that human behavior is different from the natural world and thus requires a distinct epistemological and methodological approach. As Dilthey put it in his famous phrase: “We explain nature, [but] we understand the living of the mind” (Dilthey 1900 [1894], 144). Hermeneutics argues that, while the natural world follows laws, humans seek and create meaning and always act in a specific cultural and historical context. These two

14_107-Bernard.indb 29

6/9/14 10:02 AM

30

Michael Schnegg

features were seen to be in sharp contrast to those of the natural sciences whose objects do not produce meaning and whose researchers can neglect history. The study of humans and their meaningful products in special historical circumstances should thus describe the particular and proceed along the lines of text interpretation and empathic understanding (Verstehen, in the sense of adopting the actor’s point of view). Basically, empathic understanding is invoked when we want to know the reasons or purposes underlying the meaningful behavior (including speech) of other people in the present or past. Hermeneutics proposes that the researcher engages in a thought experiment and tries to grasp the actor’s point of view by acting as though he or she were in that situation. To be sound, this thought experiment has to rest on validated background knowledge. Otherwise, one risks distorting the decision situation of the actor and mistakenly taking the self for the Other. German philosopher Martin Heidegger built on the early works of Dilthey, Schleiermacher, and Husserl and became one of the most influential advocates of the hermeneutic approach. Fundamental to his position is the distinction between Sein (translated as being) and Seiendes (usually translated as beings or entities). Let us start with the latter. Seiendes refers to the collection of physical objects that constitute the universe. It is, so to speak, the world out there, as positivists would have it. Heidegger distinguishes Sein from this physical universe as our understanding (Verstehen) of those entities. Our understanding constitutes Sein. Hence, things we do not understand or perceive do not exist. For Heidegger, distinguishing between Sein and Seiendes is an analytical trick to allow us to detect what he calls the “ontological difference”: the difference between Sein and Seiendes/Seiendem, or between what there is and what we perceive (Heidegger 1962, 2000, 344). The Sein predates our perception of a given situation. In his lecture, Heidegger uses the statement “The blackboard badly positioned” (Die Tafel steht ungünstig) to explain this relationship and thus the ontological difference. The unfavorable position of the blackboard is not a property of the blackboard as such, like its color or its size. It exists only in relation to the subject(s) that perceive(s) it. The unfavorable position of the blackboard predates the situation in which we recognize and state it because we have a common understanding of the meaning and the purpose of sitting in the classroom. The classroom situation is defined by the relationship between the audience and the instructor, the rules of communication, and the technical devices used. This knowledge guides our understanding of the objects and leads to the statement about the Sein (unlucky position) of the Seiendes (blackboard) (Heidegger 2000). Heidegger transfers his insights about the nature of knowledge into a more general paradigm of how to obtain and accumulate it. As scientists, we also have pre-knowledge about the world, which makes it impossible to describe the world as such. He proposes entering a hermeneutic circle to detect the difference between this Sein of the scientist and the Seiendes of what she or he is trying to understand (Heidegger 1962, 2000). Building on this model, Heidegger’s pupil Hans Georg Gadamer tried to establish hermeneutics as a universal theory of interpretation for all the sciences (Gadamer 2004). He made two important points about the ways we obtain knowledge. First, he argues in Truth and Method that our notion of knowledge is still largely influenced by

14_107-Bernard.indb 30

6/9/14 10:02 AM

1: Epistemology

31

Kant and Enlightenment thinking. Kant had separated the rational (Vernunft) from the nonrational, as if they were two sides of a coin. In this view, objective knowledge could only be built through scientific methods (Gadamer 2004, 36; Hekman 1983, 207). Gadamer questions this distinction when he asks rhetorically: “Is there to be no knowledge in art? Does not the experience of art contain a claim to truth which is certainly different from that of science, but equally certainly is not inferior to it?” (2004, 84). For Gadamer, there is such knowledge. He perceived it as a great deficiency to exclude all experiences from the realm of truth that have not been obtained through scientific methods. Second, and building on Heidegger, he argues that there are structures that preexist the entities and the way we perceive them in a specific situation. We cannot perceive the world in a neutral, rational way. This leads him to recognize prejudice as an integral part of knowledge construction. To put it in his words: “[T]he fundamental prejudice of the enlightenment is the prejudice against prejudice itself” (Gadamer 2004, 273). Any understanding thus involves the examination of prejudice. With Gadamer, we see clearly how the hermeneutic notion and the application of the hermeneutic circle as its main analytic tool have shifted. In the beginning, it served as a heuristic tool to analyze the relationship between the parts and the whole of a text. The aim was to grasp the meaning of both. Dilthey transformed it into a methodology to analyze the historical and cultural context in which a text and social phenomena more generally were produced. This proved especially fruitful for historical science. It became, so to speak, a tool to overcome the prejudice of the time and the context. Heidegger, and most explicitly Gadamer, moved a step further and argued that it is not possible to enter the minds of those we want to understand. There are no objects external to the inquirer; rather, those objects are produced through our inquiry. This prejudice of the researcher must become an integral part of the analysis. The works of Heidegger and Gadamer give a different answer to the question of what knowledge is and how it can be constructed. They are still in line with some positivistic claims: Yes, there is a world out there, and scientists can describe it. However, the focus has shifted. While positivists aim to explain human behavior, the focus has shifted to understanding structures of meaning. Understanding generally refers to analyzing the context in which a specific behavior is shown to enable the researcher to follow this logic. Critics of hermeneutics (Albert 1994; Stegmüller 1979) make two points. First, this is not a vicious circle that we cannot evade or even a principal problem that we cannot solve since we must always distinguish between background knowledge (assumptions) and claims that can be checked. In principle, every element of background knowledge can be criticized (but not all at the same time). Second, this dilemma is not exclusive to the humanities. It applies to any knowledge-making situation, including hypothesis testing in the natural sciences. HERMENEUTICS IN ANTHROPOLOGY

Epistemological Positions In anthropology, hermeneutical ideas were taken up by Clifford Geertz (1975, 1983, 1988, 1995), founder of an interpretive anthropology. Geertz laid down the theoretical

14_107-Bernard.indb 31

6/9/14 10:02 AM

32

Michael Schnegg

foundation of this paradigm in the much-cited essay “Thick Description,” while his analysis of the Balinese cockfight is often referred to as a best-practice example of how to apply it (Geertz 1973). Let us start with the question of what reality means to Geertz. It may come as surprise to some that his ontological position does not differ significantly from that of Malinowski and others who wrote before him. The thing to ask about a burlesqued wink or a mock sheep raid is not what their ontological status is. It is the same as that of rocks on the one hand and dreams on the other—they are things of this world. The thing to ask is what their import is: what it is, ridicule or challenge, irony or anger, snobbery or pride, that in their occurrence and through their agency, is getting said. (Geertz 1973, 10)

Geertz thus does not question the ontological status of human action (a burlesqued wink) or ethnographic observation (a mock sheep raid). He argues that the main point is not their existence but their meaning. Following Max Weber, he contends that culture is the web of significance that people spin and in which they live “and the analysis of it to be therefore not an experimental science in search of law but an interpretive one in search of meaning” (Geertz 1973, 5). According to Geertz, the main task of anthropology should be guessing at meanings, assessing the guesses, and drawing explanatory conclusions from the better guesses (1973, 20). Geertz borrows an example from the British philosopher Ryle to explain this logic: Imagine two boys rapidly contracting the eyelids of their right eyes. This behavior could be an involuntary twitch or a conspiratorial signal to a friend. A camera-like capturing of the scene would not allow capturing the meaning of it. Only an interpretation of the wider context, the behavior that preceded the contracting of the eyelids and the situation in which it happens, allows a proper, or thick description as Geertz called it referring to Ryle (Geertz 1973, 6). Geertz assumes the existence of the world as given and identifies the task of anthropologists as describing its meaning structure to distinguish the winks from twitches. This poses the problem of verification, or, to use a term favored by Geertz, “appraisal,” of an ethnographic description. Geertz argues that the measurement of validity is the “power of scientific imagination to bring us into contact with the lives of strangers” (Geertz 1973, 16). Unfortunately, he gives little concrete advice about how to tell which of the two stories describing the event either as a wink or as a twitch is correct. The essential epistemological challenge to justify that a belief is true is thus dealt with only superficially. This problem transfers to the next level when it comes to theory building. Having identified understanding and describing as central moments in anthropological research, Geertz rejects a positivistic vision of theory. According to Geertz, anthropological theory that builds on those insights stays much closer to the ethnographic ground than theories in other sciences. Theory building does not follow a cumulative curve but breaks up in disconnected bits and pieces of knowledge that do not take off from the previous knowledge but dig deeper into it. Although the approach starts from the ethnographic example and the description of phenomena, it is not inductive in the sense we have described above. New insights are not used to reject older statements; instead, a new one is included into a modified and presumably more valid version of the theory.

14_107-Bernard.indb 32

6/9/14 10:02 AM

1: Epistemology

33

Geertz seems to recognize the problem when he states that a study is an advance if it is more “incisive—whatever that may mean” than those in the past. Geertz provides no specific criteria nor does he even seem to have an idea of what these criteria for giving a more incisive account might look like (Geertz 1973, 25). The aim of theory building is thus not to generalize across cases, as comparative research in a positivistic tradition would have it, but to generalize within cases (Geertz 1973, 26). Theory provides a vocabulary for description, and concepts for the interpretation. As we have seen, the ontological question is little debated between positivist and interpretative anthropology. The epistemological position of Geertz picks up many of the ideas articulated by Dilthey and his followers. He identifies humans as the creators of significances and meanings, and identifies hermeneutic understanding (Verstehen) as an appropriate tool to analyze them. The approach begins with empirical observation, but it does not assume that the thing observed is independent of the meaningmaking observer, the other social actors present, or the history of meaningful social action that preceded the event (question two, sources of knowledge). Unlike positivistic explanation, ethnographic understanding produces knowledge claims that are very much dependent on the standpoint of the ethnographer him- or herself, rendering the search for objective generalizable knowledge claims an epistemological impossibility. Geertz’s belief that some ethnographic accounts could be more incisive than others suggests that he is less skeptical of the truth value of anthropological knowledge claims than either Heidegger or Gadamer were (question 3, limits of knowledge). While law-like statements and generalizations across cases are key in the analytic tradition, understanding specific local cultural symbols and practices is the main goal of interpretative anthropology.

Methodological Implications As we have seen, one main difference between positivistic and hermeneutic thinking lies in the sources of knowledge and the procedures with which knowledge can be obtained. In hermeneutics, Verstehen gained center stage. Meaning and symbols became the focus of the analysis. Geertz focused on institutionalized actions, like the Balinese cockfight, that represent some deeper aspect of Balinese culture. Like a Shakespearean play, the cockfight can be read and analyzed as a text that contains knowledge about the author and the time it was written. The notion of culture-astext put meaning structures at the forefront and finally broke with the behavioristic tradition in anthropology. The anthropologist was to collect those “texts” from the local population and interpret them. Much of the work done in the interpretative tradition sticks to the general idea of a correspondence theory of truth. However, the judgment of a knowledge claim is based on the believability and convincingness of the interpretation of the observation against the array of other observations that form the context for the interpretive claims. So, “truth” in this sense corresponds to the “right feel” of the interpretive analyst. As we have seen, Geertz offers relatively few criteria to judge whether a statement is true or not. This can be identified as a main obstacle for hermeneutics: sticking to the notion of describing the world without offering criteria to distinguish better from not so good descriptions and thus to justify that a belief is true. There are still more-or-less objective

14_107-Bernard.indb 33

6/9/14 10:02 AM

34

Michael Schnegg

facts in this tradition, but they do not stand alone as sources of our knowledge claims. Their assembly by the interpreter is the critical difference. Although partly new, the approach provided by Geertz did not radically question many agreed-on fieldwork practices, including long-term stay in the community, acquiring language skills, and building rapport. In terms of data collection, quantitative techniques were largely abandoned, and the analysis of situations, including extended cases, gained significance. In terms of research design, Geertz argues that the fieldworker must go to the field with a number of questions, adapt them, and leave the field with different questions. He or she should leave when the answers do not surprise anymore.

Writing Ethnography The style of writing changed with the turn to a more interpretative anthropology. Maybe the most important shift is the turn from realistic to personalized author(ity). The “I” of the fieldworker entered the text (van Maanen 2011, 74). According to the epistemology, the information cannot be separated from the ethnographer who collected it. As a result, ethnographic writing must change to include the observer; otherwise, it would be inconsistent with the epistemology from which it emerges. However, the section that describes this personal involvement is often separated from the analysis and the presentation of the ethnographic account. Again, Geertz serves as an example. His ethnographic description and analysis of the Balinese cockfight begins with a lengthy episode about how he and his wife Hildred were caught by the police as part of a larger group watching an illegal cockfight and how that transformed their relationship with the community. After the incident, the village “opened up to them,” and they were “in” (Geertz 1973, 416). At the same time, Geertz recognizes the problem of the relationship between the object of ethnographic enquiry and its representation—be it a book, a lecture, or a film— much more profoundly than did his predecessors. While he points to the difficulties of drawing a line between the object and its representation, he clearly rejects the position that the source of anthropological knowledge is not social reality but scholarly artifice (Geertz 1973, 16). As he says: “The thread is hollow” (p. 16). In general, fieldworkers who work in a hermeneutic tradition provide much more information about: (1) their relationship to the main informants; and (2) the context in which the data collection took place. Ideally, and in the hermeneutic tradition, that would allow judging and evaluating the information against the context in which it was produced. It may also convince the audience that the author’s account is more valid or more informed than accounts written by others. It is the ethnographer’s human qualities of which the reader must be convinced. Those qualities allow him or her to recognize, reflect on, and overcome the hardship of fieldwork and the biases entailed (van Maanen 2011, 74). EXCUSE: SYMBOLS AND UNIVERSALS IN ANTHROPOLOGY

Geertz was neither the first nor the only one who turned away from a more rigid scientific approach. Interpretation and meaning became key to many of his contemporaries, including David Schneider and Marshall Sahlins. Along with Geertz, Victor Turner,

14_107-Bernard.indb 34

6/9/14 10:02 AM

1: Epistemology

35

and McKim Marriott, they are sometimes referred to as the Chicago School since all of them taught at the University of Chicago at some point in the 1970s. Through his study of the American kinship system, Schneider had concluded that many of the assumptions made in anthropological theory were deeply rooted in our own understanding of kinship. He argued that our definition of kinship is largely based on assumptions that derive from the Western folk model that “blood is thicker than water.” Through defining kinship as a category, we order the world of the people we study. We classify behaviors as kinship that the people we study may conceive very differently from us. For Schneider, the solution to the problem is not to redefine kinship by including other dimensions or by focusing on doing kin instead of being kin. He turns away from kinship as a category and his aim becomes to understand the ways different definitions of kinship shape its articulation with various ethnographic data (Schneider 1984, 132). In the end, this analysis may show that kinship is a “special custom, distinctive of European culture, an interesting oddity at worst, like the Toda bow ceremony” (Schneider 1984, 202). Sahlins launched a similar attack and equally argued against universal classifications. Cultural practices are not a reaction to material circumstances and thus cannot be explained by positivistic thinking. Nothing in the material world could explain why Americans eat beef and abandon horse meat or why women wear skirts and men trousers. Not practical reason but culture orders the world. Symbolic systems are largely reproduced through the process of production and related to other processes within culture (Sahlins 1978, 207). In Culture and Practical Reason, Sahlins (1978) argues against a utilitarian interpretation of human action and shows that the rationalist paradigm had emerged with capitalism and is unsuitable for understanding societies that operate outside the logic of the market. Even further, it is misleading to distinguish the West from the rest, and perhaps the most decisive difference is that the West lives with the illusion that society is pragmatically constructed (Sahlins 1978, 210). Unlike Geertz, Sahlins and Schneider (and other symbolic anthropologists including Turner and Mary Douglas) did not relate explicitly to hermeneutics and its epistemological assumptions. Their point of departure came from within the discipline arguing that anthropology had long imposed Western categories on other societies. In doing so, they too turned away from the search for universal laws and intercultural explanations and tried to show how particular meaning structures guide and shape the daily lives of people. The consequences for doing and writing ethnography were similar. RADICAL CONSTRUCTIVISM

Radical constructivists doubt that scientists are really finding things out. Most constructivists agree with positivists that we rely on our senses to comprehend the world. However, they disagree on what consequence this insight has. Remember that positivists called for improving validity and reliability to improve measurement. To constructivists, the senses are not the precise microscopes through which we can see the world clearly. Instead, they are more like a pair of sunglasses we cannot take off. We can only perceive the world through those glasses and have no possibility to grasp an unfiltered view of reality. If we cannot say anything about the world behind those glasses, all we can do is make statements about our perception of the world. Those can be compared

14_107-Bernard.indb 35

6/9/14 10:02 AM

36

Michael Schnegg

and combined. The argument takes up the proposition introduced by Kant that we can never transcend the bounds of our own mind to see the world as a “thing-in-itself” (Kant 1998). In this view, science is viewed as just another construction or pair of sunglasses. One of the most influential radical constructivists is Ernst von Glasersfeld (1917– 2010). Von Glasersfeld did not have a university education when he started to engage with the issue of subjectivity as part of the working group of philosopher Silvio Ceccato in Milan. The group’s work on computer-aided translations was funded by the U.S. Air Force. When the funding agency asked his research group to move to the United States, von Glasersfeld followed and was hired as a professor for cognitive science after the termination of the project. In his book Radical Constructivism (1996), von Glasersfeld gives the following example to underline his point about the limits of our perception: Imagine a man born blind walking through a forest. He will most likely construct a world of obstacles and barriers (von Glasersfeld 1998). He may distinguish these barriers according to their texture and size, but the construction he has will be very different from that of a sighted person. Von Glasersfeld argues that we all construct idiosyncratic views of the world and that this construction is not an image of reality but only stimulated through it. Thus, he does not deny the existence of a world outside our imagination. However, since there is no way of knowing how closely a particular construction of reality matches reality, it becomes useless to search for the true representation of the world. This is where it becomes radically different from what I have introduced so far. To replace the search for truth, von Glasersfeld proposes the concept of viability. Actions, concepts, and conceptual operations are viable if they fit the purposive or descriptive contexts in which we use them. Thus, in the constructivist way of thinking, the concept of viability in the domain of experience, takes the place of the traditional philosopher’s concept of Truth, that was to indicate a “correct” representation of reality. This substitution, of course, does not affect the everyday concept of truth, which entails the faithful repetition or description of a prior experience. (von Glasersfeld 1996, 14)

Von Glasersfeld uses the difference between Newton’s and Einstein’s physics to explain the concept. Einstein showed that Newton’s physics describes the world properly only for speeds significantly lower than the speed of light. It can predict the flight of an aircraft and a missile. Thus, Newton’s theory is viable though substantially limited. A viable statement is useful for predicting an outcome that will happen. To come back to the example above, a construction of the world that avoids running into trees is a useful, but limited, construction of reality (von Glasersfeld 1981, 91). According to von Glasersfeld, objectivity cannot be achieved in the classical sense. He thus redefines objectivity as the intersubjective construction of reality. This shared construction emerges from similar experiences (von Glasersfeld 1996, 119) and understandings of what a useful theory would be. Using our example again, suppose a number of blind people experience the forest in similar ways and communicate, negotiate, and construct a view of the landscape that is shared by most of them. This would be a kind of understanding of reality that transcends the individual but is still in no way objectively true. So, different collective understandings of reality are possible, depend-

14_107-Bernard.indb 36

6/9/14 10:02 AM

1: Epistemology

37

ing on the degree of similarity between people in terms of the ways they perceive and process the information about the world available to them. The idea that the world is socially constructed was largely developed by the German sociologist Alfred Schütz and later extended and deepened by Berger and Luckmann in The Social Construction of Reality (Berger and Luckmann 1966; Schütz 1964). Like von Glasersfeld, Berger and Luckmann argue that the world does not exist independently of our perceptions. At the same time, we understand our acquaintances and our kin talking about business or the family. Berger and Luckmann call this the construction of social knowledge (e.g., what everybody knows about a social world) and argue that it derives from shared experiences and is tested and reinforced through communication. What holds true for common people must hold true for scientists as well. They, too, are trapped in their inability to look beyond their sunglasses, or—as Gadamer would have it—their prejudices. For radical constructivists, the knowledge-seeking process itself becomes a matter of empirical investigations—just another construction of reality. To them, epistemology was largely a theoretical science. The insight of the constructivists transformed epistemology into a partly empirical one. Science and the production of knowledge became a key interest, and, with the works of authors like Latour, Marin, Peña, and others, an entirely new discipline was founded: science and technology studies (STS), to which anthropologists have made significant contributions (Latour and Woolgar 1986 [1979]; Martin 1998; Peña 1997). Within this context, the divide between human and nonhuman entities was blurred, and Latour argued in his actornetwork-theory (ANT) that nonhuman entities should be considered as agents that shape human behavior and constitute its reality (Latour 2005). Like Latour, Michel Foucault analyzed the relationship between knowledge and power. He argues that until the eighteenth century, power was exerted through the rulers and their ability to kill. This has since shifted, and power is now exercised through knowledge that defines bodily issues including mental illness, reproduction, and health status. The term “biopolitics” refers to this new practice of power through the regulative mechanisms of health and education policies, and equally important, through shaping public discourses. While the powerful shape the discourse, the discourse again defines what is powerful. In this sense, postmodernism has a stark political perspective. Postmodernists argue that the purported value freedom and the denial to engage politically make science a silent supporter of the ruling powers and existing social inequalities. In this perspective, social research must be politically committed and help deconstruct those discourses and the powers that constitute them (Foucault 1998, 136, 2008). While these general philosophical positions influenced anthropological thinking, the impact was most profound in analysis of the fieldwork situation and the question of how to represent the Other. POSTMODERNISM IN ANTHROPOLOGY

Epistemological Positions A tipping point of the discipline was the posthumous publication of Malinowski’s diaries. They revealed an unsympathetic and often impatient fieldworker. Even more, they showed that Malinowski was well aware of the role he played in representing the Trobrianders. He put it frankly when he wrote on December 1, 1917: “Feeling of

14_107-Bernard.indb 37

6/9/14 10:02 AM

38

Michael Schnegg

ownership: It is I who will describe them or create them” (Malinowski 1967, 140). This and other passages of his diaries allowed some to fundamentally question the image of the analytical fieldworker who—unlike the travelers and the natives—was able to describe culture in objective terms, an image that Malinowski himself had created through his methodological claims and the realistic style of writing he employed (Clifford and Marcus 1986). In the debate that followed, it became evident that the fieldworker plays a very active—and sometimes even conscious—role in shaping the representation of the people and the phenomena studied. More recently, Alvarez Roldán has questioned this radical interpretation. He compared the structure and the content of Baloma, Malinowski’s first publication—written after 10 months of fieldwork—and the corresponding field notes. Alvarez Roldán argues that Malinowski clearly linked fieldwork, field notes, and the final report in a way that is reliable and valid (Alvarez Roldán 2002). A second stimulus for questioning the validity of ethnographic knowledge was Derek Freeman’s book about Margaret Mead and her work on Samoa. Five years after Mead died, Freeman argued that Mead did not really understand her informants. The same women, when interviewed by him, swore that they had lied to Mead and that she could not tell jokes from truth (Freeman 1983; Mead 1949). Much has been said and written in support of both authors. It is futile to repeat the arguments here, but on an abstract level, the discussion demonstrated once again how insecure anthropological knowledge may be and how difficult it is to make and question claims made in ethnographic monographs. Writing Culture was the first collection of essays that systematically collected those doubts and summarized the critiques (Clifford 1986a; Clifford and Marcus 1986). James Clifford, a historian, analyzed ethnographic writings not only as field reports of faraway realities but as a specific literary genre that had developed rules and strategies to communicate its messages and to claim authority. Clifford and others showed that anthropologists tend to homogenize the phenomena they study and to ignore the intracultural variations among the Nuer, the Trobianders, and the Other in general. In his introduction to Writing Culture, Clifford coined the much cited-term “partial truths” to refer to this phenomenon and to capture it analytically. He argues that “Ethnographic truths are thus inherently partial—committed and incomplete” (Clifford 1986a, 7). The postmodern debate in anthropology thus connects to the constructivist debates by pointing out that people perceive the world idiosyncratically and that we have no way of telling which perception is more and which is less true. At the same time, they deny the overlap of those perceptions and the consequences this may have, which both von Glasersfeld and Berger and Luckmann acknowledged (Berger and Luckmann 1966; von Glasersfeld 1996). Clifford takes idiosyncrasy as a starting point to call for a new way of writing ethnographies. As an example of the creative handling of different viewpoints, he cites the monograph First-time: The Historical Vision of an Afro-American People (Price 2003 [1983]). Richard Price, who had worked in Saramaka for a couple of years before starting the project, aims to reconstruct and combine the knowledge people have about their more distant past. The Saramaka had won their liberation in 1762 after a century-long fight and were among the first slaves to become emancipated in Latin America (Price 2003 [1983]).

14_107-Bernard.indb 38

6/9/14 10:02 AM

1: Epistemology

39

Price uses a twofold strategy to analyze and present his findings. In the beginning, he introduces a set of characters who told him the stories about the past. The book is divided into two parts running simultaneously across two channels that divide each page horizontally. The upper part carries the texts and the lower part the interpretation by Price, who had done extensive fieldwork in the area before turning to the project. Those stories are combined, resulting in a collection that “would amaze (and be new to) any single living Saramaka” (Price 2003 [1983], 25). Partial truths thus combine to form a totality. An earlier and not less impressive attempt to combine different voices and interpretations is the The Natural History of a Delinquent Career by Clifford Shaw (Shaw 1968 [1931]). In this astonishing book, the author combines statistical information from a census, case records from the court, live histories, and lengthy chapters written by Sidney, a delinquent, during his third year of imprisonment. Those chapters carry titles like “How I Learned to Lie and Steal” or “Becoming a ‘Big Shot’” and are not changed grammatically and orthographically. In his “tentative interpretation” (1968 [1931], 224), Shaw tries to combine these different perspectives, before he hands it over to two colleagues (Ernest Burgess and Mary Bartelme) to offer additional interpretations and discussions in the final two chapters. The two examples emphasize a shift in terms of the sources of knowledge claims. Malinowski spoke on behalf of those he studied, often representing them in modal terms rather than recognizing the internal variability. Shaw, however, uses multiple sources, statistics, informant narrative, case records, his own interpretations, and the interpretations of a well-known expert. In his book, the stories of the people studied share space with the interpretations of the ethnographer. The shift is thus from the objective single source purveyor of anthropological knowledge to a collective, multi-source set of knowledge claimants. What is at work here is not the ability to translate other cultures but rather the ability of the lone ethnographer, using a standard mode of writing and representation of the fieldwork observations, to be a justified source of knowledge claims made. This shift is toward providing multiple sources, possibly justifying multiple knowledge claims within the same piece of ethnographic writing (see also Zenker and Kumoll 2010). As we have seen, postmodernists and radical constructivists argue that we cannot describe the world in an objective manner (see Eriksen [2002] for the political dimension of the issue). In sum, postmodern anthropology has questioned our ability to translate other cultures, what Geertz proclaimed to be the anthropological endowment. The postmodern critique takes up many of the impulses from Heidegger and Gadamer. That is, all representations including science are prejudiced. Science is only one construction of reality, and not a privileged one, and must be put alongside other stories about the world. This was radicalized when not the world out there but the scientific discourses about it became the focus of the enquiry.

Methodological Implications and Writing Ethnography The postmodern critique has largely focused on the professional author and his or her authoritative role in constructing a representation of cultural phenomena. Consequently, it tried to decenter the author. To do so, it needed to develop new styles of

14_107-Bernard.indb 39

6/9/14 10:02 AM

40

Michael Schnegg

writing, and a number of different approaches have been proposed, including multivocal representations and shifting from ethnography to fiction. I have already given a number of examples of the first strategy, the multi-vocal representations (Price 2003 [1983]; Shaw 1968 [1931]). A second alternative are dialogue-based accounts. Although these forms have been long and famously proclaimed, relatively few anthropologists have actually used them (Tylor 1986). A third way that has been explored argues that the only way to get rid of the authoritative “I” of the ethnographer is to concentrate on the creative aspects positioning the account close to fiction. This transforms the ethnographic voice as one among many—and one that does not claim to be a better or more suited representation than any other. Through positioning the author center stage, the ethnographer cannot count on science as a source of power and legitimacy and becomes more vulnerable (Jacobson 1991, 119). More influential and lasting than those experiments were attempts to include the Other into the process of writing ethnographies. These and more traditional ethnographic writing can be grouped along two axes: (1) whether the author is writing about his or her own cultural context and (2) whether the author is a professional anthropologist or not (Reed-Danahay 1997, 4). The two axes produce four cases. 1. In the classical ethnographic account, the professional wrote about the Other (case 1). 2. In case 2, nonprofessionals write about their own culture. In those cases, the professional ethnographer is often involved as a coauthor. A good example of this second strategy is In the Time of Trees and Sorrows: Nature, Power, and Memory in Rajasthan, in which the U.S.-based anthropologist and her local assistant coauthor an account about the history of an Indian province. The collaboration was largely a division of labor between data collection and writing (Gold and Gujar 2002), with the assistant checking the interpretations of the anthropologist. It is obvious that the coauthor relationship in such a dyad is unlikely to be egalitarian and that this may shape the degree to which interpretations are challenged. Nevertheless, it offers a new perspective and serves as an example for a number of approaches that have been taken to include local voices into the production of texts (Bernard and Salinas Pedraza 1989; Fischer and Abedi 1990; Price 2003 [1983]; Werbner 1991). 3. In the third combination, the professional writes about his or her cultural context (case 3). Among the earliest and most famous works of those so-called autoethnographies is Facing Mount Kenya, in which Malinowski’s Kenyan student Jomo Kenyatta describes his own ethnic group, the Kikuyu people (Kenyatta 1938). Much later, in Under the Medical Gaze: Facts and Fictions of Chronic Pain, Susan Greenhalgh (2001) analyzes the eight months she spent as a patient of a highly recommended male rheumatologist, who diagnosed her fibromyalgia, a new and little understood medical disorder. With heavy medication prescribed, she drifted into a depression from which she could only escape when her former doctor convinced her that the diagnosis was wrong and that she should stop the treatment. Based on detailed diaries written during this time, Greenhalgh analyzes how she became an object of a medical system and how that medical system constructs its authority to exercise this power over her (Greenhalgh 2001; Hayano 1983; Reed-Danahay 1997). 4. The forth combination, nonprofessionals writing about other cultural situations, has not been explored in this context.

14_107-Bernard.indb 40

6/9/14 10:02 AM

1: Epistemology

41

Independent of the form of writing and representing, others have turned away from long-term anthropological fieldwork and defined the task of anthropology as one of deconstructing, in a Foucaultian analysis, dominant discourses about the world. These discourses are thought to be produced in the center of the world system and are often written in official documents. They shape and legitimize behavior at the local level or serve to suppress the marginalized. Ethnographic fieldwork that follows this line of thought takes place where these discourses are produced and reproduced including— for example, NGOs, UNESCO, and other organizations (Comaroff and Comaroff 2009; Scheper-Hughes 2000). Overall, the postmodern critique has had an impact on ethnography and on ethnographic writing. It became evident how inappropriately the internal diversity of other people is represented through generalizations about the Nuer and the Samoans. At the same time, the most radical calls for writing ethnography as poems or dialogues did not become prevalent. Ethnography has—looking at it from the present—experienced an evolution, not a revolution. CONNECTING POSITIONS WITH ARGUMENTS

Before I offer some ideas to connect the various epistemological positions, it may help to dispense with some of the common prejudices in the emotional debate about the nature and validation of anthropological knowledge. Some say: Radical constructivists and postmodernists deny the existence of a physical world. One should beat them or throw them out of a window, and they will see the world is real.

None of the approaches introduced here claims that there is no physical world surrounding us. Constructivists argue that this world stimulates our senses and the perceptions we have. However, we cannot perceive it in an unbiased way, and any attempt to do so is doomed to fail. Others respond: Positivists really believe that the world exists exactly the way we perceive it and is entirely independent of social constructions.

None of the approaches introduced here would claim that. It is entirely clear that phenomena are to a different degree constructed by human actors. While this may be less the case for a primary forest or a hill, it is certainly the case with money, with rituals, and with many other subjects of ethnographic enquiry.

Question Three: Limits of Knowledge The positions discussed differ largely in the way they access knowledge and truth (Morris 1997; Myhre 2006; Wilson 2004). The main issue in the debate is the degree to which we can sense and describe the social world as such. The issue is pretty clear for positivists: Yes, by and large we can describe the social world and need to improve our methods to better our descriptions. Adherents of hermeneutics insist that the observer cannot be separated from the information and that we have to deal with this situation

14_107-Bernard.indb 41

6/9/14 10:02 AM

42

Michael Schnegg

analytically to describe the social world. In contrast, postmodern ethnographers have argued that we all construct our own realities and that truth is idiosyncratic. Hence, it becomes useless to search for the truth value of a statement and for ways to justify it. Social constructivists are less pessimistic than most postmodern anthropologists. Von Glasersfeld (1996) argues that people construct intersubjective realities, which he refers to as objectivity. Berger and Luckmann (1966) and Schütz (1964) had developed similar positions before and offer a bridge between realistic and relativistic positions. To explain why interpretations are shared, they argue that people who grew up under similar circumstances perceive the world in similar ways. Some of it can be explained through the similar biological makeup of humans. In addition, von Glasersfeld refers to Piaget and to the power of similar experience to explain observed similarity (von Glasersfeld 1996, 119). Berger and Luckmann (1966) name the dominance of daily interactions (Alltagswelt), which is nothing more than a cipher for experiences, dayto-day interactions, and the negotiation of interpretations in social relationships (von Glasersfeld 1996, 119). Both approaches follow the same logic and argue that the world, as such, makes some perceptions more likely than others. Independent of this and following a different epistemological position, Romney and others have developed a model of culture that they called “consensus theory” (Romney and Moore 1998; Romney et al. 1986). Starting with the assumption that there is an underlying cultural truth, they argue that those who agree more know more. The model can also be used to determine when a consensus does not exist. The nonexistence of a consensus may result from: (1) each person having an idiosyncratic interpretation; (2) two groups (men and women, for example) having largely different views; (3) many sharing one view and few deviating from it; and so on (Schnegg and Lang 2008). This model was developed for analyzing very specific quantitative data, but it can serve as a blueprint for dealing with different constructions of the world methodologically and theoretically (for a somewhat comparable approach, see also Reyna 2010). Agreement may be interpreted as cultural truth, though this does not address its relations to reality as observed and experienced through other senses and data. This becomes especially evident when we are confronted with the collective denial of guilt and violence in cases of genocide (Wilson 2004). I agree that the prediscursive world exists and that it shapes our perceptions. At the same time, people—including ethnographers and informants—have different takes and produce different accounts. The main anthropological challenge now becomes to understand where and why those constructions differ and to uncover the bias they entail. As we will see below, this includes the informant, the ethnographer, and the wider academic context in which knowledge is produced. To me, this is much more challenging than stopping with Clifford’s insight that truth is partial. It is also not that different from the approach Price—cited as a best-case example by Clifford—employed: Listen to many stories and try to combine them to document what they can tell us about the underlying reality, here a past event. The postmodern critique has shown that the relationship between the ethnographer and the people interviewed is crucial and must be incorporated into the analysis. Recently, Hastrup has argued that this makes anthropological knowledge relational. It is relational because the relationship is so close that we cannot separate the objective of

14_107-Bernard.indb 42

6/9/14 10:02 AM

1: Epistemology

43

the study, the anthropologist, and the object itself. Through the intensity of the relationship, the objective gets inscribed into the object and stains the evidence we produce (Hastrup 2004, 468). The transmission of information between objects and people and between people and people is translation. In a similar manner, differentiated between three stages of “errors” in cross-cultural research are the mind of the informant, the ethnographer’s mind, and the mind of the code (Naroll 1962, 1970). The ethnographer’s sensing of the field (e.g., our observation of a festival or a landscape) is a first-order translation. In contrast, the information given by someone being interviewed includes the informant’s interpretations of the world and is a second-order translation. I use the term “bias” to refer to the modification that takes place through the translation. In any ethnographic analysis, we must acknowledge that we are confronted with information that has been translated any number of times and contains different biases. All information is thus biased—though in different degrees. How do we move on from here? To get beyond this, we can draw on the general idea proposed by Heidegger and Gadamer and encourage an approach comparable to the hermeneutic circle to understand the relationship between the world as such and the different perceptions of it. This approach contextualizes: (1) the informants and the information given by them; (2) the context, including the ethnographer, in which the information is gathered: (3) the senses and methods used to obtain it; and (4) the particular academic context in which the information is generated and analyzed. While the first three points derive more or less directly from the hermeneutic and postmodern critique of positivism, the last point has been prominently made by Bourdieu in his Homo Academicus (Bourdieu 1988). Bourdieu argues that social scientists are biased by their social origins, their adherence to a certain academic field (e.g., the specific modes of knowledge production, theories, and methodologies), and their bias as intellectuals (see also Kuhn 1970 [1962]). The latter point refers to the fact that scientists tend to perceive the world as a spectacle to be interpreted rather than a set of concrete problems to be solved (Bourdieu and Wacquant 1992, 39). These four points combined—position of the informant, context of the information, methods used, and academic context—provide the basis for a reflexive anthropology that allows us to understand and interpret the biases involved and to deal with the relational nature of anthropological knowledge that Hastrup identified (Hastrup 2004). To return to an earlier metaphor, we can only see the world through sunglasses, but the glasses are not all the same, and we have other senses. The property of those glasses is also not randomly distributed but structured by the experience and the learning of the people involved and by larger discourses. Not only do different stories told by different people or to different people contain different biases. The same holds true for different sources of evidence—like observations and introspection—that produce different accounts of the world. Comparing and communicating the statements we make, and combining them with information gained by other senses, allows us to understand the strength of those sunglasses, to incorporate this in the interpretation, and, eventually, to grasp the world beyond them by assessing the biases.

14_107-Bernard.indb 43

6/9/14 10:02 AM

44

Michael Schnegg

Communicating and reflecting this process is a key precondition to make the translation intersubjective and reliable, thus distinguishing it from art and other ways to make sense of the world. This understanding has a number of specific methodological consequences: 1. The approach calls for an actor-oriented methodology, in which we start from the individual actors of the world. We use the structural information we have to understand what may have brought the actor to tell this specific story, be it personal involvement, position in the kinship network, economic position, and the like. 2. This also implies triangulation or a mixed methods approach, using as much different data as we have. All information is biased in specific ways, and we can only understand the world behind the bias if we analyze and understand it in an interactive and reflexive way (Denzin 2005). 3. This approach would profit from making use of all our senses in ethnography, not only our eyes and ears—a strategy that Spittler termed “deep participation” (Spittler 2001).

The reflexive approach taken here has some similarity with what has become known as “critical realism” (Bhaskar 1997 [1975]; Brereton 2004). The term must not be confounded with “critical rationalism,” coined by Popper (1969a). Critical realism developed as a way to overcome some of the pitfalls of realism and relativism. It argues that the structures that underlie human behavior cannot be perceived. Different structures interact and produce specific instances of behavior. This is the only way they come to be. As Porter put it, if we accept the causal criteria of reality that to be is to be able to do, the very existence of those effects demonstrates the reality of the underlying structures (Porter 1993, 593). The analytical task becomes getting from the behavior (e.g., a murder of a migrant in the subway) to the underlying structures (e.g., racism). Structures understood in that way are a synonym for the world as such. Critical realism focuses very much on observations (Porter 1993). The approach is compatible with what I have said so far. However, it differs in its ontological premise. While critical realism does not deny the existence of a world out there, it introduces two ontological levels: social structure and human agency. Although structures and actors both have an ontological status and are interlinked, only actors and their behavior can be observed. Davies has developed detailed guidance to show the methodological implications of this approach. Many of these techniques can be helpful in deciphering the four biases introduced above (Davies 2008, 18).

Question Two: Sources of Knowledge All of these discussed approaches share the understanding that knowledge is constructed empirically. Some authors have argued that ethnography unites even the most divergent theoretical approaches in anthropology (Mintz 2000; Murdock 1971). This could have an epistemological basis and be related to the subject of our enquiry: It is hard to claim that one can understand/explain the cultural Other by ratio or introspection alone. In the anthropological debate, the most important divide runs between understanding and explanation as two strands of the empirical position. But how are they related? Stegmüller (1979) argues: (1) Conceptually, both procedures are not polar

14_107-Bernard.indb 44

6/9/14 10:02 AM

1: Epistemology

45

opposites. A valid contrast to explanation (embedding a case in a larger, generalizing model, typically using laws or law-like statements) is description (singular statements specifying what the case is in particular entities). (2) The dichotomy of Verstehen/explanation does not lead to a fertile classification. Explanation is possible in the medical sciences, psychology, economics, and other social and behavioral sciences, including anthropology; interpretation also plays a role in the natural sciences. In addition, not all natural sciences can conduct experiments (e.g., astronomy), and they are not confined to the invariant aspects of their subject matter (as Dilthey had it), but take history into account as well (e.g., natural history in biology or astronomy). So the simple distinction between the natural and the human sciences according to explanation and Verstehen disappears. (3) A more productive methodology in social sciences attempts to integrate Verstehen and explanation. In this new perspective, we can first use (empathic) understanding to develop hypotheses on the subjective meanings of actors and then consider meanings as well as the conditions of the larger context in which actors are embedded in more encompassing explanations of social and cultural phenomena (Stegmüller 1979). In this respect, some interesting methodological ideas have been put forward by sociologists, most notably Esser (1991, 1993, 1996), Lindenberg (1989, 1990, 1992), and Lindenberg and Frey (Lindenberg and Frey 1993). Following some of Weber’s early concerns, these analysts begin with the “logic of the situation” at the level of individual actors: Using Verstehen, they investigate the actors’ view of their situation, including cultural preferences, which yields information on actors’ goals and expectations. Then analysts study the economic, political, and social restrictions of the context that enable and constrain individual action. Finally, they use a refined rational actor model (as a law or law-like statement) to explain the actors’ strategic actions. They also trace the purposeful and unintended collective outcomes of individual choices on the larger system. The refinement of the rational actor model refers mainly to the fact that the economizing scheme has to incorporate social constraints to avoid the flaw of the undersocialized actor and has to include social approval in addition to material well-being as a maximizing goal (Lindenberg 1990, 1996). In a diachronic perspective, one can study the embeddedness of actors in previous institutional and cultural arrangements, and, contrastingly, the emergence of new norms and institutions as an effect of present actions (Cashdan 1990; Ensminger 1992; Ensminger and Knight 1997; Netting 1993). Thus, in a major reconstruction, sociologist Hartmut Esser (1991) demonstrates how the Verstehen-oriented social theory of Alfred Schütz (1964) corresponds with rational choice theory and can be incorporated into rational choice explanations. These social theories use empathic Verstehen as a bridge between a very general and abstract social theory, on the one hand, and information on the concrete decision situation of actors, on the other. Siegwart Lindenberg discusses heuristic strategies for cross-fertilizing these two approaches to arrive at more precise and more realistic, but still sufficiently general explanations (Lindenberg 1992, 1996). This theoretical approach challenges the former inappropriate distinction between Verstehen and explanation and the accompanying idea that both procedures exclude each other. Better theories capture both subjective meanings (i.e., the actors’ views

14_107-Bernard.indb 45

6/9/14 10:02 AM

46

Michael Schnegg

and social constraints and the institutional context and shared beliefs prevailing in society at a certain time). By integrating both kinds of information, they arrive at more grounded, in-depth explanations. This links to a much larger debate about structure and agency (Ortner 1984, 2006). Although it is beyond the scope of this analysis to give a detailed overview of this discussion, I want to underline that this counts among the most challenging and important problems in social theory. I have taken it up here under the rubrics of methodology (the link between micro/macro and Verstehen/explanation) and epistemology (critical realism), but it must also be seen as a theoretical problem that deals with the question of how structures enable and constrain social action. Historian William Sewell has offered a model to conceptualize the duality of structure and agency building on the works of Giddens and Bourdieu. The latter two both proposed that structure and agency are not opposites but two sides of a coin (Bourdieu 1977; Giddens 1979). Sewell defines structures as follows: “Structure, then, should be defined as composed simultaneously of schemas, which are virtual, and of resources, which are actual. Structures may only be perceived through social action, as Porter argues” (Sewell 1992, 13). Sewell goes on to argue that structures are not static and that many of the existing theories fail to explain how structures can change. This is where agency comes in, since “the capacity to transpose and extend schemas to new contexts . . . is inherent in the knowledge of cultural schemas that characterizes all minimally competent members of society” (1992, 18). This capacity to transform structures again relates to the resources an actor has access to. This model opens up a way to link micro and macro phenomena, structure, and agency. Social action is carried out on the basis of a certain interpretation of the situation (logic of the situation) and with certain resources and schemata at hand. It can only be understood in this context. However, when aggregated, it may lead to phenomena, like the transformation of a schema, that can be explained. Esser has referred to this type of explanation as “deep explanation.”

Question One: Nature of Knowledge As we have seen, in the anthropological debate the question about the nature of knowledge is closely related to the question about its limits. While the philosophical debate focuses on the question of what needs to be added to true beliefs to turn them into knowledge, the main anthropological concern is to justify that a belief is true and to provide better justifications than our colleagues do. I have referred here to a number of strategies commonly employed in an empirical tradition. This is where research design and methodologies enter the game. Throughout this chapter, I have argued that justification can be improved by using multiple data and multiple senses, by ensuring transparency, and by reflecting on biases that may be entailed in the translations and perceptions of both the ethnographer and the informant. This will ultimately lead to reliability and intersubjectivity and thus to statements that are more likely to be true than others. This also means we need to reflect on how our attempts to validate a statement as true translates into specific research designs and methodologies. I advocate making

14_107-Bernard.indb 46

6/9/14 10:02 AM

1: Epistemology

47

these linkages more explicit and broadening the epistemological debate in anthropology beyond the current focus on the relationship between the ethnographer and the informant. This would include analyzing the biases that occur at different stages of knowledge production and calls for enhancing methodologies, improving transparency, and reflecting about the role of all actors involved. The insight that we cannot have the whole truth should not make us despair that we cannot have any truth at all (Pina-Cabral 2009). THE FUTURE OF KNOWING IN ANTHROPOLOGY

Anthropologists have long faced a challenge that affects the way we produce and conceive of knowledge: Our informants may read our work and may or may not agree with our claims. Today, Internet-based technologies make it relatively easy and inexpensive to draw a larger public into debates about the validity of anthropological knowledge. David Mosse (2005), for example, describes the quarrels that surrounded the publication of his ethnography of a British NGO operating in India. People involved in the NGO, especially many of his former colleagues, urged him to drop some of his claims and to substantially rewrite the book (Mosse 2006). In his later analysis of the dispute, Mosse shows that the separation between field and desk is increasingly blurred and that our writing influences our social relationships with informants and collaborators—possibly strengthening, damaging, or even severing relationships in the field (Mosse 2006). This vanishing gap between field and desk is likely to start even earlier in future research. Many of our publications are now accessible via electronic sources, and more and more of our informants are connected to Internet devices, providing new opportunities to contextualize the ethnographer and know about his or her prior work, publications, and postings. At the same time, blogs and other publishing platforms open up new ways to put our knowledge claims to the test. These developments raise questions about the anticipated perception of our research by research subjects, which may impact the way knowledge is produced and presented in ways we are only beginning to explore. Thus, for the next generation of anthropologists, Bourdieu’s (1988) call to reflect critically on the structure of academic knowledge production takes on new significance and urgency. ACKNOWLEDGMENTS

This chapter was written on the basis of the chapter authored by the late Thomas Schweizer in the first edition of this book. Michael Schnegg thanks Julia Pauli for the interdecadal discussions on the topic and João Baptista, Hartmut Lang, Ted Lowe, Julia Pauli, Erwin Schweitzer, and Anastasia Weiss for their helpful comments and discussions about an earlier version of the chapter. The editors and an anonymous reviewer provided valuable comments and pointed to arguments I had omitted. REFERENCES

Achinstein, P., and S. F. Barker, eds. 1969. The legacy of logical positivism. Baltimore: Johns Hopkins Press. Albert, H. 1994. Kritik der reinen Hermeneutik. Tübingen, Germany: Mohr.

14_107-Bernard.indb 47

6/9/14 10:02 AM

48

Michael Schnegg

Alvarez Roldán, A. 2002. Writing ethnography: Malinowski’s fieldnotes on Baloma. Social Anthropology 10: 377–93. Beckermann, A. 1997. Einführung in die Logik. Berlin: de Gruyter. Berger, P. L., and T. Luckmann. 1966. The social construction of reality: A treatise in the sociology of knowledge. New York: Doubleday. Bernard, H. R. 2011. Research methods in anthropology: Qualitative and quantitative approaches, 4th ed. Lanham, MD: AltaMira. Bernard, H. R., and J. Salinas Pedraza. 1989. Native ethnography: A Mexican Indian describes his culture. Newbury Park, CA: Sage. Bhaskar, R. A. 1997 [1975]. A realist theory of science. London: Verso. Boas, F. 1961 [1940]. Race, language and culture. New York: Macmillan. Bohman, J. 1991. New philosophy of social science. Oxford: Polity Press. Bourdieu, P. 1977. Outline of a theory of practice. Cambridge: Cambridge University Press. Bourdieu, P. 1988. Homo academicus. Stanford, CA: Stanford University Press. Bourdieu, P., and L. Wacquant. 1992. An introduction to reflexive sociology. Chicago: University of Chicago Press. Brereton, D. P. 2004. Preface for a critical realist ethnology, Part I: The schism and a realist restorative. Journal of Critical Realism 3: 77–102. Bunnin, N., and E. P. Tsui-James, eds. 1996. The Blackwell companion to philosophy. Oxford: Blackwell. Campbell, B. G. 1970. Natural selection as an epistemological model. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 51–85. New York: Natural History Press. Campbell, B. G., and D. Fiske. 1959. Convergent and discriminant validation by the multitraitmultimethod matrix. Psychological Bulletin 56: 81–105. Cancian, F. 1992. The decline of community in Zinacantán: Economy, public life, and social stratification, 1960–1987. Stanford, CA: Stanford University Press. Carnap, R. 1963. Intellectual autobiography. In The philosophy of Rudolf Carnap, ed. U. A. Schilpp, 1–84. La Salle, IL: Open Court. Cashdan, E. 1990. Information costs and customary prices. In Risk and uncertainty in tribal and peasant economies, ed. E. Cashdan, 259–78. Boulder, CO: Westview. Clifford, J. 1986a. Introduction: Partial truths. In Writing culture: The poetics and politics of ethnography, ed. J. Clifford and G. E. Marcus, 1–26. Berkeley: University of California Press. Clifford, J. 1986b. On ethnographic allegory. In Writing culture: The poetics and politics of ethnography, ed. J. Clifford and G. E. Marcus, 98–121. Berkeley: University of California Press. Clifford, J., and G. E. Marcus, eds. 1986. Writing culture. Berkeley: University of California Press. Cohen, J. H. 1999. Cooperation and community: Economy and society in Oaxaca. Austin: University of Texas Press. Comaroff, J. L., and J. Comaroff. 2009. Ethnicity, Inc. Chicago: University of Chicago Press. D’Andrade, R. 1986. Three scientific world views and the covering law model. In Metatheory in social science. Pluralisms and subjectivities, ed. D. Fiske and R. A. Shweder, 19–41. Chicago: University of Chicago Press. Dancy, J., E. Sosa, and M. Steup, eds. 2010. A companion to epistemology. Chichester, UK: WileyBlackwell. Davies, C. 2008. Reflexive ethnography. London: Routledge. Denzin, N. 2005. Sociological methods: A sourcebook. Chicago: Aldine Transaction. Dilthey, W. 1900 [1894]. Ideen über eine beschreibende und zergliedernde Psychologie. In Gesammelte Schriften, Vol. 5, ed. B. Groethuysen, 139–240. Stuttgart, Germany: Teubner.

14_107-Bernard.indb 48

6/9/14 10:02 AM

1: Epistemology

49

Ensminger, J. 1992. Making a market: The institutional transformation of an African society. The political economy of institutions and decisions. Cambridge: Cambridge University Press. Ensminger, J., and J. Knight. 1997. Changing social norms: Common property, bridewealth, and clan exogamy. Current Anthropology 38: 1–24. Eriksen, T. H. 2002. Confessions of a useful idiot, or why culture should be brought back in. LBC Newsletter (Uppsala University), Uppsala, Sweden. Esser, H. 1991. Alltagshandeln und Verstehen. Tübingen, Germany: Mohr. Esser, H. 1993. The rationality of everyday behavior. A rational choice reconstruction of the theory of action. Rationality and Society 5: 7–31. Esser, H. 1996. Die Definition der Situation. Kölner Zeitschrift für Soziologie und Sozialpsychologie 48: 1–34. Fischer, M. M. J., and M. Abedi. 1990. Debating Muslims: Cultural dialogues in postmodernity and tradition. Madison: University of Wisconsin Press. Fiske, D. 1986. Specificity of method and knowledge in social science. In Metatheory in social science. Pluralisms and subjectivities, ed. D. Fiske and R. A. Shweder, 61–82. Chicago: University of Chicago Press. Flis, A. 1988. Cracow philosophy of the beginning of the twentieth century and the rise of Malinowski’s scientific ideas. In Malinowski between two worlds. The Polish roots of an anthropological tradition, ed. R. F. Ellen, E. Gellner, G. Kubica, and J. Mucha, 105–27. Cambridge: Cambridge University Press. Foucault, M. 1998. The history of sexuality: The will to knowledge. London: Penguin Books. Foucault, M. 2008. The birth of biopolitics. New York: Palgrave Macmillan. Freeman, D. 1983. Margaret Mead and Samoa: The making and unmaking of an anthropological myth. Cambridge, MA: Harvard University Press. Fuchs, M., and E. Berg. 1995. Phänomenologie der Differenz. Reflexionsstufen ethnographischer Repräsentation. In Kultur, soziale Praxis, Text, ed. by E. Berg and M. Fuchs, 11–108. Frankfurt Suhrkamp. Gadamer, H.-G. 2004. Truth and method. London: Continuum. Geertz, C. 1973. Thick description: Toward an interpretative theory of culture. In The interpretation of cultures: Selected essays ed. C. Geertz, 3–30. New York: Basic Books. Geertz, C. 1975. The interpretation of cultures. London: Hutchinson. Geertz, C. 1983. Local knowledge. New York: Basic Books. Geertz, C. 1988. Works and lives. Stanford, CA: Stanford University Press. Geertz, C. 1995. After the fact. Cambridge, MA: Harvard University Press. Geier, M. 1993. Der Wiener Kreis. Reinbek, Germany: Rowohlt. Geldsetzer, L. 1989. Hermeneutik. In Handlexikon zur Wissenschaftstheorie, ed. H. Seiffert and G. Radnitzky, 127–39. Munich: Ehrenwirth. Giddens, A. 1979. Central problems in social theory: Action, structure and contradiction in social analysis. Contemporary social theory. Basingstoke, UK: Macmillan. Gingrich, A. 2009. Evidence in socio-cultural anthropology. Limits and options for epistemological orientations. Social Analysis 53: 177–90. Gold, A. G., and B. R. Gujar. 2002. In the time of trees and sorrows: Nature, power, and memory in Rajasthan. Durham, NC: Duke University Press. Greenhalgh, S. 2001. Under the medical gaze: Facts and fictions of chronic pain. Berkeley: University of California Press. Hastrup, K. 2004. Getting it right. Knowledge and evidence in anthropology. Anthropological Theory 4: 455–72. Hayano, D. M. 1983 Poker faces: The life and work of professional card players. Berkeley: University of California Press.

14_107-Bernard.indb 49

6/9/14 10:02 AM

50

Michael Schnegg

Heidegger, M. 1962. Being and time, trans. J. Macquarrie and E. Robinson. London: SCM Press. Heidegger, M. 2000. Introduction to metaphysics, trans. G. Fried and R. Polt. New Haven, CT: Yale University Press. Hekman, S. 1983. From epistemology to ontology: Gadamer’s hermeneutics and Wittgensteinian social science. Human studies 6: 205–24. Hirsch, J. S. 2003. A courtship after marriage: Sexuality and love in Mexican transnational families. Berkeley: University of California Press. Hollis, M. 1994. The philosophy of social science. Cambridge: Cambridge University Press. Jacobson, D. 1991. Reading ethnography. Albany: SUNY Press. Janik, A., and S. Toulmin. 1973. Wittgenstein’s Vienna. New York: Simon and Schuster. Kant, I. 1998. Critique of pure reason, trans. and ed. P. Guyer and A. W. Wood. Cambridge: Cambridge University Press. Kenyatta, J. 1938. Facing Mount Kenya. London: Secker and Warburg. Kroeber, A. L. 1948. Anthropology. New York: Harcourt Brace. Kuhn, T. S. 1970 [1962]. The structure of scientific revolutions. Chicago: University of Chicago Press. Kuznar, L. A. 1997. Reclaiming a scientific anthropology. Walnut Creek, CA: AltaMira. Lang, H. 1994. Wissenschaftstheorie für die ethnologische Praxis. Berlin: Reimer. Latour, B. 2005. Reassembling the social: An introduction to actor-network-theory. Oxford: Oxford University Press. Latour, B., and S. Woolgar. 1986 [1979]. The construction of a fact: The case of TRF(H). In Chapter 3 in laboratory life: The construction of scientific facts, ed. B. Latour and S. Woolgar, 105–50. Princeton, NJ: Princeton University Press. Lewis, O. 1951. Life in a Mexican village: Tepoztlán restudied. Urbana: University of Illinois Press. Lindenberg, S. 1989. Social production functions, deficits, and social revolutions: Prerevolutionary France and Russia. Rationality and Society 1: 51–77. Lindenberg, S. 1990. Homo socio-economicus: The emergence of a general model of man in the social sciences. Journal of Institutional and Theoretical Economics 146: 727–48. Lindenberg, S. 1992. The method of decreasing abstraction. In Rational choice theory, ed. J. S. C. Coleman and T. J. Fararo, 3–20. Newbury Park, CA: Sage. Lindenberg, S. 1996. Constitutionalism versus relationalism: Two versions of rational choice sociology. In James S. Coleman, ed. J. Clark, 299–312. London: Falmer Press. Lindenberg, S., and B. S. Frey. 1993. Alternatives, frames, and relative prices: A broader view of rational choice theory. Acta Sociologica 36: 191–205. Lowie, R. H. 1959 [1937]. The history of ethnological theory. New York: Rinehart & Company. Malinowski, B. 1922. Argonauts of the Western Pacific. London: Routledge & Kegan Paul. Malinowski, B. 1964. A scientific theory of culture and other essays. New York: Oxford University Press. Malinowski, B. 1967 [republished 1989]. A diary in the strict sense of the term. Stanford, CA: Stanford University Press. Malinowski, B. 1993. Totemism and exogamy (1911–1913). In The early writings of Bronisław Malinowski, ed. R. J. Thornton and P. Skalník, 123–99. Cambridge: Cambridge University Press. Martin, E. 1998. Flexible bodies: Tracking immunity in American culture, from the days of polio to the age of AIDS. Boston: Beacon Press. McCabe, J. T. 2004. Cattle bring us to our enemies: Turkana ecology, politics, and raiding in a disequilibrium system. Human-environment interactions. Ann Arbor: University of Michigan Press.

14_107-Bernard.indb 50

6/9/14 10:02 AM

1: Epistemology

51

Mead, M. 1949. Coming of age in Samoa: A psychological study of primitive youth for Western civilization. New York: The New American Library. Mintz, S. W. 2000. Sows’ ears and silver linings: A backward look at ethnography. Current Anthropology 41: 169–89. Morris, B. 1997. In defense of realism and truth: Critical reflections on the anthropological followers of Heidegger. Critique of Anthropology 17: 313–40. Mosse, D. 2005. Cultivating development: An ethnography of aid policy and practice. London: Pluto Press. Mosse, D. 2006. Anti-social anthropology? Objectivity, objection, and the ethnography of public policy and professional communities. Journal of the Royal Anthropological Institute 12: 935–56. Murdock, G. P. 1949. Social structure. Glencoe, IL: Free Press. Murdock, G. P. 1971. Anthropology’s mythology. Proceedings of the Royal Anthropological Institute 1971: 17–24. Myhre, K. C. 2006. The truth of anthropology: Epistemology, meaning and residual positivism. Anthropology Today 22: 16–19. Naroll, R. 1962. Data quality control—A new research technique. New York: The Free Press of Glencoe. Naroll, R. 1970. Data quality control in cross-cultural surveys. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 927–45. New York: Natural History Press. Netting, R. M. 1993. Smallholders, householders. Stanford, CA: Stanford University Press. Notes. 1874. Notes and queries on anthropology: For the use of travellers and residents in uncivilized lands / Drawn up by a committee appointed by the British Association for the Advancement of Science. London: Stanford. Ortner, S. B. 1984. Theory in anthropology since the 60s. Comparative Studies in Society and History 26: 126–66. Ortner, S. B. 2006. Anthropology and social theory: Culture, power, and the acting subject. Durham, NC: Duke University Press. Outhwaite, W. 1987. New philosophies of social science: Realism, hermeneutics and critical theory. Basingstoke, UK: Macmillan Education. Papineau, D. 1996. Philosophy of science. In The Blackwell Companion to philosophy, ed. N. Bunnin and E. P. Tsui-James, 290–324. Oxford: Blackwell. Peña, D. G. 1997. The terror of the machine: Technology, work, gender, and ecology on the U.S.Mexico border. Austin: CMAS Books, Center for Mexican American Studies, University of Texas. Pina-Cabral, J. d. 2009. The all-or-nothing syndrome and the human condition. Social Analysis 53: 163–76. Popper, K. 1969a. Conjectures and refutations. London: Routledge & Kegan Paul. Popper, K. 1969b [1934]. Logik der Forschung. Tübingen, Germany: Mohr [Engl. transl. The logic of scientific discovery. New York 1959: Basic Books]. Popper, K. 1972. Objective knowledge. Oxford: Clarendon Press. Porter, S. 1993. Critical realist ethnography—The case of racism and professionalism in a medical setting. Sociology—the Journal of the British Sociological Association 27: 591–609. Price, R. 2003 [1983]. First-time: The historical vision of an African American people. Chicago: University of Chicago Press. Quine, W. V. O. 1961. From a logical point of view. Cambridge, MA: Harvard University Press. Radcliffe-Brown, A. R. 1957 [1948]. A natural science of society. Glencoe, IL: Free Press. Radcliffe-Brown, A. R. 1968 [1952]. Structure and function in primitive society. New York: Free Press.

14_107-Bernard.indb 51

6/9/14 10:02 AM

52

Michael Schnegg

Rao, A. 1998. Autonomy: Life cycle, gender, and status among Himalayan pastoralists. Oxford/ New York: Berghahn. Redfield, R. 1930. Tepoztlán. Chicago: University of Chicago Press. Reed-Danahay, D. 1997. Introduction. In Auto/Ethnography: Rewriting the self and the social, ed. D. Reed-Danahay, 1–17. Oxford: Berg. Reyna, S. 2010. Hard truths: Addressing a crisis in ethnography. In Beyond writing culture: Current intersections of epistemologies and representational practices, ed. O. Zenker and K. Kumoll, 163–86. Oxford/New York: Berghahn Books. Rivers, W. H. 1910. The genealogical method of anthropological inquiry. Sociological Review 3: 1–12. Romney, A. K., and C. Moore. 1998. Toward a theory of culture as shared cognitive structures. Ethos 26: 314–37. Romney, A. K., S. Weller, and W. Batchelder. 1986. Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–38. Sahlins, M. 1978. Culture and practical reason. Chicago: University of Chicago Press. Scheper-Hughes, N. 2000. The global traffic in human organs. Current Anthropology 41: 191–224. Schnegg, M., and H. Lang. 2008. Analyse kultureller Domänen. Eine praxisorientierte Einführung. In Methoden der Ethnographie, vol. 3, ed. H. Lang and M. Schnegg. www.methoden der-ethnographie.de (accessed March 20, 2013). Schneider, D. M. 1984. A critique of the study of kinship. Ann Arbor: University of Michigan Press. Schütz, A. 1964. Collected papers II: Studies in social theory. The Hague: Nijhoff. Sewell, W. H. 1992. A theory of structure—Duality, agency, and transformation. American Journal of Sociology 98: 1–29. Shaw, C. R., in collaboration with M. E. Moore. 1968 [1931]. The natural history of a delinquent career. New York: Greenwood Press. Soames, S. 2003. Philosophical analysis in the twentieth century, Vols. 1 and 2. Princeton, NJ: Princeton University Press. Spittler, G. 2001. Teilnehmende Beobachtung als Dichte Teilnahme. Zeitschrift für Ethnologie 126: 1–25. Stegmüller, W. 1979. Walther von der Vogelweides Lied von der Traumliebe und Quasar 3 C 273: Betrachtungen zum sogenannten Zirkel des Verstehens und zur sogenannten Theoriegeladenheit von Beobachtungen. In Rationale Rekonstruktion von Wissenschaft und ihrem Wandel, ed. W. Stegmüller, 27–86. Stuttgart, Germany: Reclam. Stocking, G. W. 1983. The ethnographer’s magic: Fieldwork in British Anthropology from Tylor to Malinowski. In Observers observed: Essays on ethnographic fieldwork. History of Anthropology Series, Vol. 1, ed. G. W. Stocking, 70–120. Madison: University of Wisconsin Press. Tylor, S. 1986. Post-modern ethnography: From document of the occult to occult document. In Writing culture: The poetics and politics of ethnography, ed. by J. Clifford and G. E. Marcus, 122–40. Berkeley: University of California Press. van Maanen, J. 2011. Tales of the field: On writing ethnography. Chicago: University of Chicago Press. von Glasersfeld, E. 1981. The concepts of adaptation and viability in a radical constructivist theory of knowledge. In Piagetian theory and research, ed. I. E. Sigel, D. M. Brodzinsky, and R. M. Golinkoff, 87–95. Hillsdale, NJ: Erlbaum. von Glasersfeld, E. 1996. Radical constructivism: A way of knowing and learning. London: Falmer Press.

14_107-Bernard.indb 52

6/9/14 10:02 AM

1: Epistemology

53

von Glasersfeld, E. 1998. Die radikal-konstruktivistische Wissenstheorie. Ethik und Sozialwissenschaften. Streitforum für Erwägungskultur 9: 503–11. Werbner, R. 1991. Tears of the dead: The social biography of an African family. International African Library. Edinburgh: Edinburgh University Press. Wilson, R. A. 2004. The trouble with truth. Anthropology’s epistemological hypochondria. Anthropology Today 20: 14–17. Wittgenstein, L. 1922. Tractatus logico-philosophicus. London: Kegan Paul, Trench, Trubner & Co. Zenker, O., and K. Kumoll. 2010. Beyond writing culture: Current intersections of epistemologies and representational practices. Oxford/New York: Berghahn Books.

14_107-Bernard.indb 53

6/9/14 10:02 AM

14_107-Bernard.indb 54

6/9/14 10:02 AM

CHAPTER 2

In Search of Meaningful Methods James W. Fernandez and Michael Herzfeld

FIELDPATHS: ON THE MEANING OF METHOD

Lost as a dead metaphor in the word “method” is the Greek hodos, “way, path, road.” Webster’s dictionary (1959, 1548) tells us that method is following some systematic course (Latin cursus, from currere, “to run”) of action. Why, given the contingencies and uncertainties of life’s constant negotiations of meaning, we should hope to follow any systematic course of action is not clear in the dictionary definition. This is an epistemological question requiring resolution for particular cases, and the very contingency of choice—imposed by the varying needs of different disciplines—belies the common assumption that methodology is necessarily dictated by a logic that stands outside culture and history. Questioning that assumption is profoundly anthropological, as is perhaps most strikingly evident in the anthropology of science (e.g., Rabinow 1996; Traweek 1988) and in ethnographic accounts of experimental research in the social sciences (Lave 2011). While science may pay collateral attention to “the observer effect” in their laboratory work, anthropology keeps it centrally in focus in our investigations. The problem is compounded in the analysis of meaning by the fact that the very object of analysis is also the medium for the exercise of the method. Much has been written about the virtues of various meta-languages of method, but none—certainly not in the fields of semiotics and semantics—has been plausibly upheld as context independent. In addition, the English language generates some confusion between meaning as a semantic property and meaning as intention (as in “meaning to do something”), while other languages and local conceptual frameworks may not make the clear, Cartesian significance between meaning-as-semantics and meaning-as-importance (as in “That doesn’t mean a thing”) (see Herzfeld 1985, 18). If the former confusion is further compounded by anthropology’s longstanding difficulty with the analysis or even the recognition of psychological inner states (“intentionality”; see Needham 1972; Rosen 1995), the latter can obfuscate our effective reading of local ideas about how people perform the tasks of generating and interpreting each other’s utterances and other actions. Such confusions become still more vexatious when we cross the line dividing supposedly pure scholarship from activism and engagement (Low and Merry 2010; Pink 2012) and from recognizing and prioritizing moral significance (see especially Kleinman 2006)—in other words, as scholarly detachment gives way to moral imperatives that reorganize the relationship among semantics, intentions, and significance in sometimes radical ways. 55

14_107-Bernard.indb 55

6/9/14 10:02 AM

56

James W. Fernandez and Michael Herzfeld

For all these complex and interrelated reasons, methodological specificity will, in most cases, be a clarification of the goal to which the path is thought to lead and of what is to be revealed in achieving that goal. We begin here in what might be taken as a reversion to the ideals of pure, philological scholarship by attempting some etymological reflections on key tropes buried in the central terms under discussion.1 This will emerge as a form of engaged practice. In a theoretical phase in which anthropology is still recovering from the determined resistance of classic structuralism to both historical specificity and textual philology, and in which ideals of scholarly purism are increasingly incompatible with the necessarily intense engagements of fieldwork, there are nevertheless good reasons for using etymology—which is both historical and, at least traditionally, philological—as a reflective methodological instrument. Attention to etymology (which can address nonverbal as much as lexical forms of expression [see Kendon 1995, 2000]) also shows that the anthropological study of meaning cannot rest on assumptions of pure reference, for these ignore the temporal instability of semantics. Meanings, even in the most severely lexicographical sense, rarely stay the same for long. The critical term is use, and the primary methodological challenge is to assemble as full an account as possible of the contexts of use for each item to which local commentators attribute referential meaning. In this way, the anthropologist treats the claims of referentiality (and other forms of pragmatic essentializing) as themselves elements of the social universe of symbol production empirically identifiable precisely because they have been disaggregated from the processes of routinization that protect them from challenge in everyday life. In this approach, the object of empirical investigation is not semantic accuracy, in some absolute sense, but the use that informants make of symbols and models. For example, arguments used to center on whether lineage segmentation was an accurate description of social reality or “merely” a folk model (Peters 1967; Salzman 1978). Once it became possible to treat it instead as a metaphor for political uncertainty (Meeker 1979, 222), it could also be asked who used it, under what conditions, and for what purposes (according to their own or others’ attributions of motive), and this made it possible to see segmentation as a pragmatic resource in marshalling historical memory for political purposes (Dresch 1986; Shryock 1997). It also underscored the constitutive role given to metaphor by effective and interested social actors (see, e.g., Festa 20072), instead of literal-mindedly rejecting the model it represented as inaccurate or unreal. As a result, instead of worrying about whether a social group really practices lineage segmentation, we can see how individuals manage the observed patter of political alliances by using the concepts of segmentation, agnatic solidarity, and the obligations of ritual kinship to legitimize intended or already accomplished acts of violence, loyalty, and factionalism (e.g., Herzfeld 1991, 126–29). Perhaps the most obvious locus for this kind of distinction between literalist and symbolic understandings of fact lies in the management of history, where any critique of factuality must appear to be corrosive of power—precisely the point that Vico was at pains to make in his New Science (see below). This is especially true for the interpretation of material remains, notably archaeological sites, because these physically imposing cultural objects force us to confront the fallacy of misplaced concreteness head-on. Thus, rather than asking which narratives about a historical site are correct,

14_107-Bernard.indb 56

6/9/14 10:02 AM

2: In Search of Meaningful Methods

57

we can learn a great deal more by examining how the various interpretations of that site are used by interested factions and individuals (Abu El-Haj 2001; Appadurai 1981; Bender et al. 1998; Bond and Gilliam 1994; Bruner and Gorfain 1984; Gable et al. 1992; Gable and Handler 1997; Herzfeld 1991). The literal content attributed to the narratives and the symbols deployed in their perpetuation (see Handelman 1990, 2004; Kapferer 1988) are interesting primarily for the uses to which they are put (see Roberts 1997). Much in such analyses will not be reducible to a verbal account of “what these symbols mean.” There are many ambiguities (Fernandez1965). To take a particularly salient example, does the echo of an etymological root that we can recognize mean that our informants recognize it, too? Or, more significantly, does their puzzlement when we ask them about this mean that the connection is, in some absolute sense, “not there”? Here we are particularly invoking what J. L. Austin (1971), writing of excuses, called “trailing clouds of etymology”—a motif to which we shall return. Without necessarily reviving Lévi-Strauss’s (1963) unfortunate distinction between “conscious” and “unconscious” models—in which the former were impoverished because the “natives” supposedly did not have “our” capacity to detach their immediate interpretations to the real, full significance of the social phenomena in question—we may certainly assume that much of what we acknowledge as poetic resonance in speech, visual, and olfactory signs, and indeed virtually every semiotic domain of culture would lose its evocative power, like an Aristophanes joke painstakingly dissected in the apparatus criticus, once people became too directly conscious of “how it worked.” To be sure, there are obvious aspects of meaning (Rappaport 1979, 173–221), or at least there seem to be some, but obviousness is itself culturally contingent (Douglas 1975; Miceli 1982). Moreover, much that is meaningful in social interaction and its attendant cultural configurations, shaped by and shaping of personal understandings, is not at all obvious—even to the actors. Furthermore, while pointing out the obvious is—obviously?—a primary task, providing paths by which inquiry can proceed in order to discover and understand that which is not so obvious is crucial to method in the study of meaning. In the most reductionist of terms, meaning is not exhausted by a listing of referential correspondences or denotations, and it would be a poor ethnography that failed to grapple with the innuendoes, ironies, and other connotative meanings by which people construed the intimacies and ambiguities of everyday existence. Indeed, avoiding these cloudy issues, however discomfiting they may be and however deeply they offend our own methodological desire for order, would be an open invitation to engage in linguistic and other forms of cultural incompetence. Third, the study of meaning is also, as a consequence of its dependence on being able to respond to such indefinable but undeniable problems, pronouncedly reflexive. The inquiry compels reflection on its own organizing tropes so as not to fall prey to unrecognized and unanalyzed path dependencies—that is, to teleological expectations that are concealed within the presuppositions of the methodology itself. Thus, any method for the study of meaning will entail reflecting on the tropes that appear to frame its argument in an interesting and intelligible way (Fernandez 1986). (A good example is the significance of our own cultural and “rational” distaste for the kinds of analytic disorder just mentioned: They are violations of a categorical order, and, as such, represent

14_107-Bernard.indb 57

6/9/14 10:02 AM

58

James W. Fernandez and Michael Herzfeld

the irruption of an awkward reality into the symbolic order of science—an instructive irony, given that symbolism is so often conceptually opposed to science.) The “frames of intelligibility” (Geertz 1973, 36) that we use are thus not to be taken for granted as inevitable and universally self-evident. And the particular tropes on which they depend and their difference from the tropes characteristic of other frames of intelligibility can be readily demonstrated, as in Salmond’s (1982, 82–86) comparison of European and Maori framing of knowledge claims in territorial terms (European), on the one hand, and sacred wealth terms (Maori), on the other. In this regard, the still resonant debate between Sahlins (1985, 1995) and Obeyesekere (1992) regarding the appropriate interpretation of the death of Captain Cook illustrates the trembling contingency of scholarly attempts to stabilize the shifting frames of intelligibility. Such interpretative indeterminacy speaks to what in his New Science (1984 [1744]), the eighteenth-century Neapolitan philosopher Giambattista Vico called “poetic wisdom”—wisdom about the experiential sources of our understanding and about the ways in which our knowledge is represented and organized by various devices of the imagination so as both to frame and, by performance, to fulfill what is taken and received as intelligible. This wisdom, for Vico (and, more recently, for Bourdieu [1977] and Jackson [1989]), is often lost in the mesmerizing abstractionism of intellectual life and has to be rescued from it. In particular, Vico, who deployed etymology against its normativizing uses by the official establishment to develop a critical, defamiliarizing, and consequently subversive method, argued that when thinkers lose sight of the material, social, and corporeal metaphors out of which their abstract ideas arise, they also lose a sense of the experiential grounding of their knowledge. Superficial intellectual enthusiasms and actual loss of knowledge are the effective outcomes. Vico argued that “in all languages the majority of expressions for inanimate things are made by means of allusions to the human body and its parts and to human sensations and human passions” (Vico [1984, 248] in Herzfeld [1987, 13, 169–70]). The probing of these allusions is basic to Vico’s notion of poetic wisdom. He argued that we must remember these crude origins of abstract reason if we are to avoid the ignorant and ultimately degrading conviction that our thought, with its claims to transcendence and sophistication, has escaped our corporeal, historical selves. This old point about poetic wisdom has been more recently developed in the cognitive theory of metaphor, which anchors the dynamic of the tropes in the prototypical corporeal and sensationanchored domains of human experience (the experiential gestalts of life) (Johnson 1987; Lakoff 1987; Lakoff and Johnson 1980). Arguments like this—whether of Vico or in the latter-day cognitive approach to metaphor—should be particularly congenial to anthropologists, for it enables them to lodge the sources of academic abstractions—the scaffolding of learned theorizing with all its attendant, often arrogant narcissism and solipsism—in the tropes of entirely local and embodied sensations and experiences as these are understood and articulated by the people we study. Indeed, such approaches have been developed by several writers, notably—and with various perspectives aimed at unshackling methodology from the surprisingly non-empirical demands of scientism—including Brady (2004), Classen (1993), Farnell (1994, 1995; Farnell and Graham (this volume), Howes (1991), Jackson (1989), and Stoller (1989, 1997). By juxtaposing our interlocutors’ wisdom and

14_107-Bernard.indb 58

6/9/14 10:02 AM

2: In Search of Meaningful Methods

59

theoretical capacity with the formal knowledge of academic writers, moreover, we may actually be able to develop more useful tools of analysis to decipher local meanings: The similarities and differences that emerge may help us understand the range of what people “mean by meaning” (see, e.g., Herzfeld [1985, 2005] for a juxtaposition of Jakobson’s [1960] “poetic function” with local concepts of meaning). In place of vague generalizations about meaning, we recast our theoretical frameworks as ethnographic data in a comparative project engaged by both our informants (who are usually, after all, trying to make sense of us) and ourselves. Recognizing, then, that the concept of method is itself enmeshed in a complex history of shifting signification permits a determinedly empirical insistence on the constant calibration of the methods we use to elicit data with our informants’ understanding of what constitutes appropriate discourse, social interaction, and intellectual activity. At the same time, and for the reasons that Vico was the first to recognize, any method must be in some sense both a path and the project of following that path. Against the outright rejection of ethnographic realism (see, e.g., Clifford 1983), we would prefer to recognize as realistic those approaches that, again consistently with Vico’s thought, acknowledged the limitations and inevitable distortions of human sensory perception, accepted all facts as representations of an always-mediated reality (rather than as nonexistent), and recognized that contingency (sometimes known in this context as serendipity) must both shape the content of what we know and—as Edwin Ardener famously insisted (2007)—revel in a provisionality that is always potentially the source of new insights. The overall approach that we recommend is thus dynamic and actively adaptive to circumstances rather than preordained and given. In the sense that method constitutes the criteria of relevance for its own objects of discovery, moreover, it is also constitutively performative. Following a path of investigation is a methodological performance in this active (or performative) sense: The choice of the path we follow determines and in some senses guarantees the truth that can be asserted as a consequence of following that particular path. In this sense, our recommendation converges with those forms of positivism that acknowledge the provisionality of knowledge and its dependence on an honest assessment of the circumstances that produced it—a phenomenon that is nowhere more evident than in the uncertainties of ethnographic fieldwork (see, e.g., Brady 2004, 623–28). Anthropologists now generally accept that good ethnographic field methods must provide a clear account of the field paths followed to obtain the data out of which the local knowledge and the eventual ethnography built on it are constructed (Sanjek 1990, 398–400; see also Fernandez 1993, 1996, 2006; Lave 2011)—a position that exposes the fashionable distinction between positivistic and interpretative styles of analysis as itself caught up in an identifiable, describable social rhetoric. It has been similarly argued that historical method must be critically aware of its path dependency, and, at the same time, of the degree to which it may be overly influenced by a teleological dynamic that overcommits it to final causes and a priori endpoints of pathlike reasoning (Goldstein 1976; Sewell 1996; Tilly 1997). There is a delicate balance here by which a method must be purposive without being deterministic—that is, allegedly immanent in what it is used to investigate and thus

14_107-Bernard.indb 59

6/9/14 10:02 AM

60

James W. Fernandez and Michael Herzfeld

self-fulfilling. Thus, while we seek to condition and control through pathlike procedures the presence of pure randomness and contingency in the undertaking of inquiry through the use of method, we do not thereby dispose of the ever-present possibility of contingency in the actual working out of human interaction. When method appears to have that effect, often indexed in English by passive-voice constructions (or locutions such as “This article will argue that . . .”) indicating an avoidance of scholarly agency and accountability, it has clearly strayed from the path of empirical inquiry and fallen into the trap of pure self-reference—the very danger against which Vico’s etymological subversions were intended to provide a provocative and persistent warning.3 Method implies systematization of procedure, usually leading to the goal of greater clarity or parsimony in the explanation of human affairs. Method is itself a project that recognizes the presence and influence of projects in the social and cultural affairs of those studied (Sartre 1963), including the presence of goals as immanent in these projects. But it entails caution about the a priori presence of final causes in its own procedures and its empirical claims depend on its remaining open to contingency—the combination of predictability and improvisation that characterizes the social interaction it studies. It remains answerable to what Marx, here perhaps following Vico (Tagliacozzo 1983), said in The Eighteenth Brumaire (Marx 1978 [1852], Section 1): “While men make their own history, they do not make it just as they please.” SOCIAL AND CULTURAL POETICS AS A METHOD FOR THE STUDY OF MEANING IN PERFORMANCE

We noted above that method usually works toward greater parsimony (powerful simplicity) of explanation, with the caveat that frequently enough, as far as the study of meaning is concerned, what is produced is an understanding of the greater complexity in the processes by which meaning is configured. The notion of poetic wisdom is sensitive to that complexity and thereby challenges parsimonious accounts, so to speak, to account for their parsimony—for their reductiveness, as it often turns out to be. For poetic wisdom is not the study of meaning in the abstract. Instead, it attends to the dynamic meaning in actual performance—the means by which significance is conveyed in actual social life. In this, it anticipates historically the development of use or action theories of meaning as opposed to an empirically indefensible reliance on the notion of pure reference. Our view is that poetic principles guide all effective and affective social interaction. By this we mean that the mind engaged in action in the social world is as much if not more a poetic instrument as it is an abstracting and concept-forming instrument (Gibbs 1994). Indeed, to the extent that there can be a poetics of precision, the effective and the affective are not necessarily distinguishable entities and only an antiempirical, preemptive logic would allow us to presuppose that they were. Conversely, affectations of numerical or monetary imprecision—as when a merchant pretends not to weigh produce carefully, or when a customer ostentatiously avoids looking at the scale—demonstrate disdain for calculation, which is considered socially disruptive. In fact, effective performances of such careful imprecision, especially by the economically disadvantaged, may be a way of creating social solidarity (e.g., within a social class or a profession) on the basis of which people can help each other—a form of social capital

14_107-Bernard.indb 60

6/9/14 10:02 AM

2: In Search of Meaningful Methods

61

that also reduces their dependence on often hostile market mechanisms in obtaining cheaper goods and services (Herzfeld 1991, 168–73). The conveying of significance in action is the heart of the matter that should concern us. And this requires us to focus on signs themselves, that is, on the wisdom that the various traditions of semiotic and symbolic inquiry in anthropology provide. Here we are obliged to recognize distinct but interrelated traditions of understanding: semantic anthropology, on the one hand (Crick 1976; Parkin 1982) (or semiotic ethnography [Herzfeld 1983]), and symbolic anthropology, on the other (Colby et al. 1981; Dolgin et al. 1977; Peacock 1975). We also must recognize that there have been two distinct but interrelated deployments of poetic wisdom: social poetics and cultural poetics. The first, social poetics, is grounded in a recognition of the role of cultural form in social life. Building on several strands—the recognition of cultural form (e.g., modes of hierarchy or of inclusion and exclusion) in interaction (Fernandez 1977a), the traditions of “use” or “action” theory in semantics (analogous to practice theory in social anthropology and largely derived from the work of J. L. Austin [1971, 1975 (1962)]), Jakobson’s concept of “poetic function” (1960), and a long and complex tradition of rhetorical analysis in several semiotic domains and intellectual traditions, social poetics is the study of how social actors play on their societies’ conventions in ways that, if successful, increase those actors’ power or status, while also sometimes contributing to cultural change. Some critics have charged that this approach can only work in agonistic societies where the competition over the control of good form is palpable. We answer that it is possible to perform nonperformance, to be boastfully modest and demure, and to make an extravagance out of restraint (whether social or aesthetic). It is vital, in pursuing a social poetics, to detach our methodology from assumptions (such as those that have dogged studies of Mediterranean or Melanesian cultures) about some cultures being “more agonistic” than others. Such assumptions too easily reflect the social norms of our own cultural milieu, thereby illustrating perfectly what we have just said about the dangers of solipsistic methodology. And there may be no actual competition present in a given situation. But the creative deformation of norms—what in another context Boon (1982) has called “the exaggeration of culture”—must always be a response to people’s understanding of the value of cultural capital (see Bourdieu 1984). Contestation thus lies at the root of the realization of human existence, even when we live in societies where the dominant ideology is a denial of any such thing. The term “cultural poetics” surely does not deny the contestatory and agonistic politics of everyday life. That dimension is evoked in the subtitle of the collection that has stimulated so much recent inquiry into cultural poetics in anthropology, Writing Culture (Clifford and Marcus 1986). Notable subsequent instances of the cultural poetics approach (Crain 1989; Limón 1994) seek to keep the practical politics of everyday life, the war of position as Limón calls it (evoking Gramsci), in focus. But the practitioners of cultural poetics desire in their interpretations to focus, to quote Limón again, “on the aesthetically salient and culturally embedded textualities and enactments” (p. 14) of social life, which is to say the cultural preoccupations of that life. They seek to probe the “political unconscious”—in Jameson’s (1981) phrasing, the “socially produced, narratively mediated and relatively unconscious ideological responses of people” to recurrent problematics and challenges to their situation (chap. 1).

14_107-Bernard.indb 61

6/9/14 10:02 AM

62

James W. Fernandez and Michael Herzfeld

The unconscious may be analytically opaque, to say the least, but failing to recognize its existence, however we label it, is equally unjustified methodologically and may lead to false assumptions about the very nature of knowledge. If we examine what works as ethnographic description, we see that our own ability to grasp cultural specificities hinges on a willingness to suspend, with all deliberate and necessary provisionality, disbelief in the possibility of understanding and expressing intersubjective relations (Leavitt 1996). Turning to the work of novelists is sometimes a useful compromise inasmuch as their imagination can help us reconstruct culturally acceptable understandings of motive and other psychological inner states without falling into the trap of romanticizing empathy (see Herzfeld 1997a). The apparent concreteness of language is a major source of this fallacy (misplaced concreteness), and this overdetermination is nowhere more evident than in discussion of (or, worse, careless uses of the term) poetics. Cultural poetics has been so heavily associated with textuality, for example, that its practitioners have often felt obliged to add the words “and politics,” whereas a clearer definition of poetics itself should have made that addition redundant: inasmuch as poetics is about creativity, it is inherently also about power.4 This last point requires some elaboration. Taken together, the approaches of social poetics and cultural poetics constitute a contemporary and comprehensive program— constellation of paths of inquiry—for the study of meaning. From the first, however, they have faced two criticisms: that they are too language based or logocentric and that they either neglect or ignore the dynamic interdependence between meaningfulness—the significance present in symbolic activity, we might say—and power relations. Above, we tried to make clear that poetics and politics have been mutually complementary from the first. Still, the lingering association of poetics with a purely textual reading of the Aristotelian discussion of rhetoric and dramatic statement, or of Jakobson’s poetic function, has tended to confine the term to the verbal and to exclude it from the raw, as opposed to the rhetorical, exercise of power. Meaning is never reducible to verbal form or restricted to language itself. We nevertheless wish to emphasize that the reflexive and recursive nature of language are main challenges to the complexity through which we must find our way and that, pragmatically, the medium of language must often be our principal guide; we should not forget that even the most determined (and determinist) materialists gain access to most of their data through conversation and archival work. We further argue that political dynamics, the play of power, even raw power, has its inescapable rhetorical or poetic component. The unjustly neglected writings of F. G. Bailey are suggestive in this regard, for they show that an inadequate performance can be constitutive of major shifts in the balance of power (Bailey 1973, 1983, 1988, 1994; see also Paine 1981). We “do things with words” (Austin 1975 [1962]) and with all manner of other signs as well. It would be very easy to read Bailey as exclusively a political analyst, but his writings suggest the mutual entanglement of power and rhetoric in the broadest sense of the latter term, More generally, we may say that both narratives and carefully observed sequences of events can, at the very least, allow us to track probable causalities, which become significant especially as they are recognized by social actors. The transgressive move of using poetics as fundamental to a social science in a sense explicitly broadened and even separated from that of poetry is, in part, intended to animate if not shock practitioners into recognizing the technical features by which

14_107-Bernard.indb 62

6/9/14 10:02 AM

2: In Search of Meaningful Methods

63

we grasp the place of the rhetorical and the expressive in human affairs. The common, restrictive, and usually logocentric sense by which the term poetics is commonly understood should not obscure the deeper and more general semiotic properties and processes of human interaction that a social and cultural poetics seeks to evoke and appraise. While the use of the term “poetics” to denote the performative aspects of textual production in anthropology (Clifford and Marcus 1986) is consistent with more general usage, and already shifts the search for poetic qualities away from the philological confines of poetry, we argue for a more radical expansion of the term to include the entire gamut of semiotic devices through which human beings not only express but also create new meanings and even new institutions and conventions. Moreover, while verbal poetry may suggest new ways of conveying the inchoateness of social life (e.g., Brady 2004) and plays a more important role in organizing it than is often acknowledged (but see, e.g., Fernandez 1998; Friedrich 1996; Mintz 1997), it is far from exhausting the range of practical possibilities suggested by the recognition of what, for want of a better term, we might dub “poeticity.” Such an approach illuminates the way that the main entities of social life, the bodily beings logocentrically denoted by the social pronouns, inchoate in their base natures, obtain social identity and the possibilities of extra-bodily social experience by poetic processes of predicating physically sensate signs on themselves. (Fernandez 1974). We speak of a “body politic,” and we can trace its vicissitudes in the disciplines and constraints imposed on individual bodies in its name (Cowan 1990). We can also approach physical monuments as ornaments of the body politic, analogous to those of the body personal, thereby treating the most obviously material manifestations—architecture, for example—as tropes of collective identity that are also personal and familial and that become understandable as such through the transmission of knowledge about their appropriate construction (e.g., Marchand 2001, 2009) as much as through their subsequent use and modification. Johnson’s (2010) analysis of Thai temple architecture in northern Malaysia explores the spatiality of what Herzfeld (2005) has called the “cultural intimacy” of institutions, arguing that widely divergent interpretations of how appropriate local versus “central Thai” forms of temple space meant that the embarrassment of difference was not always obvious and might sometimes look like pride. The success of an internal self-image is always mixed with a certain sense that it would not be successful in other contexts (and perhaps represents what Ardener [1975, 25] calls “englobing” in that sense). Irony and sarcasm may as easily be built into architecture as self-aggrandizement; if etymology, as Vico argued, can be turned against the grandiloquent uses often reserved for it by official historiography, the same can certainly be said of an arena that literally structures everyday lives but that is not so obviously open to direct contestation as a verbal claim might be. Indeed, we would argue, the approaches we develop here speak more directly to the materiality of experience than do many of the approaches that are self-styled as “materialist.” This appropriately ironic assessment is especially applicable in the sense that the embodied nature of human knowledge is embedded in relations of causality— especially as performativity—that are often channeled precisely through those phenomena that materialists dismiss as merely symbolic and therefore epiphenomenal (e.g., Bloch [1975] and, in curiously similar terms, Harris [1974]).

14_107-Bernard.indb 63

6/9/14 10:02 AM

64

James W. Fernandez and Michael Herzfeld

Poetic Function A social poetics, and by extension a cultural poetics, can usefully begin from the concept of “poetic function,” as originally formulated by Jakobson (1960). His definition was ostensibly a purely linguistic one, and all the material adduced to explore it was taken from the verbal domain (Waugh 1976), but the underlying principles— as must follow from Jakobson’s communicative model—were unequivocally social. Methodologically, this alerts us to the possibility of a Jakobsonian classification of the uses of metaphors and other tropes, defined in terms of social actors (sender, destination, audience), that would be sufficient to ground the inchoateness and ambiguity of so much figurative language in the observed details of social interaction. It also suggests that similar principles might obtain for the entire range of human semiotic domains. Moreover, formal properties emerge more from the social relationships among actors than from the sometimes “fuzzy” semantics of their language itself (see also Bourdieu 1977, 113). Jakobson (1985), for example, was much concerned with the poetic uses of symmetry, chiasmus, parallelism, and opposition—idioms ideally suited to, and perhaps originating in, the oppositional aspects of social life, and powerfully iconic of that life, but—by the same logic—often also conducive to a loss of referential meaning. (This consideration also undoubtedly undergirded his occasional collaboration with Lévi-Strauss, in whom it generated a more abstract understanding of the concept of complementary opposition.) Parallels between social relations and aesthetic elaborations, as in architecture (or, better, architectonics) (Blier 1994; Fernandez 1977b, 1992b; Preziosi 1979) are perhaps the more obvious diagrams of the poetic resonances between bodily experiences and living structures. In Newfoundland, for example, Pocius (1979) showed how the distribution of more and less hierarchically organized rug designs in formal front-room and familial kitchen spaces, respectively, reproduced the contrast in the social uses to which these two types of room were put. Another example is provided by rhyming couplets that reproduce sexual couplings (Dundes et al. 1972; Herzfeld 1985; see also Mintz 1997) in a manner sometimes explicitly recognized in the associated terminology. Here, the apparently superficial play on notions of sexuality offers an arena for playing on and contesting conventional expectations about other social relationships. Poetics is particularly attuned to the playfulness of social life—indeed, is about that playfulness: how people master the rules to demonstrate that the rules have not mastered them. The Play of Meaning At this point, we especially wish to focus attention on the notion of play. Vico himself exhibits a certain playfulness in his subversive use of etymology—his tracing of semiotic and tropological origins. Note that we say “semiotic and tropological” rather than merely “the origins of words.” This expansion of etymology enables us to expand the notion of poetics to a more technical sense than originally suggested by Vico and is more consistent with the thoughts of Jakobson, namely, to the play on the conventions of form. We thus evoke the theme of play and playfulness (as opposed to planfulness) (Fernandez 1994) as central to meaning creation and as diagnostic of it. Attention to planning in the more technical sense reveals the extent to which play may disrupt the

14_107-Bernard.indb 64

6/9/14 10:02 AM

2: In Search of Meaningful Methods

65

more repressive modalities of urban design (e.g., Mack 2012), much as the anthropology of science gives the lie to so much scientistic self-seriousness. This is, in turn, consistent with our view that meaningful knowledge of the constantly changing world is found not in classification itself and the planfulness that it represents (and that may be indeed a suppression of active knowing in favor of an imposed conformism), but in the act of reclassifying and of challenging or reconstituting existing classifications. In other words, play reveals form as it revels in it—sometimes in the moment of recasting or destroying it. In this sense, etymology is a kind of diachronic pun: It undercuts received and conventional meanings. Often used by the powerful to legitimate their power, it may also be used subversively or playfully to undercut that power. This was Vico’s understanding of etymology—a device through which the discovery of the instability of the term-referent relationship suggested the “civic disability” (Struever 1983) inherent in all human institutions rather than their capacity to resist forever the erosion of time. By noting the instability of meaning in all symbols—including, but by no means confined to, linguistic elements—Vico provided both a method for analyzing the role of meaning creation in the construction and maintenance of political authority and an illustration of the subversive activity by which cultural dynamics and change in that authority occur. A complaisant view of meaning privileging reference over use would have been of no practical use in understanding how social actors playfully manipulate the truth claims of seemingly fixed symbols to satisfy their own interest. Vico’s perspective, although conceived for very different ideological reasons and under conditions very different from those of modern anthropology, thus anticipated the present-day interest of anthropologists in action- or use-oriented theories of meaning and the constant social and cultural transformations for which they attempt to account.

The Elicitation of Nonverbal Meanings in Symbolic Interaction There is always the tendency for discussion of methods, particularly when etymological method is invoked, to become logocentric. But we must recognize that nonverbal signs are important in the flow of meaning—that they may, indeed, play a far larger role in the constitution of sense than it will ever be possible to recuperate from the data available to us, and their nonverbal character sometimes renders them particularly resistant to analysis precisely because academic modalities of analysis have traditionally been inherently verbal. Recuperating the nonverbal dimensions of ethnographic interactions thus represents a particularly important and difficult challenge, although these dimensions may be recoverable through a consideration of the ethnographer’s trajectory from communicative failure to a more successful apprehension of the role of gesture and its impact (Herzfeld 2009). Their relative absence from the literature would seem to be diagnostic of their insinuation into the realm of what Mary Douglas (1975) has called “self-evidence”—the common-sense, taken-for-granted aspects of a culture that are precisely what anthropology should defamiliarize and bring into focus. When linguistic models predominate, methods usually entail the elicitation of taxonomies and the systematic recording of violations of those taxonomies—“matter out of place,” in Douglas’s formulation (1966). When we turn away from language, the question of elicitation becomes vastly more difficult. Gesture and posture cannot be reduced to

14_107-Bernard.indb 65

6/9/14 10:02 AM

66

James W. Fernandez and Michael Herzfeld

verbal equivalents and must be indicated by notational systems of various kinds; these are often cumbersome or recondite and pose a practical challenge to comprehension and interdigitation with the verbal (see Farnell 1994; Farnell and Williams 1990). One approach entails videotaping sequences of dyadic interactions, then playing them with the sound turned off to other culturally knowledgeable actors and asking the latter to comment on the interaction, supply the words being used, and answer some specific questions about the nature of the interaction. This avoids facile equation of bodily movement with segmental speech units and permits greater access to ideological presuppositions entertained by the actors themselves. Even a cursory survey of Herzfeld’s (2004, 92–95) material from Crete, for example, shows that male commentators tended to sympathize with the artisans and female with the apprentices, thereby demonstrating the replication of ideologies of hierarchy in adjacent but discrete contexts of social life. Note, too, that the “diagramming” of social relations in the Jakobsonian sense allows us to recognize homologies of political relationship across two or more spheres (here, labor and gender). While such elicitation is to some extent intrusive, it fits increasingly with a worldwide familiarity with the technology that makes it feasible, thereby reducing the effects of shyness and embarrassment (on which, moreover, some social actors are willing to provide their own glosses). Here the cultural responses vary enormously, from lively participation to deep reluctance, the latter sometimes attributable to religious prohibitions. The method requires a set of dyads that can eliminate or at least minimize random effects. For example, given that the local term for apprentice in Crete means “foster son” as well, the inclusion of a father–son dyad in which the relationship was significantly warmer helped “materialize” the significance of kinship in affective terms. The repetition of the same scene with different voiceover and subtitle instructions to the viewer-respondents also accommodated a variety of sensibilities (e.g., visual as well as verbal attention to eye contact as well as gesture), and respondents were invited to put their own words into the actors’ mouths as a means of determining conventional expectations about such relationships. Some sessions induced elaborate reminiscences, which could then be compared with longer-term testimonies by former apprentices and artisans, and this helped establish ways in which informants could conceptually square their long-term recollections of ideal–typical relationships with the immediacy of experiences that were often very different. Especially through this comparison of the elicited responses with the verbal commentaries from a rich series of artisans’ oral work autobiographies, language itself made it possible to move the analysis beyond a purely referential and language-based model of meaning.5 Placing such analyses in the context of the longue durée can be a complex matter. The origins of nonverbal signs have usually proved more elusive than verbal etymologies, in the absence of notational systems analogous to writing. Their very evanescence makes them ideal vehicles for indirection, ambiguity, irony, and other nonreferential modes of meaning. But the adequacy of the lexicographic model of referentiality is brought even more into question by the flows of nonverbal meaning often tied up in symbols in which the absence of lexicographic methods challenges (or should challenge) the illusion of analytically reducible reference.

14_107-Bernard.indb 66

6/9/14 10:02 AM

2: In Search of Meaningful Methods

67

Difficult as it is to specify methods for the study of nonverbal meaning, it would be an impoverished anthropology that excluded from the domain of culture everything not reducible to verbal equivalents. Symbolic analysis has sought to escape the purely verbal. The difficulty is that it has tended to rest on contested claims of ethnographic authority (Clifford 1983; Sperber 1975, 1985) or on putting forth formalized systems that often enough, in circular fashion, justified their own logic, a criticism made of Lévi-Straussian structuralism (Leach 1970, 1976; see also Schwimmer 1978). RECOVERING MEANING IN USE AND EMERGENCE

Openness to nonverbal meanings brings with it a host of problems. Since all contemporary human communities whose symbolic universes we attempt to enter are composed of language-using subjects, however, we emphasize that the Vichian notion of poetic knowledge/wisdom suggests the outline of an appropriate methodology. Vico’s own use of etymology was notoriously unreliable from a purely philological standpoint, but if we recast his phylogeny of human semiosis as an ontogeny of semiotic forms—if, in other words, we suppose that all meaningful acts have a history and that the process of their formation can be reconstructed—we have propositional paths that enable us to appeal to at least two central foci: use and change over time. The first of these two operations concerns the recovery of simple image-schemata as these repose in primary experience or process. This is the “argument of images” (Fernandez 1986; see Stark 1996). Although anchored in the concept of the image (which has usually been taken to be visual but need not necessarily be so—we shall briefly explore other iconicities below), it works primarily with verbalized tropes invoked as the basis of claims to local knowledge. But if this kind of interpretation is to be applied to tropes and symbols not expressed in language or supported by verbal exegesis, the basis of interpretation must become one of observed use. This assumption of ethnographic presence and direct observation of actual use undergirds inspired studies of tropes and symbols, of which an early and extended example comprises Turner’s studies of Ndembu symbolism (Turner 1967). Unless this observational anchorage is made explicit, as it was not in Turner’s case, the regrettable appearance of arbitrariness or invention—a projection of referentiality onto an analysis that is not intended to convey it—becomes the basis of criticism (Sperber 1985). The second operation is analogous to the pragmatic turn to observations of actual change in use in ongoing action. Stimulated by the etymological impulse, it is above all historical. It is the restoration of the centrality of time in all interpretation by following the paths that particular usages follow over long periods of times. Such study gives substance to the idea that cultural form is always in a state of emergence (Bauman 1984 [1977]; Bruner 1994). Our reconstructions of meaning thus starts from the recognition that before we can say what the meaning of a given action or object situation might be, we must first establish the range of acceptable alteration in the form of the act both at present and over time. In that context, we may even conclude that meaning is itself an inappropriate term and that “meaningfulness” is the more significant and appropriate form because it implies an ongoing temporal and emergent process of successive usages that themselves have a flexible range of acceptability.

14_107-Bernard.indb 67

6/9/14 10:02 AM

68

James W. Fernandez and Michael Herzfeld

This move, which protects us from the tendency to essentialize our own categories by invoking as explicitly as possible the primacy of use and action, both complicates and strengthens description. It does so by keeping reification at bay, for it reminds us that an empirical task, both etymologically and pragmatically, is the grounded inspection of lived experience (Greek empeiria) rather than an exercise in abstract classification far removed from the social contexts in which we gather our data. INDETERMINACY: TAPPING THE HISTORICAL CONSCIOUSNESS OF PERMISSIBLE VARIATION

Tracking variation over time in the uses of signs makes us aware not only of emergence of meaning but of a considerable degree of variation (Fernandez 1965) and even indeterminacy (Giddens 1984) in the meaning of common practices and their accompanying and associated symbolic components. In anthropological literature, there has long been awareness of ranges of permissible variation, from culture to culture and time to time, in dress or performance styles, for example (Kroeber and Richardson 1952 [1940]), or in the graphing of the differential distribution of normative behaviors generally (LeVine 1973). But the study of the meanings attached to or promoted by these differences is usually lacking. Empirical observation shows, moreover, that there is a temporal dimension here in relation to which meaningfulness in action can be derived. As a modest example, a too-sudden change in the way of doing things with respect to table manners may provoke anger or contempt and thus very significant social meaning. So it is vital to link the tempo of performance (Bourdieu 1977, 6–7) to the rate of change in the prior form of the act; some acts are much more resistant to change than others, and variations have greater potentiality for meaning production both for the participants and for the observer. Analysis may thus, at a crude level, entail listing conventions and mapping them onto a social grid or a temporal chart to establish empirically the range, context, and character of permissible invention and intervention. This is clearly a complicated matter—but only the most determined Cartesian would claim that it was less empirical than the summing-up of such complex matters in the form of a referential taxonomy devoid of the ambiguity that lies at the heart of all experienced social life. Moreover, it can be checked against informant responses, both openly elicited and supposedly natural (a revealing term in itself; see, e.g., Boehm [1980], in which the methodological and cultural artifice entailed in such invocations of nature is clearly exposed). When a single action appears to constitute a solecism in certain actors but not in others, the difference indexes the other factors that contribute to the shifting, but often identifiable, flux of interpretation and meaning. We may thereby, for example, infer what factors influence informants’ judgments of each other and how predictable their effects are thought to be. Some of these ranges are codified for us. Etiquette books register, for example, both the rigidity or permissible range of behaviors and reflect the acceptable rate of longterm change. But only long-term participant observation can effectively reveal the significance of different tempos. Sudden variation or acceleration of behavior excites the search, in gossip for example, for meaningful explanations such as “foreign influence” or “degeneration of morality.” Actors teeter between social boorishness or approbation for social daring; their attribution to either categorization depends not only on what

14_107-Bernard.indb 68

6/9/14 10:02 AM

2: In Search of Meaningful Methods

69

they do and what relationship they have with those who are judging, but also on evershifting perceptions of what constitutes a permissible distortion of convention—an appraisal often influenced by the actors’ own mastery of play. There is some evidence that abrupt variation attests to superficial rigidity of norms and that a challenge to such norms can constitute status-gaining acts of daring, particularly in societies in which an ideology of opposition to authority is explicit. Cretan artisans, for example, deny that they encourage insubordination in their apprentices. The evidence suggests, however, that they usually retain for long periods of time only those apprentices who have shown sufficient daring and resourcefulness to undermine the artisans’ formal authority—a paradox that exemplifies the role of ethical and political ambiguity in many societies where agonistic relations are prominent. In a truly Vichian mode, then, observation of the shifting relationship between “etymological cognates” among modes of social comportment becomes an investigation of the kinds of semiotic stability necessary, in a given culture or society, to the effective maintenance, negotiation, or usurpation of power. The converse also holds true. “Time servers,” “hacks,” “wimps,” “bureaucrats”—these are labels reserved in many societies for those who, precisely because they follow social norms too rigidly, marginalize themselves from the fellowship of ordinary social interchange and thus from its manifold possibilities for inventive meaning creation. Etiquette books have their modern equivalent in the transnational world in the form of handbooks designed to help entrepreneurs negotiate their way through foreign cultures, but only by following their evolution through several editions (see, e.g., Mole 1990, 1992, 1995) can we trace both the authors’ learning path and the deeply hidden flexibility of each culture so surveyed. The proliferation of such books and the telegraphic instructions travelers find in guidebooks often occasion mirth from local people precisely because such formalization usually leads to excessively literal observance of rules. True mastery can only be shown by bending those rules within the shifting limits understood, supposedly, only by those who inhabit the culture’s intimate spaces. A historical awareness of what has transpired may provide actors with quite flexible templates for present action, as the range of ways that Turner’s (1972, 1974) “social dramas” can be played out indicates. The use of historical consciousness in understanding the range of possibly meaningful action and the scope of the indeterminate is always problematic. In societies with no recorded history of symbolic forms, such analyses are hardly feasible, as neither informant nor analyst will usually have the knowledge of the relevant antecedent events in a form that can easily be matched to the sense of documentary facts available—albeit from differing perspectives—to both. It is all too easy to reduce that knowledge to dismissive remarks about formulaic representation, as was customary in folklorists’ accounts of songs and tales about historical events, but this overlooks the fact that modern journalism is equally subject to formulaic design, as in the clichés from which most headlines are built. Academic discourse, similarly, often has a reductive effect on the information its users seek to convey. Of course, the reenactment of mythological events is a kind of collective remembering. But even here, the problems of translation and the establishing of “coevalness” (Fabian 1983) are so great that they may defeat the best of intentions, and they certainly leave enormous areas of ambiguity unresolved.

14_107-Bernard.indb 69

6/9/14 10:02 AM

70

James W. Fernandez and Michael Herzfeld

Most famously perhaps, or at least most recently, the Sahlins–Obeyesekere debate about the Hawaiian understanding of Captain Cook’s murder centers on just such a question of recovering and assessing interpretations from indigenous myth and ritual practice. The debate suggests the magnitude of the problem in recovering just what occurred in the events surrounding Cook’s death. For the Hawaiians, there were ritual precedents for such a death, but the intractable problem of deducing what these precedents meant and what behaviors they might provoke or determine remains, and interpretations made of them are inevitably subject to present interest (see also Borofsky 1997; Parmentier 1996; Roberts 1997). Modern nationalism is another arena in which, with extraordinary rapidity, “the facts” become inextricably entangled with constructions of the past necessarily derived from present concerns and leading to a serious questioning of the very distinction between the supposedly analytic categories of history and myth (see Abu El-Haj 2001; Hill 1988, 1996; Rappaport 1994; Steedly 1993). Recording the ways in which even such “fixed” relations as those of kinship can be renegotiated in the course of political realignment (e.g., Shryock 1997) also sheds valuable light on the ways in which the meaning of the past may be calibrated, for social actors, to the political exigencies of the present; in such cases, context-free assertions of the truth or falsehood of historical assertions are not so much untrue in themselves as simply unhelpful and uninformative. What is involved is an etymology of forms that are not necessarily or invariably verbal. The attempt here is to discover essentialized meaning in the past. Standard etymology was indeed precisely a form of essentialism in action, an effort to establish a past for words, the denotata of which, through the common error of misplaced concreteness, were represented as equivalent to essentialized objects—institution, nations, values. Its proponents sought the truth (to etumon)—the goal that made Vico’s intentionally subversive use of etymology so dangerous, since any threat to the order of words logically entailed an equivalent threat to the order of things. Nonverbal etymologies are a special case of the more general rhetorical phenomenon of self-backgrounding or self-evidence that typifies all signification based on resemblance, especially one claimed to be timeless—in short, iconicities, the seeming naturalness of which provides a refuge from analytic scrutiny and makes them ideal tools of political propaganda and especially to claims of racial, cultural, or linguistic continuity with a supposedly fixed point in the past (Herzfeld 2005, 56–73). Such iconicities are truly constitutive of cultural capital and the claims of power. In the case of linguistic etymologies, it may sometimes suffice to record who asserts them and under what peculiar circumstances, as well as to note the terms in which challenges—including claims of unscholarliness!—are mounted in turn. The idea of an etymology beyond language may strike some as a contradiction in terms, but this response in itself offers a peculiarly ironic illustration of the problem. Only a prior assumption that facticity can only be established through language and that language is capable of pure referentiality—that the relations among words are “like,” “icons of,” the things they supposedly denote—would logically make the notion of etymology exclusively verbal. We argue, to the contrary, that—as with poetics— challenging that long-cherished assumption of a self-proclaimed common sense must be part of a critical anthropology’s methodological commitment to meaning.

14_107-Bernard.indb 70

6/9/14 10:02 AM

2: In Search of Meaningful Methods

71

Notions of nonverbal etymology have been around for a long time. De Jorio (1832) provides a minor instance of the attempt to connect the visual rhetoric of modern Neapolitan gesture with the forms of gesticulation represented in ancient statuary (see especially Kendon 1995). Art historians recognize the rhetorical intentions in the making of such links easily. Gombrich’s (1979) well-known studies of visual “rhetoric” and “visual metaphor” (his terms), in pointing up implications of class and status, are perhaps a more useful illustration of what underlies our insistence that a social poetics is inevitably political, a point that was realized in an important way by the anthropologist E. W. Ardener (1971, xliv), when he pointed to the “blank banners” of resistance. These represent the outer edges of iconicity. Only the faintest traces connect them to each other and to the explicit configurations that hesitantly begin to emerge in what Scott (1990) has called the “hidden transcripts” of peasant resistance; they may be engaged in configuring the sociopolitical landscape long before they are translated into verbally explicit forms of protest and revolt. When we move to the etymologies of art, architecture, and space, and still more of gesture (to name just a few examples), the traces are more clearly marked, and the aura of antiquity and deep knowledge surrounding such etymological linkages may be also clear enough.6 Synchronically, such nonverbal etymologies may be accessible by a kind of triangulation of linkages in one area of culture: architectonics. Among the Fang of Gabon, for example, we can identify parallel structures in the architectonics of dwellings, village layout, games, and dance forms (Fernandez 1982, chap. 4). This is a replication of structures (Vogt 1965) that suggest the investment of persistent common, thematic, meanings in the deployment of bodies in multiple ways through space. Herzfeld (1991) has argued that architecture represents the clothing of the body politic and suggests in that architectonic metaphor the linkages and tensions between external display in its various forms and interior familiarities. These tensions are likely to be especially acute in societies where the physical body is itself notably an object of strong, socially organized sentiments of embarrassment (as in southern Europe). This insight may also help archaeologists overcome the absence of explicit verbal information. Concentricities between the deployment of specific decorative motifs in domestic house plans paralleling their distribution in the territories of dominant and subaltern populations, for example, may—if ethnographic parallels are any basis for judgment—be highly suggestive for deciphering the management of cultural capital and thus for the relationship between domestic and civic power (Herzfeld 1992b, 80–82); see also Gardin [1980], Gero [1991], Leone [1988], and Watson [1994], in which meaning is reviewed in many contexts, including those of class, subsistence conditions, gender, and technology). The tension between what is observed and the specificities of informant exegesis is a constant in fieldwork, particularly in such complex matters as the interpretation of meaning. But we also insist that the ideal—as, at the very least, a model for (Geertz 1973) action, and thus as consequential—is no less real than what is conventionally assigned to the latter category. Again, the turn to practice-oriented theories has fatally undercut the old division of labor between the symbolic and the material; more recently, it has also challenged the old reluctance to engage political and ethical issues in a committed fashion and at the same time has expanded (or intensified?) the

14_107-Bernard.indb 71

6/9/14 10:02 AM

72

James W. Fernandez and Michael Herzfeld

range and locus of fieldwork practices to include areas of familiar routines in the ethnographers’ home cultures (Pink 2012). Yet there is a great deal that we can do, practically speaking, to rescue provisional but illuminating understandings from this apparent chaos. Verbal exegesis is unquestionably beset by ambiguity. But it is also anchored in indexical relationships to significant parallelisms, so that variations among different exegetical texts may provide a means of access to the ways in which seemingly rigid codes provide protection and justification for a wide range of cultural invention. Perhaps the most dramatic case of how ethnography can reconfigure claims of literalness as meaning creation lies in analysis of bureaucracy, but there are many other examples that, while less extreme, show how all claims to fixed meaning are actually themselves, ipso facto, contingent interventions by social actors. SYSTEMATIC ASPECTS AND LIMITS OF MEANING ACQUISITION

In informant exegesis, it is difficult to escape the tense relationship between politics and meaning, a relationship sketched out in Abner Cohen’s pioneering Two-Dimensional Man (1974). But this tension, the messy suspension between a desire for sincere and accurate description, on the one hand, and political commitment to the established privileges of always-already body-based, ingroup-oriented, and ingroup-adapted biases, on the other, is a condition of social life, and its presence in field observation as a supposed inadequacy is, in fact, the source of its claim to produce knowledge. This struggle of social subjectivism, as we might call it—the struggle for objectivism and the out-and-out self-assertion and consequent rejection of objectivism—is a condition of both informant exegesis and testimony and eventual anthropological interpretation. This tension may appear in testimony to or exegesis in everyday life. For the ethnographer, it represents a difficulty of escaping social subjectivity; to be in some ultimately unattainable sense objective can be, as Jenkins argues (1994), a source of knowledge. For it is in this tension and this struggle for objectivity in the pursuit of meaning that anthropological knowledge most closely models the politics of its own study. There is a reflexive interrelationship in the two systems of meaning. Fieldwork, Jenkins argues, is a form of apprenticeship in which the body, or the “body in the mind,” is trained by a combination of frustration and example to expose the inadequacy and thereby enrich a purely intellectualist approach. But it is the comparison of anthropological method with observed local practices that, for Jenkins, rescues this phenomenological insight from a crass subjectivism—or from a thinly disguised objectivism (see Reed-Danahay’s [1995] critique of Bourdieu [1977, 1984], for example). Apprenticeship, not childhood socialization, is the operative model here, although anthropological research has long been characterized as a process of socialization to another culture. This distinction is methodologically useful and important, since the forms of apprenticeship in a given society can be studied to contextualize the ethnographer’s own learning processes (Coy 1989; Delbos and Jorion 1984; Kondo 1990; Lave 2011). The model of socialization, by contrast, implies both a passivity and an absence of intentionality that do not correspond to the actual experience and maturity of intention of ethnographers, however infantilized they may be made to feel; it is thus epistemologically unsound (Karp and Kendall 1982).

14_107-Bernard.indb 72

6/9/14 10:02 AM

2: In Search of Meaningful Methods

73

The metaphor of apprenticeship and the accession to artisanal excellence, with its attendant implications of situated learning (Lave and Wenger 1991), can be compared with local methods of inculcation and experienced as such. It also has the advantage of forcing anthropologists—who are rather wordy creatures on the whole—to be attentive to, if not to adopt, a set of practices in which verbality is actively discounted or even disapproved in favor of handiness, artisanship, real-world know-how (savoir faire), rather than academic know-say or say-how (see Bourdieu 1977; Dougherty 1985; Jackson 1989; Pirsig 1974; Willis 1977), or is treated as artisanal in itself (Boyer 2005b; Herzfeld 2007; Lave 2011). Anthropology is often conducted in circumstances in which to talk a good game is not really to play it at all. This, in turn, brings the ethnographer’s frustration with the often inarticulate imponderables of the ethnographic experience more directly into line with those of repressed craft apprentices who must not only suppress their questions and backchat, but must also, because of that situation, find practical ways of overcoming the dearth of verbal communication to learn anything at all—in short, they must learn by seeing and doing and not by saying. What they seem to learn, in many cases, is the kind of cunning—or, better, craft—needed to abstract practical knowledge from observation and hands-on participation very much like that described by Jenkins (1994) for the day-to-day business of fieldwork. Claims to cultural knowledge through purely verbal operations may be a black-box-opening operation of the most problematic kind. The more secure method of approximation is to have learned by apprenticeship how to do (the essence of poiesis, the proper subject-matter of poetics), to operationalize by bodily activity, what others know how to do in the world (Fernandez 1992a). In several different field situations, we have found it useful not only for rapport but simply for ethnographic learning to try and apprentice our bodily selves, we might say, to local masters. Among Fang villagers, Fernandez (1965) acquired an agricultural plot and sought not only advice on agricultural practice from elders but also participation and direction of his physical efforts in his apprentice practice. When working on the religious movement of Bwiti, Fernandez (1982) sought help in learning the bodily discipline necessary for participation in some of the easier dances in the complex dance cycle and, as a consequence, danced at times in the all-night cycle. In his work among Asturian villagers (Spain), Fernandez (1977a) also sought physical participation in agricultural activity, apprenticing himself as a mower learning to manage the scythe, and he sought to relive the miner’s experience in its most corporeal terms by entering into the mine with the various officios and following them to their various work sites (see also Jackson 1983). Herzfeld was less successful at persuading a sheep-thief to take him along on a raid. But the way in which Herzfeld’s verbalized enthusiasm for the project so disgusted the shepherd—who declared that while he had made a nóïma (a nonverbal gesture) Herzfeld had inappropriately responded with speech, thereby demonstrating his ineptness—was a valuable object lesson, perhaps more useful even than the participation in communal house-building work parties through which Herzfeld nevertheless came to acquire a more physically engaged sense of community. Coy (1989) and others have documented the advantages and limits of actual apprenticeship as a source of descriptive insight (see above).

14_107-Bernard.indb 73

6/9/14 10:02 AM

74

James W. Fernandez and Michael Herzfeld

But it is important not to romanticize one’s own experiences, common though the ground and moment may be, as a realization of those of the local informants: There is a real risk of trivializing the latter. Examination of the discourse about that experience (e.g., popular and critical responses to fictional accounts) may at least, as we have noted, furnish information about the culturally acceptable representation of such experiences. But this should not be confused with direct knowledge of the representation itself, though, unless we are also prepared—as we must be—to assume a degree of commonality in all human experience under conditions of strong physical engagement. We come to experience that engagement, not necessarily as a conscious act of self-apprenticeship aimed at gathering data (which is not, after all, our informants’ usual primary concern), but through relatively disinterested forms of emulation (see Jackson 1989, 134–35). THE CORPOREAL LOCUS OF MEANINGFUL ACTIVITY

We wish here to pinpoint what is, in effect, a method of attentiveness to bodily meaning or embodied meaning (or “body-in-the mind meaning”). This is surely appropriate to social and cultural anthropology, for much meaningfulness in social life is so embodied and implicit and is not cerebralized according to academic preferences. Indeed, one of the most cerebral descriptions of the fieldwork experience, Evans-Pritchard’s (1940, 13) celebrated account of “Nuer-osis,” clearly betrays the frustration of trying to formulate verbally unarticulated knowledge, as, more generically, does Lévi-Strauss’s (1963) equally intellectually problematic comment on conscious models. The absence of verbalization in the learning process is not a complete barrier to understanding, however, for by working alongside our informants, we can partially—and sometimes only inchoately—perceive the grounds of their social experience across differences of culture, political hierarchy, and gender, as Jackson (1989) has noted in his own attempts to grapple with the life-worlds of African peasant women (although his account presupposes a somewhat problematic concept of empathy). Such an approach works through what Fernandez (1974) identified as the basis of the social role of metaphor: It predicates a sign of known selfhood on a necessarily inchoate other. Even in the hard-headed world of business, metaphor plays a crucial role. Whether we are speaking of an analyst’s recognition of the presence of embodied metaphors of indeterminacy and specifically of gambling in the everyday conduct of business in a Greek island town (Malaby 2003) or of a Korean CEO’s invocation of the metaphor—or, more properly, the simile—of the temporal parallels between banking profits and the maturation of fine wines (reported in The Korea Times, May 24, 2012, p.12), we can see that metaphors, like symbols in politics (Bailey 1973, 1983, 1988; Kertzer 1988; Paine 1981), can provide an effective way of framing desirable outcomes and social implications at the highest level of corporate power. Such uses of tropes are direct and accessible. As Rapport (1993) remarks, however, even within one’s own culture (itself hardly an unproblematic concept), there are severe limitations to our capacity for interpretation. The possibilities through participation of reading the mostly unarticulated and also diverse and only partially intermeshing worldviews of English villagers (Rapport 1993, Part I) are necessarily limited by social conventions of silence and introversion, even though they may also,

14_107-Bernard.indb 74

6/9/14 10:02 AM

2: In Search of Meaningful Methods

75

in their very closedness, be revelatory of what those worldviews mean for social actors as they try to decipher each other’s intentions and inner lives. As Candea (2007) has noted, we too easily ignore the silences, chasms, and tacit exclusions that are actually a significant part of ethnographic experience and of the lives of those we study. We would do incalculable violence to the reality of social life if we insisted on perfect representations of social experience as the criterion of admissibility (see also Leavitt 1996); mapping the imperfect and the imponderable is no less realistic than any account of social structure or cultural norms. For actors who may be thought to share a range of cultural experiences, responses to literary form offer especially useful insight when critics imply or even articulate the criteria by which they and their readers collectively assess the plausibility of fictional accounts of psychological inner states (Cohen 1994; Herzfeld 1997a). It is in this reaching back to or below language that we can most critically understand Vico’s “poetic wisdom”—not as an origin tale, complete with a culture-hero, to explain (in the manner of an “etiological myth”) the mundane practices of our workaday world, but as an empirical recovery of the ways in which the body registers all knowledge before it is processed as increasingly abstract, disembodied representation. All human knowledge of the world beyond immediate sensory experience is mediated by signifying processes located as Vico noted, and as both phenomenologists (Csordas 1994; Merleau-Ponty 1962) and the cognitive linguists (Johnson 1987; Lakoff 1987; Lakoff and Johnson 1980) have attempted to demonstrate in more formal ways, in the body’s preobjective placement from which the self negotiates its life-world. We are taking a militantly middle-ground position here (see Herzfeld 2005), emphasizing a mode of fieldwork that focuses on the mediations of corporeal experience and that locates what has been called “the mind” where Vico located it—in the body. It makes no sense to discard such fieldwork as unempirical merely because we presently lack a full conceptual apparatus for describing the imponderable. That would be a poor excuse for, and a poor contribution to, meaningful methodology. We can provide some paths to follow and techniques to use in the project of recovering the meaningfulness of expressive forms beyond the literal interpretation of the lexical. PATHS TO AND TECHNIQUES OF MEANINGFUL INQUIRY INTO THE AGONIES OF EVERYDAY LIFE

A central form by which life is made meaningful is narrative emplotment—the characterization of human interaction and of the persons involved therein in dramatic terms. Characters are ensnared in various mixtures of tragic or comic expectations vis-à-vis themselves and others and they are consequently involved in interactions that lead, by their agonistic interaction, to various states of more or less enduring resolution, requitement, or dénouement of their affairs. In the process of the dénouement, they obtain confirmation, denial, or significant transformation. This theatrical trope—basically, the idea of social role playing—has long been persuasive as a plausible formulation for grasping the essentials of human interaction. In starkest form, social interaction is seen as the struggle for desired or feared characterization as heroes, villains, and fools (Klapp 1972). Less dramatically, Turner (1974) has taken this dramatic trope and made it into a phased method of social analysis, linking the

14_107-Bernard.indb 75

6/9/14 10:02 AM

76

James W. Fernandez and Michael Herzfeld

short-term performance of ritual time in Van Gennep’s celebrated tripartite schema to the longue durée of historical event structures. But this dramatic or narrative trope of social science understanding, or at least its teleological implications of ultimate dénouement or resolution, can be found in much less explicit form in models of human or social or cultural development. In one form or another, these all imply what can be called progressive transformations of character, dénouements of a kind. It has been noted for Western literary culture that male autobiographies tend to be more deterministically teleological than female ones (Jelinek 1980); given the new interest in auto-ethnography—the comparison of their with our constructions of their cultures (Reed-Danahay 1997)—this kind of generalization should be distinctly susceptible to cross-cultural comparison and to sensitivity to changes across time and between contrasted contexts. The master trope of teleology has certainly been a major element in anthropology’s own self-construction, which is perhaps one thing that makes it hard to recognize as distinctive or worthy of comment when we encounter it in the field. This was most obvious in classical evolutionism in its dramatic developments from savagery to civilization. Even Vico narrated a developmental model of progressive transformation over historical time of consciousness in successive social formations. He pointed up not only the key tropes of a given sociocultural stage of development but the tropological phases—metaphoric, metonymic, synechdochic, and ironic—through which these successive stages might be expected to pass. Vico’s awareness of the tropological in the study of meaning rests on notions of the tropes as part and parcel of a society’s narrative sense of its place in space and time. And, unlike the unilineal narratives of development in much applied anthropology and more largely in the colonial project and its successor ideologies (see Ferguson 1990), his was not a morality tale of ultimate triumph but a warning about the self-destructive proclivities both of absolute power and of the pretension of literal truth. For anthropologists, Vico’s warning should resonate with the dangers of “cultural fundamentalism” (Stolcke 1995), and there may be something salutary in comparing this with other, more obviously political and religious fundamentalisms (as Stolcke’s formulation clearly suggests). We may not necessarily buy into the developmental narratives themselves, either of particular analysts such as Vico or of particular societies (see Fernandez 2001; Hodgen 1964; Nisbet 1965), but they raise questions of reconstructing or deconstructing the taxonomies that constitute the local idiom of common sense with respect to both societal division and societal progression. It makes clear our obligation in the study of meaning to be alert to and describe such taxonomies in their context of use (Douglas 1966, 1975, 1992). These tropes and taxonomies are plainly central to the meaningfulness of social life. However, they do not exhaust either the semantic universe or the range of concepts and emotions to which social actors wish to give expression. Even the linguistic modeling of emotives with no verbal equivalents, “gustemes,” or the formal analysis of gesture and the like does not dispose of what is left, for clearly there are many areas of expression not easily captured by linguistic models. Take the matter of spatial communication and the meaning of space. Spatial idioms in language are plentifully available for class, status, and power analysis. The feudally derived distinction between upper and lower town in some European societies and

14_107-Bernard.indb 76

6/9/14 10:02 AM

2: In Search of Meaningful Methods

77

the roughly contemporaneous railroad extension and western town settlement in the United States produce the distinction between this side of the tracks and the other—or even a “space on the side of the road” (Stewart 1996) that is altogether marginal. Spatial figuration rather easily maps on social formation and can be analyzed taxonomically. This teleological mapping of spatial arrangements on society and vice versa is Durkheimian in the sense that they are reconstructions of the working out of the ritual space–social space teleology found in The Elementary Forms of the Religious Life (Durkheim 1976) and found in other oppositional taxonomies as well. Bourdieu (1973), Douglas (1973), and Needham (1962), for example, have all shown how the physical layout of domestic and public space may express the key complementary oppositions (notably but not exclusively those of gender) obtaining in a given society. Bourdieu (1977) has expressed dissatisfaction with such a reductionist formalism and has posited a much more complex temporal–spatial posture or readiness for interaction in the world, the habitus (although this, a concept ultimately derived from Mauss [1936], appears to be no less deterministic in important respects). More recent work on landscape (Rapport 1993, Part I; Schama 1995) and the organization of public space explores both these explicit oppositions lodged in the lexicon as well as more deeply embedded, less explicit meanings lodged in personal and social histories and in the complexities of corporeal experience in the world are explored (Bubandt 1998; Desjarlais 1992; Kleinman and Kleinman 1994; Lawrence and Low 1995; Low and Lawrence-Zúñiga 2003; Seremetakis 1993). This work shows us methods for the study of the meaning of space much more responsive to a poetics of space— and more grounded (literally)—than that intended by Gaston Bachelard (1964). Uses of space tie architectonics both to bodily movement and to political interests, and thus also provide an especially useful perspective on the dynamic interplay between state ideologies and the politics of everyday life, as is apparent in the narratives about historic sites already mentioned. Ordinary people only attribute meaning to ideology when it impinges on their everyday experiences, and the calibration of ideal–typical models of space (e.g., sacred enclosures and the haram of the home, tensions between external and internal domestic ornament as expressions of conflicting ideologies of national culture) provides an accessible arena for examining contestations of meaning between the state (or religious community) and the domestic sphere. These may be recovered through narrative: Bahloul (1996), comparing the memories of Jewish refugees from Algeria with those of the remaining Muslim inhabitants of a once-shared domestic compound, has effectively shown how to elicit the shifting meanings of space in a highly politicized context. It is the comparative perspective, grounded in a common physical space, that permits a clearer understanding of how meaning organizes the perception of space. An opening up to the methods of social poetics as propounded here is seen in the relatively recent willingness of anthropologists to use modern urban-industrial societies, not as pale or distorted sites for the fragmentation of Maussian total social facts (supposedly found in their pristine oppositional perfection in “exotic” societies), but as contexts in which we can examine our own subjective embodied knowledge about supposedly necessary equivalences between spatial forms and linguistic meanings. Here we can explore embodiments in all their complexity. In such settings, we can

14_107-Bernard.indb 77

6/9/14 10:02 AM

78

James W. Fernandez and Michael Herzfeld

more explicitly investigate the lack of fit between prescriptive ideologies of meaning and actual practice or use, thereby also obviating earlier and ethnocentric distinctions between “complex” and “simple” societies that may have been the result of our own classificatory desire for closure rather than of the realities experienced by the members of supposedly simple societies. The argument of Mary Douglas’s Purity and Danger (1966) is as applicable to the self-protectiveness of national or international bureaucracy (Barnett 1997; Herzfeld 1992a; Malkki 1995) or to the “untouchability” factor in urban ghettoization dynamics as it is to local ritual. Conversely, agency and practice complicate local and nonliterate social relations as much as they do the life of highly industrialized, geographically extensive social formations. One very promising area for further research lies in the relationship between the bureaucratic nation-state and local populations. If the state (as Weber’s ultimate embodiment of modernity) seems to generate definition and closure, ordinary people challenge this through their actions and understandings. Indeed, successful nationstates tolerate a significant degree of departure from their official norms, recognizing that failure to do so may result—as Vico would have predicted—in catastrophic consequences for the wielders of state power (Herzfeld 2005; see Boyer [2000] for a particular example of such an effect, in the former German Democratic Republic; see also Festa [2007] on Taiwan). The “complaisance,” as Jung (2010) describes it, that leads bureaucrats to play along with citizens’ and their own minor infractions is a characteristic of official action in most states, and we know, as Scott (1998, 256) reminds us with the example of the British “work-to-rule,” how disastrous an excessively literal-minded adherence to rules can be and what effective resistance to authority it can generate. Building critically on the model of cultural intimacy (Herzfeld 2005), several writers have illuminated these processes in ways that are important for seeing how even so apparently literal an institution as the bureaucratic nation-state must live with the consequences of the metaphors whereby it has constituted its own claims to logic, coherence, and legitimacy (Johnson 2010; Jung 2010; Malaby 2010; Neofotistos 2010; Soysal 2010—all contributions to a special section of the journal Anthropological Quarterly devoted to a critical assessment of cultural intimacy). POSTURES OF POETIC WISDOM IN ETHNOGRAPHIC INQUIRY AND INTERPRETATION

Revitalizing Moments The suggestiveness of the trope of path orientation in meaningful inquiry should not determine method altogether. As Vico showed, claims to open-endedness contain the seeds of their own disproof; even ambiguity can be reified as dogma. Social life oscillates, on the one hand, between established and more-or-less inflexible routines that, once learned, tend to lose their ability to convey meaning with any real immediacy and, on the other hand, moments of much more intense—because more palpably exceptional—meaning. Their intensity arises from increased uncertainty and unresolved complexity—from an increase in ambiguity, not of referential stability. Indeed, informants’ inability to provide a clear gloss on some symbolic form may index a moment of important—if temporarily unfathomable—conceptual change.7

14_107-Bernard.indb 78

6/9/14 10:02 AM

2: In Search of Meaningful Methods

79

In addition, such concerns also highlight the inadequacy of Durkheim’s (1976) reified dichotomy of sacred and profane. This is not to downplay the importance in anthropology of ethnography enriched by long-term observation and attention to the intimate details of everyday routines (thick description [Geertz 1973]), or the identification of particular events as diagnostic of systemic changes (Moore 1987). But we wish to emphasize the importance of particularly heightened moments of meaningfulness, seeking to understand the special vitality that they embody. Such events may serve as moments of social revitalization or devitalization (Wallace 1956). This is an organic trope (Nisbet 1965)—of vitality in social life—that focuses our attention, again, on the bodily experiences entailed in that life. Here the methodological challenge is to recognize through its external signs or narrative realization the embodiment by the social subject of social experience—the translation of social suffering, for example, into bodily disease (Das et al. 1997; Kleinman and Kleinman 1994) or the translation of social celebrations into a heightened sense of bodily well-being—and to relate such correspondences to larger patterns of social change or stability. Meaningfulness derives from the special reinforcement or abatement of the intensity or transformation of social relations. There is a significant literature on transformation, an important segment of which addresses the “structure of the conjuncture” or transformative event in history (Braudel 1980; Sahlins 1981). This approach suggests a method of tracking the changes in accustomed (normative) structures of behavior as a consequence of such events. Our emphasis is less on the structures involved than on the corporeal resonances arising from the reinforcement or weakening of old associations or the emergence of new ones. This interest in surprising or arrested moments or events in culture in respect of embodied experience contrasts with other approaches to meaning that engage primarily with routine, and that seem to emphasize path dependency without regard to the associational structures or contingent events involved (on contingency and gambling as a metaphor for social and economic life, see Malaby 2003). The study of meaning is bound to have to consider the nature of surprisefulness, not only with respect to the conjunction of the prevailing and changing systems of structure, but also with respect to the associated corporeal commonplaces of social interaction and the confirmation or challenges to these presuppositions. These surprise-laden incidents arguably account for differences in the quality of life from moment to moment and phase to phase, differences that constitute an important part of life’s changing meaningfulness. QUALITATIVE INQUIRY AND THE QUALITY SPACE OF LIFE

The trope of quality space for the temporal complexities of corporeal–mental life has long been recognized as a basic time–space conversion and is used in social sciences to plot the vectors of judgment by which the body in the mind weighs and assesses its experiences and, consequently, decides to act in the world (Osgood et al. 1957). And the notion of quality space has long been given by philosophers a special place as a fundamental tool of understanding of how the mind places its experiences in corporeal terms (Quine 1981). This trope, indeed, has assumed the status of a kind of elementary form of reference (Fernandez 1974) in seeking to understand in a methodological way the motivations of metaphoric predication itself. It can be understood as a trope anchored

14_107-Bernard.indb 79

6/9/14 10:02 AM

80

James W. Fernandez and Michael Herzfeld

in the primordial gestalts of human experience as this takes place in our gravitationally anchored three-dimensioned world (see Lakoff and Johnson 1980). It is as a means by which to grasp those displacements from unacceptable experiences to more satisfactory experiences that metaphoric predications are often used (Fernandez 1974, 1986). A social poetics must consider method itself under this rubric, as a form of understanding that uses a choice of tropes to organize its clarifying activity in the world. Of equal interest are other tropes of utility in the rhetoric of methodological argument, tropes of thickness, as in “thick description” (Geertz 1973), or of depth, as in deep or penetrating analysis and of movement (as in “social movements”). All these “figurations of social thought,” like the tropes of game, drama, and text (Geertz 1983), are fundamental in the arguments of methodology and methodologists, and it is their presence and evocative associations that a poetic approach takes into particular account. Even culinary tropes organize assessments of data: “thin” or “meaty” ethnography, “raw” and “cooked” data (the latter being somewhat suspect), and “processed” analysis. It would be hard to claim that such assessments lay outside the criteria of cultural value, although the device of labeling them as merely metaphorical sometimes indexes the attempt to do so. A poetics approach is not limited to analyses of the figuration of methodological argument but, as we have said, brings under scrutiny the associated commonplaces of the tropes of everyday life (including, but not privileging, that of the methodologist). Relevant here, in terms of both their cultural history and their role in everyday understanding, are analyses of tree (Fernandez 1998) and bird imagery (Friedrich 1997). We find the tree trope in a wide variety of social and intellectual circumstances, from family trees and roots and other specimens from the genealogical arboretum to tree diagrams in logic and linguistics and the quasi-totemic identification of social groups with trees—a set of associations worked out extensively in The Golden Bough (Frazer 1890). The presence of birds as figurations of the human condition has deep classical antecedents and is also a widespread trope of power in other cultures, such as, for example, those so vividly presented in mythological form in Robert Gardner’s (1983) New Guinea ethnographic film, Dead Birds. It is hardly possible to understand the New Guinea people’s sense of the qualities and perimeters of their lives, their quality space, and their life-world generally without taking into account this master trope (see also Feld 1982). But then birds in many cultures, whether in the bush or in the hand, are prime figurative references that many men and women may grasp as they reach to understand the complexities, the fulfillment, and the fatalities of their lives. The question “What is it like to be a human?” corresponds to the widespread attempt to figure out what it is like to be a tree or a bird or a bat (Nagel 1974). Such queries move decisively beyond the recognition of a particular creature as fundamental to the self-depiction of an entire people, as in Evans-Pritchard’s (1940, 19) characterization of Nuer interaction as couched in a “bovine idiom,” but they draw, directly or otherwise, on grounded insights such as his. Often, too, such figuring out takes the form of anthropomorphisms of various kinds (Guthrie 1993). The poetic possibilities of cultural self-understanding of what it is to be human—to be, that is, a human animal—are seemingly without limit, although the interplay of inventive elaboration

14_107-Bernard.indb 80

6/9/14 10:02 AM

2: In Search of Meaningful Methods

81

with perceived cultural rules—the play of invention and convention, again—defines the domain of any attempt at a cultural poetics. CONCLUSION

It is the mark of a man experienced in the world and interested in studying it to look for no more precision than the subject matter allows. —Aristotle, The Nichomachean Ethics

We end on an exclusively reflexive and recursive note, by concluding with a poetics of methodology itself. An honest methodology located almost anywhere between Popper and Foucault (if not at the solipsistic extremes beyond these authors and their followers) must recognize its own entailment in figuration and metaphor, as Max Black (1962) noted. When we say that a method is meaningful, we are assessing its significance; this is not a confusion of meaning qua semantics and meaning qua social importance. That distinction is itself contingent and culturally specific; there are cultures in which it is either unrecognized or explicitly dismissed as untenable (e.g., Herzfeld 1985). Thus, to say that a method is meaningful presupposes a social and cultural context; any methodology that claims to be supracultural (i.e., context independent) is an exercise in circularity and self-deception. A method is a kind of practice in the immediate, as open to reflexive analyses as any other practice. Its elaborations of procedures of discovery are hardly sui generis and cannot be satisfied with the pursuit of reputation and esteem among methodologists alone. An honest method must inevitably reflect on the material and social metaphors on which its practices and its ideas about its practices are based and on the social and historical conditions that enable its emergence (Herzfeld 1987; Lave 2011), conditions that may include forms of activism lying far from public view and buried in the humdrum routines of everyday life (Pink 2012). Moreover, it must take account of the relation between the commitment to formal procedure, which is characteristic of the practice of any methodology, and the facts of the contingency and open-endedness of much of life as it is lived—an inchoateness that motivates much figuration in thought and practice (Fernandez 1974). Methodology is itself motivated by contingency; it can be viewed as a coping response to the inchoate in the sense that it seeks explicit procedures and practices as free as possible from the conditions of the life that motivates it. We say “as free as possible” because we recognize that such freedom is always relative, indeterminate, and subject to empirical denial (“falsification,” in a sense expanded to a greater degree of provisionality than that intended by Popper). The view we take here, however, is that any methodology, to be judged an adequate approach to the human condition, should embrace the actuality of the contingency that besets that condition and should recognize the often necessarily unpredictable ways in which people cope with such contingencies (e.g., Malaby 2003). Methodology in anthropology, if not always in the human sciences generally, should itself be a coping practice—a way of coping with human copings with the inchoate, not a denial of this rich actuality. In other words, methodology in anthropology (in effect, method in ethnographic practice and in ethnographic interpretation and explanation) shares something fundamental with the human condition that constitutes its own subject; it,

14_107-Bernard.indb 81

6/9/14 10:02 AM

82

James W. Fernandez and Michael Herzfeld

too, entails “communities of practice” (Grasseni 2004, 2007). Method does not escape the human condition and should not mystify itself with the hope of such platonic transcendence, which is a version of the logical error of misplaced concreteness. This point has been made with some insistence by Bourdieu (e.g., 1990). He and others have repeatedly shown how highly structured theories of theorists and methodologists are themselves artifactual effects of a lack of practical mastery (Bourdieu 1977) and of a neglect of the social conditions that make social science possible (Jenkins 1994). These concerns notwithstanding, we have tried to identify some explicit procedures and processes here. Precisely because we view methodology as a form of socially embedded practice, our “militant middle ground” posture requires us to acknowledge both the importance of establishing some norms and the virtual impossibility of stabilizing them against the corrosion of time and doubt. Now that ethnographers no longer confine themselves to reputedly isolated communities but work in bureaucracies, science laboratories (e.g., Traweek 1988), advertising agencies (Mazzarella 2003), elite families (e.g., Marcus 1992), and international organizations, they have been forced to confront the inevitable partiality of their research and to turn it to practical advantage. As Stacia Zabusky, one of the few anthropologists to have commented explicitly on this methodological issue in the context of analyzing large-scale international organizations, remarks of her work in the European Space Agency: To be inside practice . . . is to be situated somewhere and somewhen in particular; but to be inside practice is also the only way to come to know, ethnographically, the social system, since the system is replicated and reproduced through agents’ everyday actions, evasions, and enunciations of value. (1995, 42)

For her own work, Zabusky chose a “crossroads” department, through which the partiality of her sample was itself made explicit and, consequently, accessible to analysis as a tangible aspect of the indeterminacy as well as the cultural values and conflicts that characterized her chosen ethnographic situation. (We should also note that the intensity and intimacy of even the most unstructured ethnographic work represents an implicitly statistical assessment as much as any numerical sampling device. And see the discussion by Jeffrey Johnson and Daniel Hruschka [this volume] of Zabusky [1995].) Given that Zabusky’s focus was on cooperation, choosing a site where the limits of that value in practice would become palpable precisely because of the intermediary position of that site was methodologically appropriate and, in the event, highly productive. It is also worth noting that fieldwork is itself always, as Zabusky’s methodological statement makes abundantly clear, a synecdochical activity, a sampling that local actors can always accuse of unrepresentativeness—thereby, perhaps, confirming the accuracy of its perceptions beneath the surfaces of propriety and cultural defensiveness. In this respect, it is worth emphasizing that methods are enmeshed in “the politics of significance,” in which objections to non-Cartesian methods are often linked with surprising fixity of purpose to nationalistic and other political ideologies (Herzfeld 1997b). Today there is a rich array of ethnographies that exemplify the attention we have recommended here to a less reductive and more tropologically focused approach that attends specifically to what we call social poetics. Notable examples include Bate

14_107-Bernard.indb 82

6/9/14 10:02 AM

2: In Search of Meaningful Methods

83

(2009), Boyer (2005a), Malaby (2003), Medeiros (2013), Neofotistos (2012), Pesmen (2000), and Živković (2011). Such a list is far from exhaustive, but it indicates the spreading enthusiasm for semantic and poetic approaches within the discipline, representing a significant expansion of possibilities. These ethnographies fully engage both the communicative interaction ongoing in particular cultures and awareness of the fleeting nature of social certainty and undisputed routine. This is work that responds to the argumentativeness and ambiguity of everyday life and utilizes the imaginative resources available for framing contestatory moments and making them intelligible. One particular development perhaps deserves particular mention here. In the last seven or eight years, the International Rhetoric Culture Project, a project coordinated by the German–American anthropological team of Ivo Strecker and Stephen Tyler, has generated an ambitious publication program in which many of the concerns adumbrated here are given methodological specificity. In the first publication in this series, Culture and Rhetoric, Strecker and Tyler (2009) set the theoretical justifications for making rhetoric, that ancient discipline of civic art and social practice, fundamental to the contemporary study of social intercommunication and cultural creation. Subsequent volumes published in 2009 to 2011 cover issues of rhetoric’s role in confronting contingency and the vicissitudes of life (Carrithers 2009a), the place of the persuasive arts in gaining allegiance to economic understandings (Gudeman 2009), and the place of rhetoric in culture creation itself (Meyer and Girke 2011). A number of volumes are still to follow that focus on the problems of practicing meaningful methods of ethnographic understanding and ethnological comprehension. The approach, which is tolerant of a considerable amount of internal variation, leaves as yet unresolved— inchoate?—the question of whether culture is ever not rhetorical, much as the earlier coinage of “expressive culture” never fully resolved the same conundrum. But what gives projects like this a degree of cogency is the effort of pushing beyond the obvious. That effort reverts to anthropology’s fundamental commitment to challenge ethnocentric and scientistic assumptions that seek to treat a culturally and historically particular understanding of common sense as universal. After all, if culture is (in the Saussurean sense) arbitrary, it must always also be contingent. And that makes claims of universality highly parochial—an irony that instantiates the importance of attending to the play of tropes in our own epistemologies. Awareness of the uses of the poetic imagination in everyday life is central to the understanding of both the situation of ethnographic subjects and ethnographers alike. The meaningful methods we espouse flourish through an alertness to the contingencies of everyday life, with concomitant attention to how ethnographers apprehend them both in the course of fieldwork and in the ethnographic afterlife of the written word— and, we should now add, in visual and sound media as well. The ethnographies that we produce are themselves arguments never far removed from the problem of figuring out what was going on in the field and thereby obliging us to reflect on the organizing tropes involved in these figurations. Methodology, we have argued, is a kind of figuring out, a coping practice, motivated by the need to understand not only a culture’s routines but also by the constancy of contingency, contestation, and, again, sheer inchoateness in everyday life. Our methods are consequently also contested, contingent, and sometimes a great deal more inchoate than the literalists among us would wish to

14_107-Bernard.indb 83

6/9/14 10:02 AM

84

James W. Fernandez and Michael Herzfeld

recognize. The difficulty of addressing the question of intersubjectivity, already alluded to, exemplifies this inescapable fact. Inchoateness may not seem a promising target or organizing metaphor for methodology. However, to ignore its radical significance would be decidedly unempirical. It is often the condition of the very possibility of method, and as often will be found to contain the seeds out of which meaningful narratives emerge to give purpose and project to existence (Carrithers 2009b). And it certainly corresponds in its refusal of arbitrary closure to such natural-science understandings as complementarity (Bohr) and uncertainty (Heisenberg). The practical issue—and methodology is about (indeed, is) a variety of practice—is simply that an accurate and empirically sound representation of social life requires some means of recognizing the “fuzzy” and “indeterminate” social matrix out of which emerge graspable and explicit meanings. Vico, whose path we have followed so assiduously here, certainly recognized a contingent sense in which truth claims could be established, for his assertion of the necessary imperfection of human knowledge, ultimately grounded though his perspective was in theological preoccupations, does not depart radically from positivistic models of falsifiability except in its anticipation of Foucault’s explorations of the relationship between discourse and power—and this is the crucial, enabling condition for the emergence of a methodologically explicit social and cultural poetics such as we have sketched here. We also insist on recognizing the kinship of our own methodology with the understandings of the people we study. Both, in fact, are rooted in tropes of various kinds—in the groping toward a match between perception and reality that is definitive of the human condition—and much can be inferred from our own embodied experience (Herzfeld 2009; Jackson 1989; Lave 2011; Leavitt 1996; Seremetakis 1993; Stoller 1997). These tropes are most clearly brought into focus by play, which is central to social life and allows people to resist the intellectual iron cage of reductionism. This applies equally to Thais repudiating the representation of death in terms of exchange value (Klima 2002), to Cretans subverting idealized harmony through their agonistic gambling (Malaby 2010, 369), to journalists drinking at their Stammtisch as they explore real-life philosophical conflicts between normativity and individuality (Boyer 2005a), and to anthropologists challenging the commonsensical horrors of bureaucratic “audit culture” (Strathern 2000). Neither the understandings of the people we study nor our own methodological choices ever reach definitive closure, which, in a Vichian epistemology, is contrary to knowledge. Argument, play, and tropes are among the most dynamic components of social life and of fieldwork alike. Ever attentive to the provisionality of our acts and deeds, then, we now propose the following stages in the research process as a plausible path for the attainment of poetic wisdom: 1. Elicitation and ethnographic learning on a model of apprenticeship (see Lave 2011), in which the ethnographer both internalizes and attempts to render explicit formal norms of social practices and interactions of the more routine kinds. 2. Recording commentaries on conformity and deviance in these norms synchronically and diachronically, with careful attention paid to the social relationships obtaining between commentators and their subjects and anthropologists and informants.

14_107-Bernard.indb 84

6/9/14 10:02 AM

2: In Search of Meaningful Methods

85

3. Probing, through imagined reconstruction and other techniques, for the images and organizing tropes that define specific norms as practical projects in the world and on which they may depend for their realization. 4. Through participant observation, relating this “poetic function” (the “play of meaning”) to political norms and practices, defined as the maintenance and contestation of systems of privilege, differential qualities of life, and hierarchized rights of choice and decision. 5. Anchoring these images and tropes in bodily experience and its vicissitudes (Carrithers 2009a), recognizing that it is from this corporeal locus that they derive their meaning, and exploring the system of entailments that constitutes the cultural logic present in social interaction and serves as the qualitative anchor of meaningfulness. (We add parenthetically that, in addition to the well-tried resources of narrative analysis, new recording devices such as video permit the exposure of field materials to multiple observation, making it possible to establish systematic features of interpretive differences by class, gender, age, cultural identity, and so forth.) 6. Tracing the more obvious narrative emplotments of social life, the social dramas, with their oppositional or agonistic workings-out and dénouements and their sometimes discomfiting entailment in encompassing tropes of revitalization, quality space, and so on.

These stages in the research process permit evaluation of what once were called the “expressive aspects” of mundane activity. For some time, anthropologists spoke of “expressive culture,” but we might now more reasonably ask, “When is culture not expressive?” In a semiotic anthropology of partially Peircean derivation (see, e.g., Fernandez 1965; Herzfeld 1983; Parmentier 1994; Silverstein 1976; Singer 1984), the symbolic is recognized only as the most explicitly meaning-conveying aspect, in contrast to the relational meanings of indexicality and the seemingly natural meanings of iconicity. In a sense, the shift from local social relations to national and ethnic identities on the grand scale is a shift from primarily indexical (“we are all related”) to primarily iconic (“we are all alike”) claims, in which the symbolic provides the grounds for contesting this view or interpreting it for specific advantages and goals (Herzfeld 2005, 29–30). Thus, we can say that the emphasis falls quite variably and in differing proportions on these three aspects of meaning, and we can only get a purchase on this slippery shape-shifting (Fernandez 1974) by remembering that they are indeed aspects of socially experienced reality. We should speak, not of icons, indexes, and symbols, but of iconicity, indexicality, and symbolism (see also Eco 1976, 190–217). This move resists the reification demanded by normativity in both reductive social science and equally reductive bureaucratic institutions. Concomitantly, it recognizes the shiftiness of human interaction and experience. But there is also an epistemological consequence. For such a perspective requires us to dispense with similar reifications of the meaningful in our own self-classification. Anthropologists have indeed long been engaged in replacing the old-fashioned segmentation of the discipline into separate fields of study, one of which is called symbolism, with a more integrative view. In this less-fragmented perspective, religious, economic, symbolic, and many other aspects of social activity are not reduced to the demarcation of separate functions that is held to typify modern, industrial, Western society. In pursuing that approach, our goal in this essay has been to suggest ways, not of

14_107-Bernard.indb 85

6/9/14 10:02 AM

86

James W. Fernandez and Michael Herzfeld

figuring out the symbolism of a given population as a finite subset of its cultural equipment, but of finding the semiotic implications, the figurations, of all social interaction. NOTES

1. Herzfeld was once challenged at a conference about his use of unnecessarily long words. The key offender turned out to be trope! Ironically, it would be hard to find a more eloquent demonstration of the inadequacy of referentiality as a model of socially recognizable. There is, in fact, a technical reason for preferring trope to the loose use of metaphor, and it is doubly ironic that those who display the most passionate allergy to trope are usually those who demand a more numerate form of precision in other areas of the discipline. 2. Festa (2007, 93 and abstract) analyzes how “men reanimate an officially orchestrated male totality, or martial imaginary, that reproduces idealized masculine values and patterns of citizenship.” 3. As editor of American Ethnologist (1995–98), Herzfeld was often struck by the effects of passive-voice constructions as a means of evading responsibility for analysis and data, and in most cases he asked authors to switch to a first-person mode. 4. It is our view that the distinction in anthropology between the two approaches, that of a social and cultural poetics as indicated, frames the fundamental dynamic of the poetics approach in anthropology. The journal Poetics (Amsterdam), which began publication in the early 1970s, although casting a wide net, has largely been devoted to the work of sociologists, social psychologists, and psychologists; only occasionally have anthropologists published in it. These scholars are interested in narrative processes and narrative devices and in measuring by various technical methods their consequences in the maintenance of social order and in the formation of public opinion. At the other extreme, Poetics Today (Tel-Aviv), also only occasionally a locus for anthropological work, is predominantly engaged in semiotic and literary criticism of verbal texts. Both journals thus share a basic concern with language, although their epistemological orientations are radically divergent. 5. This research, conducted in 1993–95, was supported by Award #9307421 of the National Science Foundation. 6. If Vico was right to suggest that all verbal expression was grounded in the immediacy of bodily experience, this grants the body a primacy emphasized in a recent cognitive linguistics (Johnson [1987] and see also, for a rather different approach, Farnell [1994, 1995] and Farnell and Graham, in this volume). Vico’s own method was strictly etymological: He traced political abstractions back to terms denoting physical images, arguing that vision itself was “immured in the body” and that, for this reason, intellectual activity could not be separated analytically from bodily sensation. 7. What the linkages mean for social actors is quite another matter—Austin’s phrase “trailing clouds of etymology” nicely captures the problem: The fact that we know that “having an accident” suggests the impersonal forces implied by the Latin verb accidit (implying something falling on one from on high) tells us absolutely nothing about the degree to which the social actors are themselves conscious of any such linkage, even though their actions may suggest it—and few anthropologists have bothered, or are adequately equipped to seek, the evidence for links of this sort. The problem is philological and psychological as well as social and cultural. REFERENCES

Abu El-Haj, N. 2001. Facts on the ground: Archaeological practice and territorial self-fashioning in Israeli society. Chicago: University of Chicago Press.

14_107-Bernard.indb 86

6/9/14 10:02 AM

2: In Search of Meaningful Methods

87

Appadurai, A. 1981. The past as a scarce resource. Man (n.s.) 16: 201–19. Ardener, E. 1971. Introductory essay: Social anthropology and language. In Social anthropology and language, ed. E. W. Ardener, ix–cii, A.S.A. Monographs, Vol. 10. London: Tavistock Publications. Ardener, E. 1975. The problem revisited. In Perceiving women, ed. S. Ardener, 19–27. London: J. M. Dent. Ardener, E. 2007. The voice of prophecy, 2nd ed. Oxford, England: Berghahn. Aristotle. The Nicomachian Ethics, Book I. Paragraph 3, trans. W. D. Ross. classics.mit.edu/ Aristotle/nichomachaen.1.i.html (accessed May 1, 2013). Austin, J. L. 1971. A plea for excuses. In Philosophy and linguistics, ed. C. Lyas, 79–101. London: Macmillan. Austin, J. L. 1975 [1962]. How to do things with words, 2nd ed., ed. J. O. Urmson and M. Sbisà. Cambridge, MA: Harvard University Press. Bachelard, G. 1964. The poetics of space. New York: Orion Press. Bahloul, J. 1996. The architecture of memory: A Jewish-Muslim household in colonial Algeria, 1937–1962. Cambridge: Cambridge University Press. Bailey, F. G. 1973. Debate and compromise: The politics of innovation. Oxford: Blackwell. Bailey, F. G. 1983. The tactical uses of passion: An essay on power, reason, and reality. Ithaca, NY: Cornell University Press. Bailey, F. G. 1988. Humbuggery and manipulation: The art of leadership. Ithaca, NY: Cornell University Press. Bailey, F. G. 1994. The witch-hunt, or, the triumph of morality. Ithaca, NY: Cornell University Press. Barnett, M. N. 1997. The UN Security Council, indifference, and genocide in Rwanda. Cultural Anthropology 12: 551–78. Bate, J. B. 2009. Tamil oratory and the Dravidian aesthetic: Democratic practice in South India. New York: Columbia University Press. Bauman, R. 1984 [1977]. Verbal art as performance. Prospect Heights, IL: Waveland. Bender, B., with Paul Aitken et al. 1998. Stonehenge: Making space. Oxford: Berg. Black, M. 1962. Models and metaphors: Studies in language and philosophy. Ithaca, NY: Cornell University Press. Blier, S. P. 1994. The anatomy of architecture: Ontology and metaphor in Batammaliba architectural expression. Chicago: University of Chicago Press. Bloch, M., ed. 1975. Political language and oratory in traditional society. London: Academic Press. Boehm, C. 1980. Exposing the moral self in Montenegro: The use of natural definitions to keep ethnography descriptive. American Ethnologist 7: 1–26. Bond, G. C., and A. Gilliam, eds. 1994. Social construction of the past: Representation as power. New York: Routledge. Boon, J. A. 1982. Other tribes, other scribes: Symbolic anthropology in the comparative study of cultures, histories, religions, and texts. Cambridge: Cambridge University Press. Borofsky, R. 1997. Cook, Lono, Obeyesekere, and Sahlins. Current Anthropology 38: 255–82. Bourdieu, P. 1973. The Kabyle house. In Rules and meanings, ed. M. Douglas, 98–110. Harmondsworth, UK: Penguin. Bourdieu, P. 1977. Outline of a theory of practice. Cambridge: Cambridge University Press. Bourdieu, P. 1984. Distinction: A social critique of the judgment of taste. Cambridge, MA: Harvard University Press. Bourdieu, P. 1990. In other words: Essays towards a reflexive sociology. Cambridge: Polity.

14_107-Bernard.indb 87

6/9/14 10:02 AM

88

James W. Fernandez and Michael Herzfeld

Boyer, D. 2000. On the sedimentation and accreditation of social knowledges of difference: Mass media, journalism, and the representation of East/West alterities in unified Germany. Cultural Anthropology 15: 459–91. Boyer, D. 2005a. Spirit and system: Media, intellectuals, and the dialectic in modern German culture. Chicago: University of Chicago Press. Boyer, D. 2005b. The corporeality of expertise. Ethnos 70: 243–66. Brady, I. 2004. In defense of the sensual: Meaning construction in ethnography and poetics. Qualitative Inquiry 10: 622–44. Braudel, F. 1980. On history. Chicago: University of Chicago Press. Bruner, E. M. 1994. Abraham Lincoln as authentic reproduction: A critique of post-modernism. American Anthropologist 96: 397–415. Bruner, E. M., and P. Gorfain. 1984. Dialogue narration and the paradoxes of Masada. In Text, play, and story, ed. S. Plattner, 56–71. Washington, DC: American Ethnological Society. Bubandt, N. 1998. The odour of things: Smell and the cultural elaboration of disgust in eastern Indonesia. Ethnos 63: 48–80. Candea, M. 2007. Arbitrary docations: In defence of the bounded field-site. Journal of the Royal Anthropological Institute 13: 167–84. Carrithers, M., ed. 2009a. Culture, rhetoric, and the vicissitudes of life. New York: Berghahn. Carrithers, M. 2009b. Story seeds and the inchoate. In Culture, rhetoric, and the vicissitudes of life, ed. M. Carrithers, 34–52. New York: Berghahn. Classen, C. 1993. Worlds of sense: Exploring the senses in history and across cultures. London: Routledge. Clifford, J. 1983. On ethnographic authority. Representations 1: 118–46. Clifford, J., and G. E. Marcus, eds. 1986. Writing culture: The poetics and politics of ethnography. Berkeley: University of California Press. Cohen, A. P. 1974. Two-dimensional man: An essay on the anthropology of power and symbolism in complex society. Berkeley: University of California Press. Cohen, A. P. 1994. Self-consciousness: An alternative anthropology of identity. London: Routledge. Colby, B. N., J. W. Fernandez, and D. Kronenfeld. 1981. Toward a convergence of cognitive and symbolic anthropology. American Ethnologist 8: 442–50. Cowan, J. K. 1990. Dance and the body politic in northern Greece. Princeton, NJ: Princeton University Press. Coy, M., ed. 1989. Apprenticeship: From theory to method and back again. Albany: SUNY Press. Crain, M. M. 1989. Ritual memoria popular y proceso político en la sierra ecuatoriana. Quito: Corporación Editora Nacional: Ediciones Abya-Yala. Crick, M. 1976. Explorations in language and meaning: Towards a semantic anthropology. New York: John Wiley. Csordas, T. J. 1994. Embodiment and experience: The existential ground of culture and self. Cambridge: Cambridge University Press. Das, V., A. M. Kleinman, and M. Lock, eds. 1997. Social suffering. Berkeley: University of California Press. de Jorio, A. 1832. La mimica degli antichi investigata nel gestire napoletano. Naples, Italy: Fibreno. Delbos, G., and P. Jorion. 1984. La transmission des savoirs, Collection ethnologie de la France. Paris: Editions de la Maison des Sciences de l’Homme. Desjarlais, R. 1992. Body and emotion: The aesthetics of illness and healing in the Nepal Himalayas. Philadelphia: University of Pennsylvania Press.

14_107-Bernard.indb 88

6/9/14 10:02 AM

2: In Search of Meaningful Methods

89

Dolgin, J. D., D. S. Kemnitzer, and D. M. Schneider, eds. 1977. Symbolic anthropology: A reader in the study of symbols and meanings. New York: Columbia University Press. Dougherty, J. W. D. 1985. Directions in cognitive anthropology. Urbana: University of Illinois Press. Douglas, M. 1966. Purity and danger: An analysis of concepts of pollution and taboo. London: Routledge & Kegan Paul. Douglas, M. 1973. Rules and meanings: The anthropology of everyday knowledge. Harmondsworth, UK: Penguin. Douglas, M. 1975. Implicit meanings: Essays in anthropology. London: Routledge & Paul. Douglas, M. 1992. How classification works: Nelson Goodman among the social sciences. Edinburgh: Edinburgh University Press. Dresch, P. 1986. The significance of the course events take in segmentary systems. American Ethnologist 13: 309–24. Dundes, A, J. W. Leach, and B. Özkök. 1972. The strategy of Turkish boys’ verbal dueling rhymes. In Directions in sociolinguistics: The ethnography of communication, ed. J. J. Gumperz and D. Hymes, 130–60. New York: Holt, Rinehart and Winston. Durkheim, E. 1976. The elementary forms of the religious life, 2nd ed. London: Allen and Unwin. Eco, U. 1976. A theory of semiotics. Bloomington: Indiana University Press. Evans-Pritchard, E. E. 1940. The Nuer: A description of the modes of livelihood and political institutions of a nilotic people. Oxford: Clarendon Press. Fabian, J. 1983. Time and the Other: How anthropology makes its object. New York: Columbia University Press. Farnell, B. M. 1994. Ethno-graphics and the moving body. Man (n.s.) 29: 929–74. Farnell, B. M. 1995. Do you see what I mean?: Plains Indian sign talk and the embodiment of action. Austin: University of Texas Press. Farnell, B. M., and D. Williams. 1990. The Laban script. Canberra: Australian Institute of Aboriginal and Torres Strait Islander Studies. Feld, S. 1982. Sound and sentiment: Birds, weeping, poetics, and song in Kaluli expression. Philadelphia: University of Pennsylvania Press. Ferguson, J. W. 1990. The anti-politics machine: Development, depoliticization, and bureaucratic power in Lesotho. Cambridge: Cambridge University Press. Fernandez, J. W. 1965. Symbolic consensus in a Fang reformative cult. American Anthropologist 67: 902–27. Fernandez, J. W. 1974. The mission of metaphor in expressive culture. Current Anthropology 15: 119–45. Fernandez, J. W. 1977a. Poetry in motion: Being moved by amusement, by mockery, and by mortality in the Asturian countryside. New Literary History 8: 459–83. Fernandez, J. W. 1977b. Fang architectonics. Working Papers in the Traditional Arts, No. 1. Philadelphia: Institute for the Study of Human Issues. Fernandez, J. W. 1982. Bwiti: An ethnography of the religious imagination in Africa. Princeton, NJ: Princeton University Press. Fernandez, J. W. 1986. Persuasions and performances: The play of tropes in culture. Bloomington: Indiana University Press. Fernandez, J. W. 1992a. What I learned from the parrot’s egg and the bull who crashes in the kraal: The senses of time binding and turn taking in being with the Other. In Dialectical anthropology: Essays in honor of Stanley Diamond, ed. C. Gailey, 209–16. Gainesville: University of Florida Press. Fernandez, J. W. 1992b. Architectonic inquiry: Review article. Semiotica 89: 215–26.

14_107-Bernard.indb 89

6/9/14 10:02 AM

90

James W. Fernandez and Michael Herzfeld

Fernandez, J. W. 1993. Advice to the perplexed ethnographer in an age of soundbites. American Ethnologist 20: 179–84. Fernandez, J. W. 1994. Spielerisch un planvoll (Playful and Planful): Zur Theorie der Tropen in der Anthropologie. Historische Anthropologie 2: 1–19. Fernandez, J. 1998. On trees of knowledge of self and other in culture: Models for the moral imagination. In The social life of trees, ed. L. Rival and M. Bloch, 81–110. London: Berghahn. Fernandez, J. W. 2001. The irony of complicity and the complicity of irony in development discourse. In Irony in action: Anthropology, practice and the moral imagination, ed. J. W. Fernandez and M. T. Huber, 54–85. Chicago: University of Chicago Press. Fernandez, J. W. 2006. Down the garden path: On “path-ologies” of inquiry and of “progress” in understanding (Por el sendero del huerto: De “las patologías” de la investigación y sobre el “progreso” del conocimiento). Revista de Antropología Social (Madrid) 15: 273–304. Festa, P. 2007. Mahjong agonistics and the political public in Taiwan: Fate, mimesis, and the martial imaginary. Anthropological Quarterly 80: 93–125. Frazer, J. G., Sir. 1890. The golden bough: A study in comparative religion. London: Macmillan. Friedrich, P. 1996. Culture in poetry and the poetry in culture. In Culture/contexture: Explorations in anthropology and literary studies, ed. E. V. Daniel and J. N. Peck, 37–38. Berkeley: University of California Press. Friedrich, P. 1997. An avian and aphrodisian reading of the Odyssey. American Anthropologist 99: 306–20. Gable, E., R. Handler, and A. Lawson. 1992. On the uses of relativism: facts, conjecture, and black and white histories at colonial Williamsburg. American Ethnologist 19: 791–805. Gardin, J.-C. 1980. Archaeological constructs: An aspect of theoretical archaeology. Cambridge: Cambridge University Press. Geertz, C. 1973. The interpretation of cultures: Selected essays. New York: Basic Books. Gero, J. 1991. Engendering archaeology: Women and prehistory. Social Archaeology Series. Oxford: Basil Blackwell. Gibbs, R. W. 1994. The poetics of mind: Figurative thought, language, and understanding. Cambridge: Cambridge University Press. Giddens, A. 1984. The constitution of society: Outline of the theory of structuration. Cambridge: Polity Press. Goldstein, L. J. 1976. Historical knowing. Austin: University of Texas Press. Gombrich, E. H. 1979. The sense of order: A study in the psychology of decorative art. The Wrightsman Lectures, No. 9. Ithaca, NY: Cornell University Press. Grasseni, C. 2004. Skilled vision: An apprenticeship in breeding aesthetics. Social Anthropology 12: 41–55. Grasseni, C., ed. 2007. Skilled visions: Between apprenticeship and standards. Oxford: Berghahn. Gudeman, S., ed. 2009. Economic persuasions. New York: Berghahn. Guthrie, S. E. 1995. Faces in the clouds: A new theory of religion. New York: Oxford. Handelman, D. 1990. Models and mirrors: Towards an anthropology of public events. Cambridge: Cambridge University Press. Handelman, D. 2004. Nationalism and the Israeli state: Bureaucratic logic in public events. Oxford: Berg. Harris, M. 1974. Cows, pigs, wars and witches: The riddles of culture. New York: Random House. Herzfeld, M. 1983. Looking both ways: The ethnographer in the text. Semiotica 46: 151–66. Herzfeld, M. 1985. The poetics of manhood: Contest and identity in a Cretan mountain village. Princeton, NJ: Princeton University Press. Herzfeld, M. 1987. Anthropology through the looking-glass: Critical ethnography in the margins of Europe. Cambridge: Cambridge University Press.

14_107-Bernard.indb 90

6/9/14 10:02 AM

2: In Search of Meaningful Methods

91

Herzfeld, M. 1991. A place in history: Social and monumental time in a Cretan town. Princeton, NJ: Princeton University Press. Herzfeld, M. 1992a. The social production of indifference: Exploring the symbolic roots of Western bureaucracy. Oxford: Berg. Herzfeld, M. 1992b. Metapatterns: Archaeology and the uses of evidential scarcity. In Representations in archaeology, ed. J.-C. Gardin and C. S. Peebles, 66–86. Bloomington: Indiana University Press. Herzfeld, M. 1997a. Portrait of a Greek imagination: An ethnographic biography of Andreas Nenedakis. Chicago: University of Chicago Press. Herzfeld, M. 1997b. Anthropology and the politics of significance. Social Analysis 41: 107–38. Herzfeld, M. 2004. The body impolitic: Artisans and artifice in the global hierarchy of value. Chicago: University of Chicago Press. Herzfeld, M. 2005. Cultural intimacy: Social poetics in the nation-state, 2nd ed. New York: Routledge. Herzfeld, M. 2007. Deskilling, “dumbing down,” and the auditing of knowledge in the practical mastery of artisans and academics: An ethnographer’s response to a global problem. In Ways of knowing: New approaches in the anthropology of experience and learning, ed. M. Harris, 91–110. Oxford: Berghahn. Herzfeld, M. 2009. The politics of gesture: Reflections on the embodiment of ethnographic practice. Ethnography 10: 131–52. Hill, J. 1988. Rethinking history and myth: Indigenous South American perspectives on the past. Urbana: University of Illinois Press. Hill, J. 1996. History, power, and identity: Ethnogenesis in the Americas, 1492–1992. Iowa City: University of Iowa Press. Hodgen, M. T. 1964. Early anthropology in the sixteenth and seventeenth centuries. Philadelphia: University of Pennsylvania Press. Howes, D., ed. 1991. The varieties of sensory experience. Toronto: University of Toronto Press. Jackson, M. 1983. Knowledge of the body. Man (n.s.) 18: 327–45. Jackson, M. 1989. Paths toward a clearing: Radical empiricism and ethnographic inquiry. Bloomington: Indiana University Press. Jakobson, R. 1960. Linguistics and poetics. In Style in language, ed. T. A. Sebeok, 350–77. Cambridge, MA: MIT Press. Jakobson, R. 1985. Verbal art, verbal sign, verbal time. In Collected essays of Roman Jakobson, ed. K. Pomorska and S. Rudy, assisted by B. Vine, 35–68. Minneapolis: University of Minnesota Press. Jameson, F. 1981. The political unconscious: Narrative as a socially symbolic act. Ithaca, NY: Cornell University Press. Jelinek, C. E., ed. 1980. Women’s autobiography: Essays in criticism. Bloomington: Indiana University Press. Jenkins, T. 1994. Fieldwork and the perception of everyday life. Man (n.s.) 29: 433–55. Johnson, I. 2010. New mosaics on old walls: Intimacy and architecture along the Malaysian-Thai frontier. Anthropological Quarterly 83: 239–77. Johnson, M. 1987. The body in the mind: The bodily basis of meaning, imagination, and reason. Chicago: University of Chicago Press. Jung, Y. 2010. The inability not to follow: Western hegemonies and the notion of “complaisance” in the enlarged Europe. Anthropological Quarterly 83: 317–53. Kapferer, B. 1988. Legends of people, myths of state: Violence, intolerance, and political culture in Sri Lanka and Australia. Washington, DC: Smithsonian Institution Press.

14_107-Bernard.indb 91

6/9/14 10:02 AM

92

James W. Fernandez and Michael Herzfeld

Karp, I., and M. B. Kendall. 1982. Reflexivity and fieldwork. In Explaining human behavior, ed. P. Secord, 249–73. Los Angeles: Sage. Kendon, A. 1995. Andrea de Jorio: The first ethnographer of gesture? Visual Anthropology 7: 371–90. Kendon, A., ed. 2000. Gesture in Naples and gesture in classical antiquity: A translation of La mimica degli antichi investigata nel gestire napoletano (Gestural expression of the ancients in the light of Neapolitan gesturing). Bloomington: Indiana University Press. Kertzer, D. I. 1988. Ritual, politics, and power. New Haven, CT: Yale University Press. Klapp, O. E. 1972. Heroes, villains, and fools: Reflections of the American character. San Diego, CA: Aegis. Kleinman, A. M. 2006. What really matters: Living a moral life amidst uncertainty and danger. New York: Oxford University Press. Kleinman, A. M., and J. Kleinman. 1994. How bodies remember: Social memory and social experience of criticism, resistance, and delegitimation following China’s Cultural Revolution. New Literary History 25: 707–23. Klima, A. 2002. The funeral casino: Meditation, massacre, and exchange with the dead in Thailand. Princeton, NJ: Princeton University Press. Kondo, D. K. 1990. Crafting selves: Power, gender, and discourses of identity in a Japanese workplace. Chicago: University of Chicago Press. Kroeber, A. L., and J. Richardson. 1952 [1940]. Three centuries of women’s dress fashions, a quantitative analysis. Anthropological Records 5: 111–54. (Reprinted in The nature of culture, 358–72, Chicago: University of Chicago Press.) Lakoff, G. 1987. Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press. Lakoff, G., and M. Johnson. 1980. Metaphors we live by. Chicago: University of Chicago Press. Lave, J. 2011. Apprenticeship in critical ethnographic practice. Chicago: University of Chicago Press. Lave, J., and E. Wenger. 1991. Situated learning: Legitimate peripheral participation. Learning in doing. Cambridge: Cambridge University Press. Lawrence, D., and S. M. Low 1995. Built environment and spatial form. Annual Review of Anthropology 19: 453–505. Leach, E. R. 1970. Claude Lévi-Strauss. Modern masters. New York: Viking Press. Leach, E. R. 1976. Culture and communication: The logic by which symbols are connected. An introduction to the use of structuralist analysis in social anthropology. Cambridge: Cambridge University Press. Leavitt, J. 1996. Meaning and feeling in the anthropology of emotions. American Ethnologist 23: 514–39. Leone, M. P. 1988. The recovery of meaning: Historical archaeology in the eastern United States. Anthropological Society of Washington Series. Washington, DC: Smithsonian Institution Press. LeVine, R. A. 1973. Culture, behavior, and personality. Chicago: Aldine. Lévi-Strauss, C. 1963. Structural anthropology. New York: Basic Books. Limón, J. E. 1994. Dancing with the devil: Society and cultural poetics in Mexican-American south Texas. Madison: University of Wisconsin Press. Low, S. M., and D. Lawrence-Zúñiga. 2003. Locating culture. In The anthropology of space and place: Locating culture, ed. Setha M. Low and Denise Lawrence-Zúñiga, 1–47. Malden, MA: Blackwell. Low, S. M., and S. E. Merry. 2010. Engaged anthropology: Diversity and dilemmas. Current Anthropology 51(S2): 203–26.

14_107-Bernard.indb 92

6/9/14 10:02 AM

2: In Search of Meaningful Methods

93

Mack, J. 2012. Producing the public: Architecture, urban planning, and immigration in a Swedish town, 1965 to the present. Ph.D. diss., Harvard University. Malaby, T. M. 2003. Gambling life: Dealing in contingency in a Greek city. Urbana: University of Illinois Press. Malaby, T. M. 2010. The second life of institutions: Social poetics in a digital state. Anthropological Quarterly 83: 355–71. Malkki, L. H. 1995. Purity and exile: Violence, memory, and national cosmology among Hutu refugees in Tanzania. Chicago: University of Chicago Press. Marchand, T. H. J. 2001. Minaret building and apprenticeship in Yemen. Richmond, VA: Curzon. Marchand, T. H. J. 2009. The masons of Djenné. Bloomington: Indiana University Press. Marcus, G. E. 1992. Lives in trust: The fortunes of dynastic families in late twentieth-century America. Boulder, CO: Westview Press. Marx, K. 1978 [1852]. The eighteenth Brumaire of Louis Bonaparte, 1st ed. Beijing: Foreign Languages Press. Mauss, M. 1936. Sociologie et anthropologie. Paris: Presses Universitaires de France. Mazzarella, W. 2003. Shoveling smoke: Advertising and globalization in contemporary India. Durham, NC: Duke University Press. Medeiros, A. 2013. Two sides of one river: Nationalism and ethnography in Galicia and Portugal. Oxford: Berghahn. Meeker, M. E. 1979. Literature and violence in north Arabia. Cambridge: Cambridge University Press. Merleau-Ponty, M. 1962. Phenomenology of perception. New York: Humanities Press. Meyer, C., and F. Girke, eds. 2011. The rhetorical emergence of culture. New York: Berghahn. Miceli, S. 1982. In nome del segno: Introduzione alla semiotica della cultura. Palermo, Italy: Sellerio. Mintz, J. R. 1997. Carnival song and society: Gossip, sexuality, and creativity in Andalusia. Oxford: Berg. Mole, J. 1990. Mind your manners: Culture clash in the European single market. London: Industrial Society. Mole, J. 1992. Mind your manners: Managing culture clash in the European single market. London: Nicholas Brealey. Mole, J. 1995. Mind your manners: Managing business cultures in Europe, new ed. London: Nicholas Brealey. Moore, S. F. 1987. Explaining the present: Theoretical dilemmas in processual ethnography. American Ethnologist 14: 727–36. Nagel, T. 1974. What is it like to be a bat? The Philosophical Review 83: 435–50. Needham, R. 1962. Structure and sentiment: A test case in social anthropology. Chicago: University of Chicago Press. Needham, R. 1972. Belief, language, and experience. Oxford: Basil Blackwell. Neofotistos, V. P. 2010. Cultural intimacy and subversive disorder: The politics of romance in the Republic of Macedonia. Anthropological Quarterly 83: 279–315. Neofotistos, V. P. 2012. The risk of war: Everyday sociality in the Republic of Macedonia. Philadelphia: University of Pennsylvania Press. Nisbet, R. A. 1965. Social change and history: Aspects of the Western theory of development. New York: Oxford University Press. Obeyesekere, G. 1992. The apotheosis of Captain Cook: European mythmaking in the Pacific. Princeton, NJ: Princeton University Press. Osgood, C. E., G. J. Suci, and P. H. Tannenbaum. 1957. The measurement of meaning. Urbana: University of Illinois Press.

14_107-Bernard.indb 93

6/9/14 10:02 AM

94

James W. Fernandez and Michael Herzfeld

Paine, R. 1981. Politically speaking: Cross-cultural studies of rhetoric. Philadelphia: Institute for the Study of Human Issues. Parkin, D., ed. 1982. Semantic anthropology. A.S.A. Monographs, Vol. 22. London: Academic Press. Parmentier, R. J. 1994. Signs in society: Studies in semiotic anthropology. Bloomington: Indiana University Press. Parmentier, R. J. 1996. Haole-ing in the wind: On the rhetoric of identity anthropology. Anthropological Quarterly 69: 220–30. Peacock, J. L. 1975. Consciousness and change: Symbolic anthropology in evolutionary perspective. New York: John Wiley. Pesmen, D. 2000. Russia and soul: An exploration. Ithaca, NY: Cornell University Press. Peters, E. 1967. Some structural aspects of the feud among the camel-herding Bedouin of Cyrenaica. Africa 37: 261–82. Pink, S. 2012. Situating everyday life: Practices and places. Los Angeles: Sage. Pirsig, R. M. 1974. Zen and the art of motorcycle maintenance: An inquiry into values. New York: Morrow. Pocius, G. L. 1979. Hooked rugs in Newfoundland: The representation of social structure in design. Journal of American Folklore 92: 273–84. Preziosi, D. 1979. Architecture, language and meaning: The origins of the built world and its semiotic organization. Approaches to Semiotics, No. 49. The Hague: Mouton. Quine, W. V. 1981. Theories and things. Cambridge, MA: Harvard University Press. Rabinow, P. 1996. Making PCR: A story of biotechnology. Chicago: University of Chicago Press. Rappaport, J. 1994. Cumbe reborn: An Andean ethnography of history. Chicago: University of Chicago Press. Rappaport, R. A. 1979. Ecology, meaning, and religion. Richmond, CA: North Atlantic Books. Rapport, N. 1993. Diverse world-views in an English village. Edinburgh: Edinburgh University Press. Reed-Danahay, D. E. 1995. The Kabyle and the French: Occidentalism in Bourdieu’s theory of practice. In Occidentalism: Images of the West, ed. J. G. Carrier, 61–84. Oxford: Oxford University Press. Reed-Danahay, D. E., ed. 1997. Auto/ethnography: Rewriting the self and the social. Oxford: Berg. Roberts, M. 1997. Histories. International Social Science Journal 49: 373–85. Rosen, L., ed. 1995. Other intentions: Cultural contexts and the attribution of inner states. Santa Fe: School of American Research Press. Sahlins, M. 1981. Historical metaphors and mythical realities: Structure in the early history of the Sandwich Islands Kingdom. ASAO Special Publications, No. 1. Ann Arbor: University of Michigan Press. Sahlins, M. 1985. Islands of history. Chicago: University of Chicago Press. Sahlins, M. 1995. How “natives” think: About Captain Cook, for example. Chicago: University of Chicago Press. Salmond, A. 1982. Theoretical landscapes: On cross-cultural conceptions of knowledge. In Semantic anthropology. A.S.A. Monographs, Vol. 22, ed. D. Parkin, 65–87. London: Academic Press. Salzman, P. Carl. 1978. Does complementary opposition exist? American Anthropologist 80: 53–70. Sanjek, R. 1990. Fieldnotes: The makings of anthropology. Ithaca, NY: Cornell University Press. Sartre, J.-P. 1963. Search for a method, trans. and intro. H. Barnes. New York: Knopf. Schama, S. 1995. Landscape and memory. New York: Knopf. Schwimmer, E., ed. 1978. The yearbook of symbolic anthropology, Vol. 1. London, C. Hurst & Co.

14_107-Bernard.indb 94

6/9/14 10:02 AM

2: In Search of Meaningful Methods

95

Scott, J. C. 1990. Domination and the arts of resistance: Hidden transcripts. New Haven, CT: Yale University Press. Scott, J. C. 1998. Seeing like a state: How certain schemes to improve the human condition have failed. New Haven, CT: Yale University Press. Seremetakis, C. N. 1993. Memory of the senses: Historical perception, commensal exchange and modernity. Visual Anthropology Review 9: 2–18. Sewell, W. H. 1996. Three temporalities: Towards an eventful sociology. In The historic turn in the human sciences, ed. T. J. McDonald, 245–80. Ann Arbor: University of Michigan Press. Shryock, A. 1997. Nationalism and the genealogical imagination: Oral history and textual authority in tribal Jordan. Berkeley: University of California Press. Silverstein, M. 1976. Shifters, linguistic categories, and cultural description. In Meaning in anthropology, ed. K. H. Basso and H. A. Selby, 11–55. Albuquerque: University of New Mexico Press. Singer, M. B. 1984. Man’s glassy essence: Explorations in semiotic anthropology. Bloomington: Indiana University Press. Soysal, L. 2010. Intimate engagements of the public kind. Anthropological Quarterly 83: 373–99. Sperber, D. 1975. Rethinking symbolism. Cambridge: Cambridge University Press Sperber, D. 1985. On anthropological knowledge: Three essays. Cambridge: Cambridge University Press. Stark, F. 1996. Communicative interaction, power and the state: A method. Toronto: University of Toronto Press. Steedly, M. M. 1993. Hanging without a rope: Narrative experience in colonial and postcolonial Karoland. Princeton, NJ: Princeton University Press. Stewart, K. 1996. A space on the side of the road: Cultural poetics in an “other” America. Princeton, NJ: Princeton University Press. Stolcke, V. 1995. Talking culture: New boundaries, new rhetorics of exclusion in Europe. Current Anthropology 36: 1–24. Stoller, P. 1989. The taste of ethnographic things: The senses in anthropology. Philadelphia: University of Pennsylvania Press. Stoller, P. 1997. Sensuous scholarship. Philadelphia: University of Pennsylvania Press. Strathern, M., ed. 2000. Audit cultures: Anthropological studies in accountability, ethics, and the academy. London: Routledge. Strecker, I., and S. Tylor, eds. 2009. Culture and rhetoric. New York: Berghahn. Struever, N. 1983. Fables of power. Representations 4: 108–27. Tagliacozzo, G., ed. 1983. Vico and Marx: Affinities and contrasts. Atlantic Highlands, NJ: Humanities Press. Tilly, C., ed. 1997. Roads from past to future. Lanham, MD: Rowman & Littlefield. Traweek, S. 1988. Beamtimes and lifetimes: The world of high energy physicists. Cambridge, MA: MIT Press. Turner, V. 1967. The forest of symbols: Aspects of Ndembu ritual. Ithaca, NY: Cornell University Press. Turner, V. 1974. Dramas, fields, and metaphors: Symbolic action in human society. Ithaca, NY: Cornell University Press. Vico, G. 1984 [1744]. The new science of Giambattista Vico, trans. T. G. Bergin and M. H. Fisch. Ithaca, NY: Cornell University Press. Vogt, E. Z. 1965. Structural and conceptual replication in Zinacantan culture. American Anthropologist 58: 264–81. Wallace, A. F. C. 1956. Revitalization movements: Some theoretical considerations. American Anthropologist 58: 264–81.

14_107-Bernard.indb 95

6/9/14 10:02 AM

96

James W. Fernandez and Michael Herzfeld

Watson, R. S., ed. 1994. Memory, history, and opposition under state socialism. Santa Fe: School of American Research Press. Waugh, L. R. 1976. Roman Jakobson’s science of language. The Hague: Peter de Ridder Press. Webster’s new international dictionary of the English language. 1959. 2nd ed. Unabridged. Springfield, MA: G. and C. Merriam Co. Willis, P. E. 1977. Learning to labour: How working class kids get working class jobs. Farnborough, UK: Saxon House. Zabusky, S. E. 1995. Launching Europe: An ethnography of European cooperation in space science. Princeton, NJ: Princeton University Press. Živković, M. 2011. Serbian dreambook: National imaginary in the time of Milošević. Bloomington: Indiana University Press.

14_107-Bernard.indb 96

6/9/14 10:02 AM

CHAPTER 3

Research Design and Research Strategies Jeffrey C. Johnson and Daniel J. Hruschka

We need a powerful mode of argumentation, a mode that ensures we can represent our representations in credible ways. In such worlds, a systematic argument enjoys a star-spangled legitimacy. We need a way to argue what we know based on the process by which we came to know it. That’s what I seek, not as the only possible representation that our field can offer, but as an essential lever to try and move the world. —Michael A. Agar (1996, 13)

In a complex world of competing arguments, who is to be believed or trusted? Are data themselves, independently of how they were conceived and collected, to be viewed as proper evidence for making a case? Although some may be swayed by the elegance of a well-written essay, for many it is crucial to know something about the author, his or her motivations, and the experiences, skills, and methods of investigation before passing judgment on the conclusions. In Agar’s statement above, we get the impression that a credible argument should be systematic and based on a process that informs us about how researchers came to know what they know. Research design is the careful planning and implementation of this process of knowing. A priori planning of all phases of research (including analysis and writing) can benefit one’s research in several ways. First, planning helps us deal with (and take advantage of) inevitable contingencies in the field. Research projects rarely, if ever, unfold seamlessly, and a plan provides a backbone for making (and justifying) tough decisions about site selection, participant sampling, and data collection and analysis in the face of inevitable bumps and detours. Second, clearly specifying the plan as well as its imperfect implementation lets other researchers check under the hood, identify potential biases, and judge for themselves whether the results are supported by the data and the methods. This transparency helps the reader determine how robust the findings are when research procedures change in presumably minor ways (reliability), how well the findings might extend to other people or groups (generalizability), and how well the findings really represent what we think they do (validity). Exposing these imperfections can be disconcerting, but it ensures that others can judge a researcher’s claims and can build on his or her research. Third, an idealized plan clarifies how methods of data collection and analysis are linked to specific concepts and processes in one’s theory or theories. Finally, on a practical level, good research design is also essential for research grants and contracts. There is much variation in what funding agencies and foundations 97

14_107-Bernard.indb 97

6/9/14 10:02 AM

98

Jeffrey C. Johnson and Daniel J. Hruschka

expect regarding research design. One agency may require a detailed description of the proposed project, paying attention to the research design logic of science (e.g., validity, reliability, hypotheses, etc.). Others may require a thorough exploration of the research problem and site but require less detail about the methods of data collection and analysis. All funding agencies expect a well-organized outline of the proposed project—one that meets the design expectations of peer reviewers and agency personnel. A distinction needs to be made between what is sometimes called the laundry-list component of research versus research design. The laundry-list component is important. It involves details about getting into and out of the field situation, making travel arrangements, getting proper government permissions, making contacts at the field site, arranging for living accommodations, and so on. Design, on the other hand, involves the methodological and analytical planning that contributes to the credibility, validity, believability, replicability, or plausibility of any study. In this chapter, we concentrate on elements of design important for generating valid results or a believable account. We describe the benefits of research design, outline how cultural anthropologists have traditionally approached study design in their research, and discuss how cultural anthropologists approach research design today. Then we describe key elements in research designs, focusing on choice of comparison groups and sampling, and present case studies of common designs in anthropology—including exploratory designs, comparing individuals within populations, case-control designs, two-site comparisons, large-scale cross-population comparisons, field experiments, and longitudinal studies. THE BENEFITS OF DESIGN

Evidence for the powerful benefits of research design is all around us. For example, the invention of the control/treatment design of clinical trials allowed researchers in the twentieth century to evaluate competing therapies and to select the ones that worked best. One result is that infectious childhood diseases that killed thousands of young people a century ago are today only a memory in industrialized countries. The lessons learned from controlled experimentation are also applied today to the policy arena, where groups are in conflict over resources or because of social inequalities (Johnson and Pollnac 1989; Porter 1995). Members of such competing groups—such as largescale commercial producers, commodity producers, environmental groups, and real estate developers—believe strongly in their positions. They have evidence, often anecdotal, that their positions are credible. Without some unbiased means for assessing the evidence, the truth is only a matter of who has the most political clout. The outcry for a ban on nets in tuna fishing is a famous example. Environmental organizations launched campaigns to ban nets in tuna fishing because dolphins are often caught incidentally in that fishery. Media campaigns in the United States showing pictures of dolphins being caught in nets (generally not in U.S. waters) contributed to Florida’s totally banning fishing nets—even though no marine mammals were threatened by the use of nets in Florida waters. Thus, in the absence of a research design that addresses a specific problem or goal, the policy emerges merely from interactions between groups of differing political, ideological, social, and economic backgrounds.

14_107-Bernard.indb 98

6/9/14 10:02 AM

3: Research Design and Research Strategies

99

There was a similar concern over the unintended but regular catch of harbor porpoises by net fishers in New England (Schneider 1996). This case led to a systematic test of a technology that might ameliorate the problem. Wildlife conservationists petitioned the U.S. federal government in 1991 to declare harbor porpoises a threatened species. In response, the fishing industry proposed the voluntary use of “pingers,” an underwater acoustic device to keep porpoises from their nets. The effectiveness of the device, however, was in question, and there was no firm evidence in the literature about it. Fishers petitioned the federal government to fund a study of pinger effectiveness. The study used the classic control/treatment design in which catch rates for a set of nets with pingers were compared to catch rates for set of nets without pingers. In the first experiment, the control net caught 10 porpoises while the treatment net caught none. Some conservationist groups claimed the study was biased in that the treatment nets were placed in areas known not to have large numbers of porpoises. So another study was conducted placing experimental treatment and control nets in the same proximity. This time, the treatment net caught only 1 porpoise while the control net caught 32. Some environmental groups were still concerned that evidence with more statistical power was needed. Lobbying efforts by fishers yielded more funds for a larger, more comprehensive study involving more than 10,000 fishing nets. Both control and treatment nets were outfitted with pingers, but only the pingers on treatment nets would activate once placed in the water. Thus, fishers were blind as to which nets were control nets and which were treatment—a classic double-blind experimental design. Again the evidence was impressive: The treatment nets caught 2 porpoises (1 was thought to be deaf), while the control nets caught 25. The issue is still under debate, but this series of studies illustrates how the elements of research design—and subsequent revisions to the research design—help muster evidence in light of competing beliefs and philosophies. In each successive study, investigators tried to control for as many extraneous variables as possible so that the hypothesized effect could be assessed (i.e., the effectiveness of using pingers compared to not using pingers). The logic of the research design contributed to the production of credible results. Although the power of experimental design is evident, attention to its application in anthropology—particularly cultural anthropology—has been limited. Some early exceptions include Brim and Spain’s (1974) book on hypothesis-testing designs, Pelto and Pelto’s (1978) book on research methodology in cultural anthropology, Naroll and Cohen’s (1973) A Handbook of Method in Cultural Anthropology, which has several chapters that address issues in research design (LeVine 1973; Sechrest 1973; Spindler and Goldschmidt 1973), and the first edition of this handbook (Bernard 1998). Bernard (2011) has elaborated in more detail on issues of design, but his treatment is necessarily limited given his task of describing the range of methods available to anthropologists. BOAS, MALINOWSKI, AND RESEARCH DESIGN IN THE SCIENTIFIC TRADITION

Despite anthropologists’ long-held interest in attention to detail, the structure for gathering those details has been documented and explicated much less thoroughly than have the details themselves. Franz Boas and most of his students advocated a natural science logic in the collection of ethnographic materials and a true concern for the collection of reliable data that could lead to the production of valid theory.

14_107-Bernard.indb 99

6/9/14 10:02 AM

100

Jeffrey C. Johnson and Daniel J. Hruschka

Notwithstanding his concern for scientific method, Boas was more explicit about his methods of data analysis than about his methods of fieldwork and data collection (Boas 1920; Ellen 1984). Malinowski, another great contributor to anthropological inquiry, was also concerned with the aims of science and methodological rigor. However, his earliest contributions were more a demonstration of the value of ethnographic writing—his “unusual literary sense” (Lowie 1937, 231)—than of methodological details of ethnographic fieldwork (Ellen 1984). A good example of this tension between the stated early concerns for the methods of science and the actual use of such methods in ethnography comes from correspondence between Boas and his student Margaret Mead during her first fieldwork in Samoa. As Orans (1996) describes it, Mead wrote to Boas with her concerns about possible violations of scientific principles in the data she had collected to that point. She wrote of her doubts about the comparability of cases and about her ability, or even the need, to do a quantitative comparison of the similarity of attitudes among the adolescent girls in her study. She had concerns—and we believe she thought her mentor, Boas, would feel similarly—as to whether a valid comparison of this type could be made, given the selection process for her sample of girls. The constraints of field research may lead one to stray from the idealized prescriptions of a research design, but Mead was actually attempting to forgo the research procedures advocated by Boas and others. Orans says: “What she wants is permission to present data simply as ‘illustrative material’ for the representativeness of which one will simply have to take her word” (p. 127). What is most surprising is Boas’s response to Mead. He writes: I am very decidedly of the opinion that a statistical treatment of such intricate behavior as the one that you are studying will not have very much meaning and that the characterization of a selected number of cases must necessarily be the material with which you operate. Statistical work will require the tearing out of its natural setting, some particular aspects of behavior which, without that setting may have no meaning whatever. A complete elimination of the subjective use of the investigator is of course quite impossible in a matter of this kind but undoubtedly you will try to overcome this so far as that is all possible. (Orans 1996, 128)

This response is important for at least two reasons. First, it demonstrates the differences between the stated scientific objectives of ethnographic work, as advocated by Boas, and the actual practice of ethnographic research. There appears to be a perception that a systematic treatment of the data will have to be abandoned to preserve context and meaning. Ironically, this concern for context and meaning over methodological rigor, particularly for those in search of theoretical foundations (i.e., the Boasian idea of data leading to the construction of theory), would ultimately hinder the comparability of data from different ethnographic sources (see Moran [1995] for a discussion of this issue, and see Ember et al., this volume). Second, Boas’s concern for contextual meaning over the statistical analysis of data was prophetic. The concern for understanding human behavior in context is one of anthropology’s strengths. However, thinking of quantification as incompatible with an attention to context and meaning has often clouded discussions about research design.

14_107-Bernard.indb 100

6/9/14 10:02 AM

3: Research Design and Research Strategies

101

Boas’s final sentence in his response to Mead illustrates that even at this early stage the issue of subjectivity of ethnographic research was of concern. There was faith, though, that awareness of the potential biases associated with the investigator’s subjectivity meant that it could be dealt with in some reasonable way. A further irony is that the one thing that might have lessened potential subjectivity biases—the use of standardized methods—was rejected outright because there was a concern (perhaps incorrect) that meaning might be compromised. Mead’s position on these various elements of research design provided fuel for the continuing discussions about how successful her original findings were in addressing the questions she posed, how valid her interpretations were, and whether the research design was adequately rigorous (Brim and Spain 1974; Freeman 1983; Orans 1996). Thus, while early British and U.S. anthropologists advocated the scientific method in ethnographic research, there is little evidence that they considered appropriate research design issues when they actually did the research. As Urry (1984) sees it: In Britain the claims that anthropology not only studied a distinctive body of data but also that it possessed a sophisticated methodology to collect these data, was an important factor in the establishment of anthropology as a discipline. This was less necessary in America where, by the late nineteenth century, anthropology was already established in universities, museums and government agencies. But in spite of claims to scientific methodology, particularly in the British tradition, there are surprisingly few details about actual methods anthropologists used in the field, beyond a few first principles and illustrative anecdotes. There was a wide belief among British anthropologists that fieldwork could not be taught to new recruits, but could only be experienced by individuals in the field. In the American tradition texts provided what was regarded as an objective body of data, whereas the British tradition was more a matter of subjective experience. It is a strange paradox in the development of field methods that the scientific study of other cultures has been built upon such a foundation. (p. 61)

There is much anecdotal evidence for a staunch belief, across the British and U.S. traditions, in the trial-by-fire method of training ethnographers. This belief likely supports the current lack of formal training in methods and research design in cultural anthropology. Agar (1980) and Bernard (2011) relate stories about Kroeber’s recommendations regarding how ethnographic research is taught and conducted. One story concerns Charles Wagley’s teaching of a field methods course while the other concerns a graduate student at Berkeley asking Kroeber for methodological advice before going to the field. According to Agar, student folklore has it that Kroeber stated tersely to the nervous student “I suggest you buy a notebook and a pencil” (Agar 1980, 2). Even in the late 1960s, when concern for methodological rigor was probably at its peak in anthropology, many treatments of research methods and design in the literature played down the need for more systematic methods and design detail, including the operationalization of concepts, the reliability and generalizability of findings, and the assessment of competing claims through formal model comparison and hypothesis testing (LeVine 1973). A good example of this is a book by Thomas Rhys Williams (1967) published in the Spindlers’ series on field methods. Williams writes:

14_107-Bernard.indb 101

6/9/14 10:02 AM

102

Jeffrey C. Johnson and Daniel J. Hruschka

I believe that only someone wholly involved and fully immersed in fieldwork can really communicate the essence of cultural anthropology to students or general readers. And since I have indicated here that research in culture involves a great deal of unique personal experience for the anthropologist, I have taken the position that it is probably unlikely there can be a rigorous, systematic, and formal presentation of methods in the study of culture like those of the natural sciences and that there are overriding concerns among many sociologists, psychologists, and economists. I find this stance comfortable, for it is my conviction that so long as prime theoretical concerns in the study of culture are an attempt to record and understand the native’s view of his culture and the objective and historical realities of culture, then methods for field study will have to reflect the end purpose of making a whole account of a part of the human experience. (pp. 64–65)

Early on, LeVine (1973) and others (Johnson 1990) made the point that the nature of fieldwork, in terms of its requisite huge investments in time and geographical focus, has often limited the attractiveness of more formal research designs because of its commitment to studying specific problems in a specific way. The realities of fieldwork often dictate the need to change the problem focus or, finding that the proposed hypotheses are inappropriate to the cultural setting under study, the need to somehow salvage the research with a description of what is really going on there. Laboratory and survey researchers have some flexibility to change the problem focus and study populations in light of emerging problems, but fieldworkers are limited in their ability to do so. Thus, the idea of researchers “putting all their eggs in one basket” may have limited the a priori formulation of problems in fieldwork (LeVine 1973, 184). Further, the huge investment in time and resources limited another important goal of science, that of replication, since ethnographers resist being expected to replicate someone else’s work. The “my natives” or “my village” mentality of some and the fact that careers are made by discovering new theories or describing exotic less well-known cultures has certainly inhibited replication efforts (Johnson 1990). CONTEMPORARY RESEARCH STRATEGIES IN CULTURAL ANTHROPOLOGY

As cultural anthropologists plan and conduct research, there are several distinct goals that they can pursue, a variety of criteria they may choose for judging good work, and numerous research strategies for fulfilling these specific criteria. Among possible goals, researchers can aim to accurately describe a social situation, vividly convey another person’s perspective, teach a lesson, or develop and test general explanations for behavior. Debates sometimes erupt about the primacy of these goals, but they need not be at odds. Cultural anthropologists also draw from a diverse set of criteria for evaluating good work. Here is a list of some of these criteria: 1. 2. 3. 4. 5. 6. 7.

14_107-Bernard.indb 102

Does the account vividly convey the situation? Is the description sufficiently rich? Is the narrative compelling? Do the local actors agree with the findings or interpretation? Is the description accurate by some objective criteria? Is the description consistent with all data? Are rival explanations or interpretations considered?

6/9/14 10:02 AM

3: Research Design and Research Strategies 8. 9. 10. 11. 12.

103

Do the methods appropriately capture the intended concepts and variables? Are the methods reliable? Can the findings be replicated? Are findings generalizable? Do the findings advance or challenge established theory?

To create work that meets these criteria, or at least appropriate subsets of these criteria, researchers have drawn from a number of research strategies. These include extended fieldwork in a setting, crafting morally compelling narratives, triangulating observations and data to minimize bias, using systematic sampling of populations, and planning study designs that can discriminate between different explanations for human behavior. Although we may be tempted to classify researchers or traditions as preferentially valuing specific criteria, it is an empirical question whether such clean divisions exist. In fact, anthropology may be unique among disciplines in its tolerance for such a diverse set of criteria. That said, solid research design is especially crucial for satisfying the criteria of accurate description (#5), consistency with data (#6), discrimination between alternative explanations (#7), reliability (#8), validity (#9), replicability (#10), and generalizability (#11). These are also some of the key criteria used for judging scientific work more generally. Research design can involve both qualitative and quantitative data, objective and subjective measurements, and biological and cultural components in the same project. Research design is also an important element of both exploratory/inductive and evaluative/deductive phases of scientific inquiry (see the section on Research Design in Scientific Inquiry, below). Regardless of whether you are pursuing a more exploratory or a more explanatory agenda, research design requires advanced planning about how you will generate reliable, valid, and generalizable findings or how the study can discriminate between alternative explanations for the phenomenon being studied. This involves meticulous attention to questions about sampling (to accurately represent a population and to improve generalizability), about measurement and triangulation (to ensure reliability and validity), about analyzing the data and reporting results, and about the kinds of data required to rule out or rule in specific hypotheses. A key principle in designing the process is to ward off as many threats to validity as possible by using appropriate methodological and analytical checks. The vicissitudes of real life in a field site nearly always requires some adjustments will be made to such advanced planning. Nonetheless, decisions about these adjustments still rely on the same concerns about minimizing threats to validity within the constraints of the field site, and being skilled, or at least experienced, in research design will be important in making those adjustments. Similar to approaches critical of science, the design process is based on extreme skepticism about the researcher’s ability to minimize bias. For example, the research design process is highly skeptical of meeting the criteria stated above by relying exclusively on a single ethnographer and his or her memory or field notes as the single instrument of measurement. Research design aims to develop and implement measures that minimize bias. Many ethnographic studies have an exploratory and descriptive aim. But within this large genre, authors vary greatly in the research strategies they follow. For example,

14_107-Bernard.indb 103

6/9/14 10:02 AM

104

Jeffrey C. Johnson and Daniel J. Hruschka

Zabusky’s (1995) ethnographic study of cooperation in European space science takes “the form of mutual exploration rather than unidirectional examination” (p. 46). She contrasts her study with research on cooperation conducted by experimental psychologists, emphasizing the cultural and social orientation of her work and the importance of considering context (social, cultural, political, etc.) in her analysis. Following in the “thick description” tradition of Clifford Geertz, Zabusky clearly believes in some kind of ethnographic authority. In a short methodology section, she discusses the challenge of conducting participant observation research in this rather complex, geographically dispersed, cross-cultural setting. She also discusses the rationales for selecting the site and the group she studied, problems of working in a linguistically and technically diverse social milieu, the use of semi-structured and unstructured interviews, and the effect of her role as ethnographer on informant relations and data quality. Although Zabusky doesn’t talk specifically about design or about concerns for potential threats to validity, there is implicit concern for such issues throughout her ethnography. In contrast to Zabusky, there is a body of work in anthropology that is more extreme in its rejection of systematic design issues in favor of vivid accounts (#1), rich description (#2), and compelling narratives (#3) as well as novelty in representing the human subjects (including the ethnographer). Ramos (1995), for example, published an ethnography based on a rewrite of her 1972 dissertation, with additional ethnographic insights. She rejects the “anthropological austerity” of her original work in favor of an “intersubjective understanding” that captures the “flavor” of her ethnographic encounter with the Yanomami. To her, the original work was “oldfashioned and theoretically unsophisticated” and had to be replaced by a more reflexive work. This contrast between the old and the new reflects the increased variation in epistemological emphasis in the field that has developed over the last 50 years. As Ramos sees it, “I found myself making forays into the self-conscious meanderings of reflexive anthropology in order to shift the axis of analysis from the skeletonlike dissertation to the flesh and blood of ethnography” (p. 6). Along with this shift came the freedom to be less concerned by the issues of bias and validity or the need for working systematically, thus allowing for a less restrictive ethnographic narrative. Although Ramos discusses informant interviewing and various sources of data, her introduction is largely devoted to discussions of her reliance on her own memory in writing the ethnography and the shift in the narrative between synchrony and diachrony. Thus, there is little discussion of research design and methods of data collection as might be found in work in the systematic tradition. Instead, Ramos emphasizes the emergent and reflexive nature of data and the literary strategies used in producing the ethnographic product. Other examples in this vein include Panourgia’s (1995) use of “we” and “they” in her “Athenian Anthropography” and Behar’s (1993) use of montage in her collaboration with a single woman in the telling of that woman’s life story. Another tradition in anthropology that gained prominence in the last 30 years focuses on human behavior, speech, and culture as texts to be analyzed by principles similar to those in literary analysis. A parallel tradition focused on problems of representation, delving into experimental writing strategies that include such approaches as montages, evocative representations, polyvocal texts, and even ethnographic fic-

14_107-Bernard.indb 104

6/9/14 10:02 AM

3: Research Design and Research Strategies

105

tions (Denzin and Lincoln 1994). While systematic analytical paradigms are primarily concerned with threats to validity, recent interpretive paradigms are focused more on threats to believability—as in “Do you believe my story?” (Tyler 1991, 85)—or, in critical theory, threats to trustworthiness (Kincheloe and McLaren 1994). A different concern is threats to compellingness—whether the story is worth telling or reading (Tsing 2005). The methodological focus is on engagement and representation. How does the researcher obtain suitable immersion into the cultural context of the actor(s) to represent it? How does one write a text that conveys the understanding gained from such an immersion in a believable and compelling way? A great deal of innovation in descriptive, ethnographic research has involved moving away from place-based inquiries in traditional or small-scale societies toward new populations and social situations and larger processes of globalization. For example, multi-sited ethnography is a research strategy that follows a topic, social problem, or object through field sites that may be geographically or socially distant from one another (Marcus 1995). Nancy Scheper-Hughes (2004) used this approach to study the global traffic in human body parts by “following the bodies” to the diverse places where human organs and tissue are extracted, brokered, transported, and transplanted (p. 32). A key goal of this research was to make public (p. 37) the harvesting, selling, and distribution of human body parts; the research shows the value of ethnography in revealing covert social worlds. Scheper-Hughes observed practices and social interactions in these diverse sites, conducted open-ended interviews with key informants, and administered structured questionnaires with vulnerable populations. To expose these practices concerning body parts, she identified key players in the trade—surgeons, kidney sellers and buyers, kidney hunters, and kidney brokers—who would grant interviews. To document various views of the organ trade as well as the participation of vulnerable populations in the organ trade, Scheper-Hughes conducted surveys of squatter settlements in South Africa, five villages in Moldova, and residents of a large slum in the Philippines. Each of these pieces of data provides material for describing this particular social world. Another ethnography in the global framework is one by Anna Tsing (2005). Whereas Scheper-Hughes focuses on key intermediaries in global connections (sellers, buyers, brokers), Tsing examines connections and interactions between local and global actors to tell the story of rainforest exploitation in Indonesia in the 1990s. To do this, she focuses on the awkward interactions of key actors and agencies. Tsing uses observation, interviews, and journalistic and archival data as raw material for these stories. She devotes considerable care to crafting these stories—to make the reader “feel the rawness of the frontier” (p. 28) and to ensure the story is compelling (p. 25). A key strategy in her ethnography is to point out observations and anecdotes that don’t fit (i.e., that “interrupt”) dominant narratives about capitalism and globalization (p. 270). The observations on which these stories are based presumably derive from Tsing’s personal experience. However, the work makes no reference to checks on reliability and validity threats—for example, a possible bias toward telling stories in a way that accentuates divergences from dominant narratives to support the overarching thesis about cultural friction. More recently, anthropologists have expanded their ethnographic inquiries into subject matter and populations that fall well outside of anthropological traditions.

14_107-Bernard.indb 105

6/9/14 10:02 AM

106

Jeffrey C. Johnson and Daniel J. Hruschka

Most notable among these is Tom Boellstorff’s (2008) study of virtual worlds in his ethnography of Second Life. He advocates the development of a virtual anthropology, as opposed to just simply virtual ethnography, that would bring a unique anthropological perspective to the study of online and virtual worlds. In this ethnographic study of Second Life, Boellstorff wants to understand social interaction through participant observation. As one might imagine, participant observation in this context poses its own set of unique challenges and opportunities. Boellstorff carefully frames the bounds of inquiry in terms of interaction among participants in the virtual space, with little concern for the nature of their lives and interactions in their non-virtual worlds. He creates a character and participates in interactions with others engaging in conversations and teleporting to virtual locations (e.g., a friend’s house). His data are derived from casual conversations, observations, formal interviews, and even virtual focus groups, being careful to inform all encountered that he is an anthropologist doing a study of Second Life. Boellstorff’s conception of participant observation here involves the discovery of “culture through nonelicited, everyday interactions” (p. 72). This distinction is critical to his methodological approach in that, in his mind, the use of elicitation techniques for interviewing (as might be found in cognitive anthropology) only helps in uncovering the cultural rules for interaction, which he sees as an impoverished model of culture, rather than a deeper understanding of “culture in virtual worlds” (p. 66). In emphasizing this epistemological and methodological tension, he contrasts what he calls epistome (science or knowledge) from techne (craftsmanship), reflecting an attempt to understand the complex whole rather than just simple knowledge and beliefs. Karen Ho’s (2009) book on financial firms on Wall Street is another example of an ethnographic study in a nontraditional setting. Anthropologists have been notoriously absent in the study of the corporate world, particularly finance. Having worked on Wall Street in several jobs, Ho used her extensive network of friends and acquaintances to make contacts and conduct interviews—in a sense, “snowballing” her way through the financial world of Wall Street. As in some of the earlier ethnographies focusing on the complexity of global connections and processes, her ethnography sought to look simultaneously at both the globalization of capital markets and the strategies of financial actors. Using participant observation as more of a true participant rather than a simple observer (Johnson et al. 2006), Ho engaged in a study of the powerful or engaged in “studying-up.” Participant observation was important in that it provided a referent to better situate the talk of powerful informants. Ironically, the usual means for dealing with the imbalance of power between ethnographer and those studied, giving voice to the native, could be problematic in this case in that giving her informants voice could over-privilege the powerful. These examples offer only a glimpse of the range of possible research strategies that can be used to do research and write ethnographies. In some cases, producing a picture of human life of a population or for a phenomenon of interest is an exploratory enterprise with an implicit concern for methodological issues. In other cases, anthropological research is concerned more with the strategies and methods of ethnographic presentation and with the reflexive character of the ethnographic enterprise. Thus, traditional methods sections are replaced by discussions on how to read the work or on the particular methods used in

14_107-Bernard.indb 106

6/9/14 10:02 AM

3: Research Design and Research Strategies

107

writing the ethnography itself. In some cases, more authoritative ethnographic methods are combined with literary devices to develop a more compelling argument. This combination of more ethnographic authority and ethnographic crafting has been the more recent trend, as illustrated by the last few ethnographic examples. Again, these innovations in anthropology do not necessarily need to be at odds with systematic inquiry. We now focus primarily on research designs aimed at the scientific concerns laid out earlier, including accurate description (#5), consistency with data (#6), discrimination between alternative explanations (#7), reliability (#8), validity (#9), replicability (#10), and generalizability (#11). For further discussion of research strategies in the interpretive mode, see Fernandez and Herzfeld (this volume). RESEARCH DESIGN FOR SYSTEMATIC INQUIRY

In some social science disciplines, like psychology, the design of research is driven by the method of analysis: Analysis-of-variance models and multi-group comparisons (factorial designs) may dictate the whos, whats, and wheres of a given project. In sociology, multiple regression models, structural equation models, and path analytic models (all related analytical techniques) have influenced the design of survey research. Ethnography, referred to as the anthropological method by William Foote Whyte (1984), has influenced the nature of design in cultural anthropology, but in profoundly different ways. While the analytical techniques most often used in psychology, sociology, and economics often led to rather standard designs, in anthropology the eclectic nature of ethnography leaves the design of research more open ended. There are generally no ethnographic analytical techniques driving the design, although ethnography has been variously associated with a number of qualitative methods. There has not even been firm consensus on what ethnography really is (Johnson 1990; Van Maanen 1988), and consensus is even more fleeting today. The good news is that ethnographic research is amenable to a wide range of research designs, including the use of multiple designs within a single ethnographic context. This allows for flexibility, multiple tests of a theory, and increased chances for various types of validity, triangulation, and potential for high levels of innovation and creativity. This is particularly true today, given the large number of tools available for assisting the researcher in managing and analyzing text data (see Wutich et al., this volume). Currently, the qualitative analysis of text and discourse is no longer restricted to either interpretive or exploratory approaches but can also be used in hypothesis testing and explanatory research. Figure 3.1 reveals that the overall research process is more than just a matter of study design. There is no substitute for a good theory, and there is a critical need to link theory, design, data collection, analysis, and interpretation in a coherent fashion. Design, however, is the foundation of good research. No amount of sophisticated statistics, computer intensive text analysis, or elegant writing can salvage a poorly designed study. Hurlbert (1984) emphasizes this in a classic paper on the design of field experiments in ecology: “Statistical analysis and interpretation,” he says, “are the least critical aspects of experimentation, in that if purely statistical or interpretive errors are made, the data can be reanalyzed. On the other hand, the only complete remedy for design

14_107-Bernard.indb 107

6/9/14 10:02 AM

108

Jeffrey C. Johnson and Daniel J. Hruschka

Figure 3.1. Relationship between exploratory and explanatory approaches within the overall ethnographic research process.

or execution errors is repetition of the experiment” (p. 189). Redoing an experiment because of fundamental design errors is one matter; redoing a year-long ethnographic field study because of such errors is quite another. Figure 3.1 also shows that the research process involves a simultaneous concern for the development of empirical statements from theory (e.g., hypotheses), the operationalization of theoretical concepts (e.g., meaningful and reliable measures), design (e.g., groups to be studied), data collection (e.g., qualitative vs. quantitative), and data

14_107-Bernard.indb 108

6/9/14 10:02 AM

3: Research Design and Research Strategies

109

analysis (e.g., multiple regression and text analysis). Theoretical knowledge is derived either from earlier studies or from exploratory work. Which theoretical concepts are measured and which are bracketed, the levels at which theoretical concepts are measured (e.g., nominal or ordinal), the types of sampling strategies used, and the application of appropriate types of analysis must all be considered as a part of the design. For example, the particular structure of an empirical statement or hypothesis will partially determine the manner in which theoretical concepts are operationalized and eventually analyzed (Stinchcombe [1987] provides an excellent discussion of how empirical statements are derived from theory). Most importantly, Figure 3.1 illustrates the complementary relationship between exploratory and explanatory approaches in ethnographic (and more generally scientific) investigations. Theory must, at some point, be derived from observations in the world. Thus, exploratory, inductive inquiry is crucial for theory development. Darwin built his theory of natural selection on his systematic (and frequently qualitative) description of organisms around the world. The modern theory of prion diseases arose in part from systematic exploratory investigation of South Fore horticulturalists in Papua New Guinea, a group afflicted with what is now known to be a prion disease. In addition to developing new hypotheses and constructing theories, exploratory research may contribute to the production of reliable and valid measures, provide information essential for constructing comparison groups, facilitate construction of structured questions or questionnaires, or provide information necessary for producing a sound probability or non-probability sample. Today, there are many tools for conducting systematic exploratory research in cultural anthropology, including sampling strategies to increase the representativeness of one’s descriptions, behavioral observation and interviewing techniques for collecting valid and reliable information, and techniques for analyzing qualitative and quantitative data. Thus, research design is more than just methods of data collection and analysis. It involves constructing a logical plan that links all the elements of research together so as to produce the most valid possible description of a situation or assessment of a theoretical framework or parts of it, given a set of realistic constraints (e.g., cost, scope, geographical setting, etc.). The purpose of research design is to ward off as many threats to validity as possible. In the case of exploratory descriptive work, this means ensuring that one’s findings are reliable and valid. In the case of explanatory designs, this also means considering and assessing alternative explanations for one’s findings. Research design requires careful attention to detail and often an admission concerning the potential weakness of a given design. Outside the laboratory, a multitude of influences can threaten the validity of any conclusions. In natural settings, particularly fieldwork, there is no perfect design that can control for all possible extraneous effects at once. Recognition of these limitations doesn’t invalidate a study’s results. Rather, it creates an open forum that can contribute a lot to important theoretical and methodological debates. Without attention to good design and methodological detail, researchers leave themselves open to one of the worst criticisms of all—of being “not even wrong” (Orans 1996). In other words, a lack of design and methodological detail makes it next to impossible to fairly and adequately assess the validity of any study’s conclusions such that rightness or wrongness may not even be debatable.

14_107-Bernard.indb 109

6/9/14 10:02 AM

110

Jeffrey C. Johnson and Daniel J. Hruschka

Here we outline two key considerations when planning a study. The first is choosing appropriate comparison groups and interpreting differences. The second is how to sample groups and participants to make one’s findings suitably representative and generalizable.

Planning and Interpreting Comparisons Many questions of interest to cultural anthropologists are comparative. How and why do people around the world differ in how they think and act? What explains diversity in social, economic, and political systems? How does exposure to a specific economic, social, or political system, or one’s position within such systems influence thought, behavior, and well-being? How do these systems change over time, what causes these changes, and what consequences do these changes have for people on the ground? These are all questions that require comparison—across groups, over time, across individuals from different groups. Indeed, it is difficult to find research questions in anthropology that do not require some, perhaps implicit, comparison between groups. For this reason, selecting appropriate comparison groups is an important part of many research designs. Experiments are a relatively extreme example of making comparisons that can help us think about comparisons in general. In a true experiment, a researcher randomly assigns individuals to one of two (or more) treatments (e.g., bariatric surgery or not), and compares an outcome between these two groups (e.g., weight change). Ideally, the random assignment means that the only difference between the two randomly chosen groups is the different treatment they received. Thus, any difference in outcome between these two groups should have resulted from the treatment. For this reason, experiments can help rule out a number of alternative explanations for differences in outcomes between groups, including the effects of extraneous factors (i.e., unmeasured variables that might affect the dependent variable), the effects of selection (i.e., comparison groups differ because of the way they were selected and not due to the treatment), the effects of reactive measurement (i.e., the measurement procedure itself caused a change in the dependent variable), or interaction effects involving selection (i.e., when selection interacts with other factors to create erroneous findings). These and other sources of error are also potential rival explanations, and randomized experiments are best at eliminating these rival explanations. Although designs of this type are often impossible in anthropological fieldwork, the principles of experimentation are instructive and are a guide for understanding potential sources of error when comparing groups, even in a non-laboratory setting. We borrow terminology from Kleinbaum et al. (1982) in constructing a typology of research designs. Included are experiments, quasi-experiments, observational study designs, and what we refer to as natural experiments. Experiments involve the random allocation of subjects to different treatments or conditions and afford the most control over distorting effects from extraneous factors. A quasi-experiment compares groups exposed to different conditions or treatments but lacks random assignment of group members. Nonrandom assignment lays any comparison between groups open to validity threats and reduces our ability to make causal inferences. Observational studies involve neither random assignment of members to comparison groups nor the manipulation by the observer of independent variables.

14_107-Bernard.indb 110

6/9/14 10:02 AM

3: Research Design and Research Strategies

111

Table 3.1 describes examples from observational and quasi-experimental study designs discussed by Kleinbaum et al. (1982) and Cook and Campbell (1979). More details can be found in these and other sources (e.g., Robson 2002). Natural experiments are similar to quasi-experiments except that the manipulation of independent variables occurs naturally or is unplanned rather than artificial or directed. Thus, comparison groups may be chosen on the basis of different levels of exposure to some naturally occurring or human-induced phenomena (e.g., natural disaster, war, or the building of a dam). Cook and Campbell (1979) make a similar distinction but refer to these kinds of natural experiments as “passive-observational studies.” Anthropologists involved in development and evaluation research are most likely to use this design. True experiments are, of course, rare in anthropology (but see Harris et al. [1993] for an example of a true experiment in a field setting). Even in quasi-experiments, it Table 3.1. Examples of Basic Research Designs Relevant to Anthropologists Observational Designs Cohort Study

Cross-Sectional Study

Case-Control Study

Static-Group Comparison

Often referred to as a panel study, this is a longitudinal design where individuals are followed through time. May involve comparison groups subjected to different treatments or exposed to different conditions. Often referred to as a survey study, it generally involves a random sample of a target population. Stratified sampling is often used to ensure adequate sampling of comparison groups. Although study factors are not controlled directly, designs of this type allow for the statistical control of variables during analysis. For some study factor (like an outcome variable), it compares a group of cases in which members have some characteristic of interest with one or more groups in which the characteristic of interest is absent. It is assumed that both groups come from the same underlying population. Often, members of the groups are matched on one or more variables. A variant of the cross-sectional design in which a treatment group(s) (i.e., members exposed to some variable of interest) is compared with a comparison or control group whose members are not exposed to the variable of interest.

Quasi-Experimental Designs One group posttest only design

Posttest only nonequivalent groups design

Pretest/posttest nonequivalent groups design

Interrupted time series design

14_107-Bernard.indb 111

Pretest observations are made on a single group. The group receives a treatment of some type and posttest observations are made. Experimental and comparison or control group are determined without random allocation of group members. Experimental group receives treatment while the control group does not. Posttest observations are made and groups are compared. Experimental and comparison or control group is determined without random allocation of group members. Pretest observations are made on both groups. Experimental group gets the treatment while control group does not. Posttest observations are made and groups are compared. One experimental group in which a series of observations is made both prior to some treatment and after the treatment.

6/9/14 10:02 AM

112

Jeffrey C. Johnson and Daniel J. Hruschka

is often difficult to manipulate independent variables directly while still preserving aspects of context that anthropologists hold on to dearly. And, traditionally, the most common designs used by anthropologists have been observational. However, with careful attention to design and ethnographic context, quasi-experimental and natural experimental designs can be applied to anthropological field settings, particularly in evaluation research and development research. Johnson and Murray (1997), for example, used a quasi-experimental design to evaluate the use of fish aggregation devices (FADS) in small-scale fisheries development projects. Two fixed fishing structures (piers) were pretested for differences in catch rates. Then, the fish aggregrating structures—umbrella-like units suspended in the water column—were alternately placed at the piers and individual fishers were interviewed simultaneously during randomly selected times at both the treatment piers (with the devices) and the control piers (without the devices). Johnson and Murray compared and determined catch rates. Whether a study is observational, experimental, or quasi-experimental, most research designs in the explanatory mode are comparative. Experimental designs compare treatment and control groups. Longitudinal studies compare individuals and groups at different time points. Cross-cultural studies compare behavior across populations. In anthropological fieldwork, these designs and others can be used in tandem to test or explore components of a theory. For example, in their study of preschool children, Johnson et al. (1997) used a longitudinal cross-sequential design (also known as a panel study or cohort study), which involved periodic interviews and observations of a cohort of preschool children carried out over the course of the year. By doing this, they were able to make comparisons between children at one point in time and comparisons of the same child with him- or herself over time. The importance of comparative thinking in ethnographic work cannot be overemphasized. Discussing common-sense knowing in evaluation research, Campbell (1988) describes the challenge of assessing change without careful attention to study design and comparative cases: The anthropologists have never studied a school system before. They have been hired after (or just as) the experimental program has got under way, and are inevitably studying a mixture of the old and the new under conditions in which it is easy to make the mistake of attributing to the program results which would have been there anyway. It would help in this if the anthropologists were to spend half of their time studying another school that was similar, except for the new experimental program. This has apparently not been considered. It would also help if the anthropologists were to study the school for a year or two prior to the program evaluation. (This would be hard to schedule, but we might regard the current school ethnographies as prestudies for new innovations still to come.) All knowing is comparative, however phenomenally absolute it appears, and an anthropologist is usually in a very poor position for valid comparison, as their own student experience and their secondhand knowledge of schools involve such different perspectives as to be of little comparative use. (p. 372; emphasis added)

The purpose of experimental design is to ward off threats to validity, although this is not as straightforward as it sounds. There are several types of validity—face, construct, statistical conclusion, internal, external, and so on. In one way or another, various

14_107-Bernard.indb 112

6/9/14 10:02 AM

3: Research Design and Research Strategies

113

Table 3.2. Threats to Internal Validity in Quasi-Experimental Designs History Testing Instrumentation Regression

Mortality Maturation Selection Selection by Maturation Interaction Ambiguity about Causal Direction Diffusion of Treatment Compensatory Equalization of Treatments Compensatory Rivalry

Change due to unmeasured or unobserved factors (spuriousness) Change resulting from experience gained by subjects as a consequence of measurement Change resulting from varying the way study participants are tested When selection of participants is atypical or extreme on a given measure, subsequent measures will become less extreme and there will be regression toward the mean Changes due to participants dropping out of the study Change in study participants over time due to factors unrelated to expected effects Observed effects due to nonrandom assignment of members and nonequivalence of groups Predisposition of selected group members to grow apart When time-order and causal direction is ambiguous Change due to one group receiving all or a portion of treatment meant for another group Tendency toward giving all groups the same treatment Participants’ perceptions (e.g., threats) that affect performance not a part of the treatment

study designs, in combination with other considerations, such as the operationalization of theoretical constructs and sampling, are better or worse at dealing with each. Here, we stress the importance of thinking through how validity threats have influenced and will influence observations or data (for a more in-depth discussion of how these types of validity can impact study conclusions, see Cook and Campbell 1979). Potential errors and bias creep in at various steps in the research process. It’s your job to contain these errors. In research design, forewarned is forearmed. Tables 3.2 and 3.3 give examples of threats to internal and external validity as discussed in Cook and Campbell (1979) for quasi-experimental designs. Internal validity is concerned with the approximation to the truth within the research setting—requiring study variables to covary but without spurious or unintended causes. External validity is concerned with the approximation to the truth as expanded to other settings—that is, with the generalizability of research findings. The threats in Table 3.2 deal with extraneous factors that may account for the presence or absence of a hypothesized effect (thus, contrasting validity with invalidity). In the quasi-experimental case, this means changes between pre- and posttest, but this way of thinking can be expanded to include hypothesized effects dealing with differences, similarities, or associations whether diachronic or synchronic.

Table 3.3. Threats to External Validity in Quasi-Experimental Designs Selection Setting History

14_107-Bernard.indb 113

Problems with generalizing due to the selection process for study subjects (e.g., nonrepresentative) Problems with generalizing due to the nature of the study setting (e.g., setting atypical) Problems with generalizing to either the past or the future

6/9/14 10:02 AM

114

Jeffrey C. Johnson and Daniel J. Hruschka

Cook and Campbell (1979) detail how each of the quasi-experimental designs in Table 3.1 are better or worse at dealing with each of the threats to validity that are found in Tables 3.2 and 3.3. For example, the pretest/posttest nonequivalent groups design controls for some internal threats to validity but does not do well at controlling for changes due to how groups members were selected (selection maturation), changes due to how individuals were tested (instrumentation), changes due to the selection of individuals with extreme pretest measures leading to regression toward the mean (regression), and changes due to local events not a part of the study (history). Each of these threats may hamper a researcher’s ability to assess the contribution of a hypothesized effect to any changes observed. Similarly, threats to external validity, such as problems stemming from biased samples or research in atypical or unique settings, can hamper the generalizability of one’s findings. Kleinbaum et al. (1982) offer a similar discussion of the strengths and weaknesses of observational designs in terms of controlling for threats to both internal and external validity. Other sources of potential bias affecting internal validity include sampling error (i.e., chance), nonresponse, the use of imprecise measures, data recording errors, informant inaccuracies, and interviewer effects (see Bernard 2011; Pelto and Pelto 1978). Careful attention to sampling, whether probabilistic (Babbie 1990) or non-probabilistic (Johnson 1990; Guest, this volume), is essential. Measurement, operationalization of theoretical concepts, and type of analysis used are other important factors. How reliable are your measures in terms of precision, sensitivity, resolution, and consistency? Are they valid, particularly with respect to accuracy and specificity in that they actually measure what they are intended to measure? Attention and concern with the potential sources of error, whether stemming from how the study was designed, how the data were collected (e.g., face-to-face interviews or mail-out surveys), or how the data were analyzed (e.g., statistical conclusion validity) will help lead to the production of solid evidence. However, a particularly important source of errors is due to problems with proper sampling and informant selection.

Sampling With limited time and resources, a researcher can only interview or observe a select set of events, communities, or individuals. However, the researcher may also hope that the findings from this sample reflect what one would find in a much broader population. Steps taken to ensure that a sample is representative of that larger population are crucial for making such generalizations. Or a researcher may want to catalogue the maximum diversity of views in a population rather than getting a representative portrait. In this case, a researcher would aim to select individuals expected to be maximally diverse in their views. Depending on a researcher’s specific goals, there are many sampling strategies that help meet those goals. When generalization to a target population is the objective, you should strive to define a sampling universe or frame from which you will select individuals, events, or communities and then develop a system for selecting a sample that accurately reflects the broader population. This usually entails a random sample of some kind, but can also involve stratifying strategically on sex, age, or ethnicity to guarantee appropriate representation of different groups. There is a vast literature on sampling theory and

14_107-Bernard.indb 114

6/9/14 10:02 AM

3: Research Design and Research Strategies

115

random sampling procedures, including discussions of sample sizes (see, e.g., Bernard [2011, 2012] for a summary, and the classic text by Babbie [1990] for detailed discussion of sampling issues). Cook and Campbell (1979) discuss two sampling models for increasing external validity in quasi-experiments. These models don’t necessarily involve random selection and are consequently less powerful than are random samples. In one approach, the model of deliberate sampling for heterogeneity, target classes of units, whether classes or categories of persons, places, times, or events, are deliberately chosen to represent the range found in the population. Thus, testing for a treatment effect across a wide range of classes in the set of all possible classes (including both extremes and the modal class) in the population allows the researcher to say something about how the effect holds in a range of settings. While this might not be generalized to the population as a whole, it does inform the researcher if an effect holds across wide-ranging classes within the population. The logic behind this model can be extended beyond the quasi-experimental case to observational studies. Kempton et al. (1996), for example, used a static-group comparative design sampling across a range of groups that were hypothesized to vary with respect to their environmental values. Kempton et al. interviewed a range of informants from members of Earth First! (a radical environmentalist group) to dry cleaning shop owners (who depend on toxic chemicals for their business). For some populations, it may be impossible to develop a sampling frame from which to draw a sample. In these cases, there are a variety of solutions, including intercept sampling, snowball sampling, random walks, quota sampling, and purposive sampling. Each of these approaches has potential problems, and most do not allow for generalizations about a population since they involve elements of unknown error even if the method involves some form of random selection criteria (e.g., random selection of locations in which to intercept respondents). Non-probability sampling methods have come to be associated with qualitative approaches or for the selection of ethnographic informants, particularly key informants or consultants (Johnson 1990; Miles and Huberman 1994; Werner and Schoepfle 1987; Guest, this volume). In some cases, a researcher may not be interested in generalizing to a population but may just want to know whether two subgroups obtained from a snowball sample differ with respect to some variable of interest. In that case, much of the bias in the sample is a matter of the logic used in the original selection of sample seeds, and any statistical analysis of the data must take into account violations of assumptions for the particular statistical test to be employed (e.g., independence of observations or random sample from a population). Such matters are particularly germane for observational designs using various social network approaches (see Borgatti et al. 2013). How samples are chosen is an important element of any research design. If you are interested in generalizing to a given population, random sampling of some kind is essential. If generalization is not a primary goal, then sampling requirements may be relaxed. In most cases, if you can use a random sample, do it! No matter what the sampling method, you should be explicit about how you chose the sampling units. This increases the chances of detecting potential bias and also makes replication feasible. Replication

14_107-Bernard.indb 115

6/9/14 10:02 AM

116

Jeffrey C. Johnson and Daniel J. Hruschka

is extremely important to external and other types of validity, such as construct validity. Random sampling has been a primary requirement in the proper application of parametric statistics. If you don’t use random sampling, pay careful consideration to possible violations of assumptions for a given statistical test. Recent developments in randomization and computer-intensive methods of statistical analysis involve less restrictive assumptions concerning the data (e.g., assumption of a random sample from a population or skewed, sparse, or small sample sizes), opening the way for the development of new test statistics particularly suited for the problem at hand (Edgington and Onghena 2007; Johnson and Murray 1997; Noreen 1989). These new approaches seem particularly well suited for the imperfect world of ethnographic research, where the rather restrictive assumptions of parametric analysis are often difficult to meet. But it is critical to remember the connection between theory, design (including sampling), and data analysis from the beginning, because how the data were collected, both in terms of measurement and sampling, is directly related to how they can be analyzed. RESEARCH DESIGN IN ANTHROPOLOGICAL PRACTICE: SYSTEMATIC RESEARCH STRATEGIES

The following examples illustrate some of the issues discussed so far. These examples show how the interplay of exploratory and explanatory approaches is crucial for the development and testing of theory and can also aid in guarding against threats to validity (Robson 2002). This is particularly evident in what has more recently been the emphasis on the use of mixed methods in anthropological research, and in the social sciences more generally (see, e.g., Journal of Mixed Methods Research). Here we describe seven different kinds of study design used in cultural anthropology today—exploratory designs, comparisons of individuals within societies, two-community comparisons, case control designs, large-scale cross-population comparisons, field experiments, and longitudinal designs.

Using Exploratory Research to Identify Locally Relevant Categories As seen in Figure 3.1, exploratory research and descriptive research are often essential components of an overall explanatory research design. In a series of papers, Koester (1996) and his colleagues (Koester et al. 1996) offer excellent examples of the role of participant observation in more clearly defining the set of HIV risk behaviors surrounding injection drug use. In most of the prior research on injection drug users (IDUs) and HIV risk, the primary risk factor was viewed in terms of direct needle sharing. Thus, most large epidemiological studies of IDUs focused mainly on direct behaviors in attempts to understand seroconversion rates and other risk factors. Based on participant observation among IDUs, Koester (1996) identified nine other behaviors that were outside the realm of the direct sharing of a single syringe by two or more IDUs. Termed “indirect sharing,” these nine behaviors can promote the transmission of HIV among IDUs, who, although not sharing needles directly, often share water for mixing of drugs or for rinsing syringes, share drug-mixing containers (cookers and spoons), share cottons for filtering, and share the actual drug solution itself. These find-

14_107-Bernard.indb 116

6/9/14 10:02 AM

3: Research Design and Research Strategies

117

Figure 3.2. Overall design framework for the Koester study.

ings are undeniably important for larger epidemiological work that examines elements of IDUs’ behaviors and such things as producing valid models of seroconversion. In a subsequent study, Koester et al. (1996) used these additional distinctions in sharing to look at the prevalence of injection-related HIV risk behaviors among several subpopulations of injection drug users (see Figure 3.2). A major component of the study was the comparison of IDUs who engaged in both direct sharing and indirect sharing with IDUs who engaged in indirect sharing only and those who neither shared directly nor indirectly. Statistical tests of group differences provided a greater understanding of the risk factors associated with the different types of behavior. This is a good example of the application of exploratory research in the production of better measures of potentially important explanatory variables.

Comparing Individuals in a Society: Experts, Novices, and Ethnobiology Ethnobiologists have long debated whether folk biological classifiers are natural historians who compare animals on the basis of their morphological characteristics or pragmatists who compare on the basis of the utility of organisms. Boster and Johnson (1989) explored this issue in an ethnobiological study of fish. Were individual informants classifying organisms on the basis of form or function? Boster and Johnson used a static group comparison design to compare several groups of expert fishermen with a group of novice fishermen. This is analogous to treatment and control groups without the random assignment of subjects to experimental units and where the treatment is implied rather than researcher directed (i.e., natural differences in experience with fish). In the comparison, both culture and language were held constant while experience with fish was varied. Four groups—from North Carolina, East Florida, West Florida, and Texas—were sampled to examine the effects of different kinds of experience since there are regional variations in species abundance. To ensure that experts were, in fact, experienced recreational fishermen, the rosters of sport fishing clubs in each region were sampled at random. The selection of control group subjects, by contrast, involved a purposeful selection procedure in which potential subjects were screened for recreational fishing experience. Using a questionnaire to gain background information, 15 college undergraduates who had the least amount of recreational fishing experience were selected from two introductory anthropology classes. These students were the control group. Each of the four expert groups comprised 15 subjects chosen at random from a larger sample of recreational fishermen.

14_107-Bernard.indb 117

6/9/14 10:02 AM

118

Jeffrey C. Johnson and Daniel J. Hruschka

Figure 3.3. Overall design framework for the Boster and Johnson study.

Thus, the groups to be compared consisted of five groups of 15 subjects, four comprised of experts and one of novices. All the groups were shown cards with artists’ renderings and the common names of 43 marine species commonly found from North Carolina to Texas. Individuals were asked to perform an unconstrained judged similarity of the fish—a free pile sort (see Weller, this volume, and Weller and Romney 1988). Further, beliefs about the use and functional characteristics of the fish obtained from extensive ethnographic interviews were turned into a sentence-frame completion task described by Weller and Romney (1988). Finally, a measure of morphological similarities was determined, using taxonomic distances between pairs of fish. Boster and Johnson (1989) used statistical and graphical methods to evaluate whether experts’ and novices’ judgments of fish, at the aggregate and individual levels, were closer to the morphological characteristics of fish or the uses of fish. Using statistical and descriptive inference, Boster and Johnson concluded that whether informants use form or function for classification depends on the knowledge base of the informants and the methods used to test their knowledge (see Figure 3.3). Lack of random assignment of subjects to treatment and control groups and pretest observations limit the ability to make causal inferences in this case. But the in-depth ethnographic background research, the particular structure of the hypothesis, and the overwhelming reliability of informant responses make for more confidence in the possible validity of the study’s conclusions.

Case Control Design: Folk Illness and Disease Risks In a recent paper, Baer et al. (2012) used a case-control design to study the relationship between the ethnomedical diagnosis of a folk illness and disease risks. The primary objective of the research was to understand the relationship between the ethnomedical diagnosis of the folk illnesses susto (fright) and/or nervios (nerves) and the risk for developing type 2 diabetes. Understanding the relationship could be useful in health-care provider screening for diabetes. The authors selected Guadalajara, Mexico’s second-largest city, as the study site, based on their earlier work that showed a widespread belief in susto and nervios in this region. In addition, type 2 diabetes is becoming an increasing problem in Mexico more generally. The research questions emerged from the earlier work, some of it descriptive and exploratory, stemming from interviews eliciting the causes, symptoms, and treatments for susto. In these interviews, respondents kept talking about a perceived

14_107-Bernard.indb 118

6/9/14 10:02 AM

3: Research Design and Research Strategies

119

Figure 3.4. Overall design framework for the Baer et al. study.

link between susto and diabetes (see Figure 3.4). Further, previous studies using onegroup-only designs (looking at only diabetics) had suggested a relationship between folk illness and diabetes. Respondents were drawn from a family practice clinic serving approximately 110,000 patients with mostly working-class backgrounds. The first sample group involved patients who went to the clinic, had been diagnosed with type 2 diabetes following their thirtieth birthday, and had been diagnosed more than one year ago. Patients with type 1 diabetes were excluded from the study. The comparison group consisted of patients at the clinic who were 30 years old or more who had not been diagnosed with type 2 diabetes at the time of the study. Both groups were asked questions about their perceptions of the link between susto/nervios and diabetes and whether they have ever suffered from susto/nervios. The type 2 diabetes group was also asked whether they believed susto/nervios was the cause of their diabetes. Finally, blood glucose levels were measured for all study participants. The authors recruited 836 patients, 811 of whom were interviewed. One potential problem was the possibility that the folk illness might have occurred following the development of diabetes rather than before its onset. To minimize this problem, the study limited the period for type 2 diabetes diagnosis to less than five years before the interview. This reduced the diabetic comparison group from 811 to 239 patients. As in the previous susto example, to help control for any confounding factors such as educational background and social class, members of the non-diabetic group, who were drawn from the patient population who went to the clinic for general health problems such as colds, colitis, bronchitis, and so on, were matched to the diabetic group on as many sociodemographic factors as possible, although there were some differences (e.g., gender, marital status). In comparisons of the diabetic and non-diabetic groups, there were no significant differences in the prevalence of susto; nor was the prevalence of susto significantly higher for those with undiagnosed diabetes. The same was true for nervios. The study design, using comparison groups and matched controls, helped cast doubt on the hypothesis that susto was an underlying cause of type 2 diabetes. To find that out, we would need either a true experiment or longitudinal design such as a cohort or panel study to determine if the risk factor preceded the outcome. However, the importance of this research is that the study design allowed for a fairer test of the hypothesis concerning susto as a risk factor in the development of diabetes than that afforded by a simple one-group-only design.

14_107-Bernard.indb 119

6/9/14 10:02 AM

120

Jeffrey C. Johnson and Daniel J. Hruschka

Control and Treatment in a Two-Community Comparative Design One of the central concerns of medical anthropologists has been to better understand the relationship between health-related behaviors and native perceptions about illness. Young and Garro’s (1982) now classic investigation of treatment choice in two Mexican communities is an example of a static-group comparison in which the presence or absence of the treatment is based on selection criteria not directly under the control of the researchers. One of the primary purposes of the research design was the consideration of competing hypotheses—the hallmark of good research design—and the testing of the primary hypothesis as an example of descriptive inference, as opposed to statistical inference. Descriptive inference is an approach highly suited for much anthropological research. An important issue in this area of research concerns the factors influencing the use of Western treatments among non-Western populations. One explanation views use tied to congruence between a client’s medical beliefs and scientific medical theory: The higher the congruence, the more likely the client will choose a physician’s treatment. Termed the “conceptual-incompatibility” hypothesis, a number of studies have suggested that such a congruence was the primary determinant of treatment choice among Third World peoples. Young and Garro took a different stance, stressing physician accessibility as the most important determinant of physician use. An important element of this position is that traditional medical beliefs are not a barrier to choice of physician treatment. The research design included the comparison of two Mexican communities that were similar in terms of cultural traditions and economies but varied in terms of access to Western medical services. The town of Pichataro had restricted access (a 20-minute bus ride from Uricho), while the town of Uricho had easy access. From a random sample of approximately 10% of the households in each of the towns, Young and Garro collected data on the number of illnesses that had occurred during the previous two months and the treatment each had received. Later, the researchers collected triad data and what they call term-frame data on informants’ perceived similarity of illnesses. Young and Garro (1982) tested the two main hypotheses in sequence. They had to establish differences in treatment choice behavior in the two communities before they could assess any hypotheses concerning differences in beliefs. Using a standard chisquare test, the authors found a significant difference in the frequency distribution of treatment alternatives between the two towns, with the exception of folk curers. Thus, the two communities seemed to differ in their use of Western medical services. This established, Young and Garro could then test the second hypothesis relating to the similarity in beliefs between the two communities. Ironically, in statistical terms, the authors have more interest in the null hypothesis of no difference in beliefs than in the alternative hypothesis of a difference in beliefs between groups. Using multidimensional scaling, Young and Garro (1982) compared the belief data and found striking similarities in the medical beliefs of communities. They conclude: On the basis of the data from the triads study and the term-frame interviews, we see little reason to reject the “null hypothesis” of no significant differences between the responses of the two groups of informants. This leads us to the conclusion that the substantial variation apparent in the use of a physician’s treatment between the two samples, a con-

14_107-Bernard.indb 120

6/9/14 10:02 AM

3: Research Design and Research Strategies

121

Figure 3.5. Overall design framework for the Young and Garro study.

sequence of differential access to such treatment, occurs without corresponding degrees of variation in resident’s attitudes and beliefs about illness. (p. 1462)

The authors’ careful attention to research design and analytical issues contributed to the production of impressive evidence that casts doubt on the validity of the conceptualincompatibility hypothesis. Note that the analysis used to test the hypothesis concerning similarities in beliefs involved descriptive inference, not statistical inference. Despite the authors’ claims of finding no significant difference, there was no real way, at least when the study was originally conducted, to assess the extent to which any differences were significant in the sense of statistical probability. Recent developments in statistical procedures allow us to assess the similarities in aggregated judged-similarity matrices between the two communities (see Handwerker and Borgatti, this volume, and Hubert 1987). In Young and Garro’s case, a visual inspection of the graphical representations of the data could lead to no other conclusion than that there was little or no difference in beliefs between the two communities (see Figure 3.5). This distinction is important, particularly with regard to anthropological research, in that hypothesis-testing research can be done without narrowly restricting it to analytical methods using statistical inference. There are, of course, threats to validity in this study. Because respondents weren’t randomly assigned into comparison groups, it’s difficult to know the influences of confounding variables on physician utilization and beliefs about illness. It is unrealistic to suppose that Young and Garro could have randomly assigned community members to the different comparison groups to control for confounding variables and then subjected their informants to the treatments of interest. That said, in a latter section, we review how Paul Farmer attempted to mitigate such threats to validity through a field experiment in Haiti asking similar questions. Given a lack of pretest observations, we can only assume that beliefs were similar prior to the availability of physicians in Uricho. In lieu of equalization through randomization, Young and Garro (1982), through extensive ethnographic background research, produced groups that, although nonequivalent in the quasi-experimental sense, shared similarities with regard to a number of important characteristics. This isn’t perfect, but a greater in-depth exploratory understanding and an explicit discussion of design can enhance our chances for the production of valid explanations.

Large-scale Cross-population Comparison People around the world differ remarkably in how they think, talk, and behave. Understanding the roots of this striking cross-population variation has been one of

14_107-Bernard.indb 121

6/9/14 10:02 AM

122

Jeffrey C. Johnson and Daniel J. Hruschka

the core concerns of anthropology since its inception. Why are some groups more collectivist than others? What explains differences in norms of fairness? Why do different groups have different forms of kinship and social organization? A long line of research in anthropology has used comparative, cross-population analysis to answer these questions, sometimes focusing on population variation in specific regions of the world (e.g., Edgerton 1971) and sometimes using worldwide samples. More recently, a number of researchers have developed novel forms of data collection for crosspopulation analysis. An exemplar in this emerging tradition is research by Joe Henrich and colleagues (2006) that examines how people make decisions to share with others in different cultural settings. The research was a response to work in behavioral economics and evolutionary psychology that tried to explain why people share substantial amounts of money with anonymous strangers, even though the recipients will never find out who gave them the cash. To explain this phenomenon, which had largely been observed among U.S., European, and college populations, researchers argued that this was a psychological relic from our evolutionary past. They posited that we evolved in an environment where we rarely encountered strangers, so we give to strangers today because such an anonymous situation doesn’t make sense to us. The prediction from this explanation is that all humans should show similar biases toward sharing with strangers. To test this prediction, Henrich and colleagues (2006) brought the same experiments used by behavioral economists to small-scale societies around the world. People were asked to allocate 10 real dollars (or the local equivalent) to another anonymous person. The researchers found striking cross-population variation in people’s offers, all the way from stingy (giving next to nothing) to hyper-fair (giving more than half). These findings raised serious questions about explanations for such giving based on a pan-human–evolved psychology. The researchers also found in two independent studies (Henrich et al. 2010; Hermann et al. 2008) that the degree to which a population depended on markets (% of calories purchased) accounted for a large part of variation in how much people shared with strangers. Specifically, populations interacting more with markets showed more fair offers on average, a finding more consistent with adaptation to local social and cultural environments (Henrich et al. 2010). Since that time, similar studies have examined cross-cultural differences in willingness to punish others (Henrich et al. 2010; Herrman et al. 2008) and willingness to violate a rule to help friends and community members (Hruschka 2010). Figure 3.6 shows the overall design framework for this study. Such research is challenging. It requires long-term collaboration across a number of field sites and it can be difficult to identify measures that are meaningful across diverse societies. If you are studying friendship, for example, what word would you use for friend in the language of each society studied (Hruschka 2010), or does the measure used by the researcher have the same meaning (Gelfand et al. 2011)? However, rather than being a reason to avoid such explorations, dealing with these issues can teach us a great deal about where human populations diverge and where they are similar.

14_107-Bernard.indb 122

6/9/14 10:02 AM

3: Research Design and Research Strategies

123

Figure 3.6. Overall design framework for the Henrich et al. studies.

Field Experiments Observational data can help us understand how people think and act in different social and cultural situations. However, it is sometimes difficult to infer from observational data alone what causes people to do one thing or another. Suppose that we find that poor individuals are less likely to take medical treatment for tuberculosis than are wealthy individuals. We might conclude that lack of resources affects people’s behavior. It is also possible, though, that underlying variables, such as cultural beliefs about sorcery, might be the real drivers. One way to address such potential confounding by another variable is to conduct a field experiment. A field experiment randomly assigns people to a treatment condition and a control condition so that they ideally differ only on whatever is included in the treatment. Then, if we see a difference in behavior between the two groups, we can be fairly confident that the difference resulted from the treatment (rather than other unmeasured variables). Paul Farmer conducted just such an experiment at a clinic in rural Haiti to find out how to improve adherence to tuberculosis treatment (Farmer 1999). Treating tuberculosis requires an extended regime of antibiotics. If not followed completely, the patient’s condition can deteriorate and drug-resistant strains of TB can result. Despite the availability of treatment at the clinic, people were still dying of the disease, and practitioners proposed two theories. Community health workers pointed to economic barriers to completing regular treatment as well as problems of treating TB when people are malnourished. Staff doctors and nurses, on the other hand, argued that patients often stopped taking pills in part because they believed TB was caused by sorcery rather than microbes. To assess which factor would be most important in improving care, Farmer assigned TB patients to two groups of 50 individuals each (Farmer 1999, 219). One group got free treatment. The other got free treatment, but members were also eligible for a stipend of $30 per month for the first three months, $5 in travel expenses for coming to the clinic, nutritional supplements, and daily visits from a community health worker for the first month. To assess any potential impact of sorcery beliefs on treatment, each participant was interviewed about how he or she perceived the causes and treatments for TB (see Figure 3.7). The two groups differed very little in terms of age and sex, although economic indicators suggested that the treatment group was slightly poorer than the control group. Nearly all participants felt that sorcery played some role in their illness. However, a year after the study began, there were big differences in treatment outcomes. The treatment group had a 100% cure rate, whereas the control group had a cure rate of less than

14_107-Bernard.indb 123

6/9/14 10:02 AM

124

Jeffrey C. Johnson and Daniel J. Hruschka

Figure 3.7. Overall design framework for the Farmer study.

50%. The treatment group showed much greater weight gain (10 pounds vs. 2 pounds), and more members of the treatment group returned to work after one year of treatment (92% vs. 48%). After 18 months, one person from the treatment group had died, but not from TB. In the control group, six patients had died. These dramatic results indicated that small changes in economic barriers (and perhaps social influence from community health workers) could dramatically change treatment behavior and health outcomes. Moreover, sorcery beliefs played little role in the outcomes. Armed with such powerful results, the clinic began treating all patients with the combined package (Kidder 2004). Such field experiments are always subject to constraints of the field setting. For example, Farmer did not assign patients completely at random. Individuals in the treatment group were selected from a single region, whereas individuals in the control group were selected from patients not from that sector. However, checks on other variables for comparability across groups provide additional guarantees that the assignment was effectively random. Large-scale field experiments can be incredibly costly and time consuming to implement. However, their benefits in determining causality have led social science researchers to make increasing use of them in areas of critical importance for health and well-being (Banarjee and Duflo 2011). The same principle of randomization can also be applied at shorter time scales to understand how priming individuals with information or resources changes their behavior or responses. How does framing someone as a friend or a stranger change someone’s willingness to help or trust that person (Cronk 2007; Hruschka 2010) or how does thinking about a higher power make us more or less likely to share with others (Shariff and Norenzayan 2007)? Such experiments complement the findings and hypotheses generated by observational data.

Repeated Measures Design: The Evolution of Network Structure As discussed earlier in this chapter, experimental and longitudinal designs are necessary for determining causal relationships. Whereas true experiments are not always possible, or for that matter desirable, longitudinal designs are quite appropriate for testing hypotheses in many ethnographic contexts. Ethnography, by its very nature, involves extended periods of time in one or more given field sites. Thus, they provide the opportunity to collect data over time and, possibly, across multiple groups. In addition, unlike the requisite artificial conditions imposed by experiments (i.e., manipulation of

14_107-Bernard.indb 124

6/9/14 10:02 AM

3: Research Design and Research Strategies

125

independent variable while controlling for all other variables), longitudinal research in field settings allows for the study of social and behavioral phenomena in situ. Although anthropologists tend to spend months, sometimes years, in the field collecting data, few of them use panel or cohort designs. Gravlee et al. (2009) carefully document this and point out that anthropologists rarely employ systematic longitudinal designs in their work. The article provides a good discussion of the ins and outs of systematic panel and cross-sectional panel designs, informative discussions of sources of potential errors in such designs, and anthropological examples of work in this vein. In research that spanned well over eight years, Johnson, Boster, and Palinkas (2003; Johnson, Palinkas, and Boster 2003; Palinkas, Johnson, and Boster 2004; Palinkas et al. 2004) were interested in the influence of the emergence of informal social roles on the evolution of small-group network structure and ultimately on group well-being and performance. In essence, they conducted a kind of natural experiment by studying populations that were quite isolated over an extended period of time so that the emergent properties of the groups could be more readily studied with less interference from extraneous influences. Their basic theoretical proposition was that groups that formed more cohesive networks (i.e., core–periphery networks) would have higher morale and would be more productive. Furthermore, the presence of various informal social roles helps facilitate social cohesion. To test these hypotheses, they went to Antarctica to study small-group dynamics at polar research stations. The design of the study was quite simple and involved repeated interviews with winter-over crew members at Antarctic stations during the winter (an interview each month over the eight–nine months of the winter). Although a single group at a single station could have been studied over time (i.e., a single case-study design), it would have posed a number of problems for adequately testing the hypotheses of interest. To be able to draw conclusions at the group level of analysis, as opposed to simply the individual-actor level, there needed to be observations on multiple cases (i.e., across multiple years). Therefore, the design involved three observations at a station (i.e., the group level) that included the study of three separate groups per station each of a year’s duration. In addition, the researchers were interested in the role of culture in group formation. They therefore included five different cultures in the overall design. These included the Americans at the Amundsen-Scott South Pole Station, the Russians at Vostok Station, the Chinese at the Great Wall Station, the Poles at the Arktowski Station, and the Indians at the Maitri Station. This is similar to the rationale for comparisons discussed in the cross-population comparison section earlier and allowed for the study of variation both within and between the various stations and cultures and facilitated the study of social networks at the group level, the dyadic or tie level, and the individual-actor level (see Borgatti et al. [2013] for a review). Prior to the data collection at the stations, Johnson et al. (2003) conducted a series of semi-structured interviews with former winter-overs to elicit informal social roles recognized by the winter-overs themselves. In addition, the social network questions were developed in consultation with winter-over crew members early in the study to maximize cultural understanding and appropriateness of the social network questions (see Figure 3.8). These early exploratory interviews were important for producing a condensed survey instrument to be administered during the monthly winter

14_107-Bernard.indb 125

6/9/14 10:02 AM

126

Jeffrey C. Johnson and Daniel J. Hruschka

Figure 3.8. Overall design framework for the Johnson et al. study.

interviews. The researchers did not want to risk over-burdening the crew members with questions not central to the testing of their research hypotheses (i.e., not a datafishing expedition). A shorter, more theoretically relevant survey instrument would help ensure a sustained level of study participation over the course of the winter. A huge threat in repeated measures designs involves individuals dropping out of the study prematurely (Gravlee et al. 2009). Missing data are also critically important for social network studies. The research found that winter-over groups in which various informal social roles (e.g., clown, expressive leaders) emerged over the course of the winter had more cohesive social networks and higher morale and individual level psychological well-being. In addition, these findings were consistent across the five cultures, suggesting some degree of cultural universality in human group dynamics. Since the researchers were interested in social group dynamics, particularly in aspects of emergent group properties, a longitudinal design was essential for them to be able to draw the conclusions they did. SUMMARY

This review of research design and strategies in cultural anthropology only scratches the surface of the research designs, hybrid designs, and combinations of designs possible within an ethnographic context. The strength of the ethnographic approach is its ability to incorporate a wide range of methods, strategies, and designs within a single enterprise, all combining in ways to improve the chances for credible and valid results. As anthropologists, we should take full advantage of both our current understanding of research design and these new developments to produce a “powerful mode of argumentation.” It is mostly through attention to these concerns that anthropology and anthropologists will have the opportunity to, as Agar says “move the world” (1996, 13). REFERENCES

Agar, M. 1980. The professional stranger: An informal introduction to ethnography. New York: Academic Press. Agar, M. 1996. Schon Wieder? Science in linguistic anthropology. Anthropology News 37(1): 13.r. January.

14_107-Bernard.indb 126

6/9/14 10:02 AM

3: Research Design and Research Strategies

127

Babbie, E. 1990. Survey research methods, 2nd ed. Belmont, CA: Wadsworth. Baer, R. D., S. C. Weller, J. G. Garcia, and A. L. S. Rocha. 2012. Ethnomedical and biomedical realities: Is there an epidemiological relationship between stress-related folk illnesses and type 2 diabetes? Human Organization 71: 339–47. Banarjee A. V., and E. Duflo. 2011. Poor economics: A radical rethinking of the way to fight global poverty. New York: Public Affairs Press. Behar, R. 1993. Translated woman. Boston: Beacon Press. Bernard, H. R., ed. 1998. Handbook of methods in cultural anthropology. Walnut Creek, CA: Sage. Bernard, H. R. 2011. Research methods in anthropology: Qualitative and quantitative approaches, 5th ed. Walnut Creek, CA: AltaMira. Bernard, H. R. 2012. Social research methods: Qualitative and quantitative approaches, 2nd ed. Thousand Oaks, CA: Sage. Boas, F. 1920. The methods of ethnology. American Anthropologist 22: 311–21. Boellstorff, T. 2008. Coming of age in Second Life: An anthropologist explores the virtually human. Princeton, NJ: Princeton University Press. Borgatti, S., M. Everett, and J. C. Johnson. 2013. Analyzing social networks. London: Sage. Boster, J. S., and J. C. Johnson. 1989. Form or function: A comparison of expert and novice judgments of similarity among fish. American Anthropologist 91: 866–89. Brim, J. A., and D. H. Spain. 1974. Research design in anthropology: Paradigms and pragmatics in the testing of hypotheses. New York: Holt, Rinehart and Winston. Campbell, D. T. 1988. Qualitative knowing in action research. In Methodology and epistemology. for social science: Selected papers, ed. E. S. Overman, 360–76. Chicago: University of Chicago Press. Cook, T. D., and D. T. Campbell. 1979. Quasi-experimentation: Design and analysis for field settings. Chicago: Rand McNally. Cronk, L. 2007. The influence of cultural framing on play in the trust game: A Maasai example. Evolution and Human Behavior 28: 352–58. Denzin, N. K., and Y. S. Lincoln. 1994. Entering the field of qualitative research. In Handbook of qualitative research, ed. N. K. Denzin and Y. S. Lincoln, 1–19. Thousand Oaks, CA: Sage. Edgerton, R. 1971. The individual in cultural adaptation: A study of four East African peoples. Berkeley: University of California Press. Edgington, E., and P. Onghena. 2007. Randomization tests, 4th ed. London: Chapman and Hall/ CRC. Ellen, R. F. 1984. Introduction. In Ethnographic research: A guide to general conduct, ed. R. F. Ellen, 1–12. London: Academic Press. Farmer, P. 1999. Infections and inequalities. Berkeley: University of California Press. Freeman, D. 1983. Margaret Mead and Samoa: The making and unmaking of an anthropological myth. Cambridge, MA: Harvard University Press. Gelfand, M. J., J. L. Raver, L. Nishii et al. 2011. Differences between tight and loose cultures: A 33-nation study. Science 332: 1100–104. Gravlee, C. C., D. P. Kennedy, R. Godoy, and W. R. Leonard. 2009. Methods for collecting panel data: What can cultural anthropology learn from other disciplines? Journal of Anthropological Research 69: 453–83. Harris, M., J. G. Consorte, J. Lang, and B. Byrne. 1993. Who are the Whites: Imposed census categories and the racial demography of Brazil. Social Forces 72: 451–62. Henrich, J., R. McElreath, A. Barr et al. 2006. Costly punishment across human societies. Science 312: 1767–70. Henrich, J., R. McElreath, A. Barr et al. 2010. Markets, religion, community size, and the evolution of fairness and punishment. Science 327: 1480–84.

14_107-Bernard.indb 127

6/9/14 10:02 AM

128

Jeffrey C. Johnson and Daniel J. Hruschka

Hermann, B., C. Thöni, and S. Gächter. 2008. Antisocial punishment across societies. Science 319: 1362–67. Ho, K. 2009. Liquidated: An ethnography of Wall Street. Durham, NC: Duke University Press. Hruschka, D. J. 2010. Friendship: Development, ecology and evolution of a relationship. Berkeley: University of California Press. Hubert, L. J. 1987. Assignment methods in combinational data analysis. New York: Marcel Dekker. Hurlbert, S. H. 1984. Pseudoreplication and design of ecological field experiments. Ecological Monographs 54: 187–211. Johnson, J. C. 1990. Selecting ethnographic informants. Qualitative Research Methods Series, Vol. 22. Thousand Oaks, CA: Sage. Johnson, J. C., C. Avenarius, and J. M. Weatherford. 2006. The active participant observer: Applying social role analysis to participant observation. Field Methods 18: 111–34. Johnson, J. C., J. S. Boster, and L. Palinkas. 2003. Social roles and the evolution of networks in isolated and extreme environments. The Journal of Mathematical Sociology 27: 89–122. Johnson, J. C., M. Ironsmith, A. L. Whitcher et al. 1997. The development of social networks in preschool children. Early Education and Development 8: 389–406. Johnson, J. C., and J. D. Murray. 1997. Evaluating FAD effectiveness in development projects: Theory and praxis. In Fish aggregation devices in developing fisheries: Potential and pitfalls, ed. R. Pollnac and J. Poggie, 143–58. Kingston, RI: ICMRD. Johnson, J. C., L. A. Palinkas, and J. S. Boster. 2003. Informal social roles and the evolution and stability of social networks. In Dynamic social network modeling and analysis, ed. R. Brieger, K. Carley, and P. Pattison, 121–32. Washington, DC: The National Academies Press. Johnson, J. C., and R. Pollnac, eds. 1989. Managing marine conflicts. Special issue of Ocean and Shoreline Management 12(3). Kempton, W., J. S. Boster, and J. A. Hartley. 1996. Environmental values in American culture. Cambridge, MA: MIT Press. Kidder, T. 2004. Mountains beyond mountains. New York: Random House. Kincheloe, J. L., and P. L. McLaren. 1994. Rethinking critical theory and qualitative research. In Handbook of qualitative research, ed. N. K. Denzin and Y. S. Lincoln, 138–58. Thousand Oaks, CA: Sage. Kleinbaum, D. G., L. L. Kupper, and H. Morgenstern. 1982. Epidemiologic research: Principles and quantitative methods. Belmont, CA: Lifetime Learning Publications. Koester, S. 1996. The process of drug injection: Applying ethnography to the study of HIV risk among IDUs. In AIDS, drugs and prevention: Perspectives on individual and community action, ed. T. Rhodes and R. Hartnoll, 133–48. London: Routledge Press. Koester, S., R. E. Booth, and Y. Zhang. 1996. The prevalence of additional injection relation HIV risk behaviors among injection drug users. Journal of Acquired Immune Deficiency Syndromes and Human Retrovirology 12: 202–7. LeVine, R. A. 1973. Research design in anthropological field work. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 183–95. New York: Columbia University Press. Lowie, R. H. 1937. The history of ethnological theory. New York: Rinehart. Marcus, G. 1995. Ethnography in/of the world system: The emergence of multi-sited ethnography. Annual Review of Anthropology 24: 95–117. Miles, M. B., and A. M. Huberman. 1994. Qualitative data analysis, 2nd ed. Thousand Oaks, CA: Sage. Moran, E. F., ed. 1995. The comparative analysis of human societies: Toward common standards for data collection and reporting. Boulder, CO: Lynne Rienner.

14_107-Bernard.indb 128

6/9/14 10:02 AM

3: Research Design and Research Strategies

129

Naroll, R., and R. Cohen, eds. 1973. A handbook of method in cultural anthropology. New York: Columbia University Press. Noreen, E. W. 1989. Computer-intensive methods for testing hypotheses. New York: John Wiley. Orans, M. 1996. Not even wrong: Margaret Mead, Derek Freeman, and the Samoans. Novato, CA: Chandler and Sharp. Palinkas, L. A., J. C. Johnson, and J. S. Boster. 2004. Social support and depressed mood in isolated and confined environments. Acta Astronautica 54: 639–47. Palinkas, L. A., J. C. Johnson, J. S. Boster et al. 2004. Cross-cultural difference in psychosocial adaptation to isolated and confined environments. Aviation, Space, and Environmental Medicine 75: 973–80. Panourgia, N. 1995. Fragments of death, fables of identity. Madison: University of Wisconsin Press. Pelto, P. J., and G. H. Pelto. 1978. Anthropological research: The structure of inquiry, 2nd ed. Cambridge: Cambridge University Press. Porter, T. M. 1995. Trust in numbers. Princeton, NJ: Princeton University Press. Ramos, A. R. 1995. Sanuma memories. Madison: University of Wisconsin Press. Robson, C. 2002. Real world research: A resource for social scientists and practitioner-researchers, 2nd ed. Hoboken, NJ: John Wiley & Sons. Scheper-Hughes, N. 2004. Parts unknown: Undercover ethnography of the organs-trafficking underworld. Ethnography 5: 29–73. Schneider, D. 1996. Alarming nets. Scientific American (September): 40–42. Sechrest, L. 1973. Experiments in the field. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 196–209. New York: Columbia University Press. Shariff, A. F., and A. Norenzayan. 2007. God is watching you: Priming God concepts increases prosocial behavior in an anonymous economic game. Psychological Science 18: 803–9. Spindler, G., and W. Goldschmidt. 1973. An example of research design: Experimental design in the study of culture change. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 210–19. New York: Columbia University Press. Stinchcombe, A. L. 1987. Constructing social theories. Chicago: University of Chicago Press. Tsing, A. 2005. Friction: An ethnography of global connection. Princeton, NJ: Princeton University Press. Tyler, S. A. 1991. A post-modern instance. In Constructing knowledge: Authority and critique in social science, ed. L. Nencel and P. Pels, 78–95. London: Sage. Urry, J. 1984. A history of field methods. In Ethnographic research: A guide to general conduct, ed. R. F. Ellen, 35–62. London: Academic Press. Van Maanen, J. 1988. Tales of the field: On writing ethnography. Chicago: University of Chicago Press. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Qualitative Research Methods Series, Vol. 10. Thousand Oaks, CA: Sage. Werner, O., and G. M. Schoepfle. 1987. Systematic fieldwork, Vol. 2. Thousand Oaks, CA: Sage. Whyte, W. F. 1984. Learning from the field: A guide from experience. Newbury Park, CA: Sage. Williams, T. R. 1967. Field methods in the study of culture. New York: Holt, Rinehart and Winston. Young, J. C., and L. Y. Garro. 1982. Variation in the choice of treatment in two Mexican communities. Social Science and Medicine 16: 1453–65. Zabusky, S. E. 1995. Launching Europe: An ethnography of European cooperation in space science. Princeton, NJ: Princeton University Press.

14_107-Bernard.indb 129

6/9/14 10:02 AM

14_107-Bernard.indb 130

6/9/14 10:02 AM

CHAPTER 4

Ethics Carolyn Fluehr-Lobban

Matters of ethics are an ordinary, not extraordinary, part of anthropological practice. Considerations of ethics are often raised after some incident has occurred or when an anthropologist feels discomfort about the conduct or progress of fieldwork. The anthropologist’s methods may be questioned, or his or her position in the social group may be challenged. Access to resource persons may have changed dramatically or even been denied. Research goals may appear suddenly unattainable and, confused or desperate, the anthropologist may wonder what has gone wrong and why. When an anthropological researcher and persons studied respect and trust one another, there are usually positive feelings and outcomes on both sides. However, if an anthropologist has had little or poor training for ethical engagement in the field, it will be difficult to resolve the everyday dilemmas that are part of the practice of the discipline. For example, the anthropologist may not have been well prepared for the complexities of the social and political environments of the people to be studied. Or the anthropologist may receive funding from a private or public foundation, governmental or nongovernmental, without having fully considered the potential conflicting demands and responsibilities between the funder and the people studied. During the research, when the field situation changes and the anthropologist’s position becomes ambiguous, the art of negotiating and repositioning oneself resolves to ethical principles and choices. There have been some celebrated cases of alleged ethical misconduct involving major political events of the day. These occurred in Latin America in 1960s, in Vietnam in the 1960s and 1970s, and in the People’s Republic of China in the 1980s. In the post– September 11, 2001 era, in which concerns over the security of America and Americans has predominated foreign travel and research, the environment of anthropological research has become more complex and challenging. Historically, ethics and professional responsibilities have been sensitive and controversial topics in anthropology. Moreover, because ethical conduct may be perceived as overlapping with morality or personal principles, anthropologists, like other human beings, may be reluctant to talk about ethics for fear of having done something wrong. It used to be that just raising the issue of ethics in fieldwork caused a defensive reaction, with the person being queried feeling that he or she was being accused of alleged wrongdoing rather than simply asking how ethics were a part of the fieldwork plan. In this chapter on ethics and methods in anthropological research, the focus is on how to make ethical considerations an integral part of the ordinary, day-to-day practice of our craft. 131

14_107-Bernard.indb 131

6/9/14 10:02 AM

132

Carolyn Fluehr-Lobban

HISTORY OF ETHICS IN ANTHROPOLOGY

The history of ethical discourse in cultural anthropology intersects with national and international politics and the changing contexts and paradigms of fieldwork. Various incidents have precipitated crises that have forced anthropologists to hold a mirror up to themselves and to be reflective about the consequences of our actions. Until recently, historical concern with matters of ethics has been more reactive than proactive, more defensive maneuvering than an affirmative tackling of ethical issues. However, significant changes have taken place in the last two decades, with two codes of ethics proffered by the American Anthropological Association (AAA), in 1998 and 2011, and a second round of the debate about the appropriate relationship between anthropologists and government brought about by the wars in Iraq and Afghanistan. I shall try to balance the broad historical picture of crises about ethics and anthropology with these recent developments in the profession. The first formal statement on ethics was developed by applied anthropologists in 1949, following discussions that began in the immediate post–World War II era. The Society for Applied Anthropology Committee was chaired by Margaret Mead and saw its role as creating a document for general use by applied anthropologists. Nevertheless, fundamental issues of research—such as the relationship between the anthropologist and research collaborators, employers, nongovernmental, or voluntary organizations, and the protection of informant confidentiality—are all significantly present in the statement. The AAA produced its first statement on ethics nearly two decades later, in 1967, in the context of the Vietnam War. The first ethics “incident” took place in World War I. It involved Franz Boas, America’s founding anthropologist and the designer of the unique holistic brand of U.S. anthropology. A prolific writer and experienced ethnographer, Boas also had a strong civic sensibility and was a frequent contributor to numerous nonanthropological periodicals. In a letter published in The Nation, Boas, widely known as a pacifist, objected to the wartime activities of four anthropologists who had combined intelligence gathering with their research. By accident incontrovertible proof has come to my hands that at least four men who carry on anthropological work, while employed as government agents, introduced themselves to foreign governments as representatives of scientific institutions in the United States, and as sent for the purpose of carrying on scientific research. They have not only shaken the truthfulness of science, but have also done the greatest possible disservice to scientific inquiry. (Boas 1919)

For this letter of protest, Boas was censured formally by the Anthropology Society of Washington and isolated in general by the anthropological establishment for having drawn the general public’s attention to the activities of several anthropologists working as government intelligence gatherers. The issue eventually subsided and Boas’s career was not adversely affected. Sensitivities may have persisted, however, for Boas did not resurrect the subject even though the same situation might have arisen during the years of World War II while he was still alive. Boas’s censure was only lifted by a special resolution introduced at the AAA business meeting in 2004.

14_107-Bernard.indb 132

6/9/14 10:02 AM

4: Ethics

133

Throughout the twentieth century and into the twenty-first, the issue of anthropologists working for the U.S. government and, after 9/11/2001, for the U.S. military, not only retained its potency but took on new significance. The AAA’s first formal statement on ethics in 1967 was drafted in the wake of the U.S. Department of the Army’s Project Camelot, after revelations that counterinsurgency research in Latin America was a major focus of the proposed social science research. The AAA’s first statement makes explicit reference to avoiding the use of the name of anthropology or the title of anthropologist as a cover for intelligence activities (AAA Statement on Ethics 1971). Federal funds made available through the Department of Defense’s National Security Education Program (NSEP) for training in foreign language and international studies established in 1991 were controversial, as NSEP funds were for undergraduate study abroad and fellowships for graduate students to strengthen the national capacity in international education in critical world areas and were linked to future careers in intelligence and related government service. Although funding for international studies had been declining and the Department of Defense monies were tempting, the NSEP was criticized by virtually every major area studies association—including the African Studies Association, the Middle East Studies Association, the Latin American Studies Association, and the Asian Studies Association. Further, their members and potential student applicants were cautioned about the risks of the relationship of foreign area studies scholars with military and intelligence-gathering organizations and priorities. The AAA, while it cautioned its members to negotiate and understand clearly their relationship to their own and host governments, did not take any specific action or pass any resolution regarding the creation of NSEP. The broadly based reaction to the Department of Defense’s NSEP is a reminder of the seriousness of the issue of the use or appearance of the use of anthropology or any discipline as a cover for intelligence gathering. It demonstrated the continuing relevance of Boas’s admonition of 1919 to avoid such ties to defense, military, and intelligence agencies. The issue reemerged in the post–9/11 era almost exactly as it did in the 1990s, with reaction to the Minerva program to enhance graduate education in foreign cultures of interest to U.S. security, especially in the Middle East and Islamic world. The profession was presented with another challenge when a front-page article appeared in the New York Times in 2007 detailing the work of anthropologist “Tracy,” an embedded social scientist with the new U.S. Army human terrain teams (HTT) serving as a field cultural intelligence adviser in counterinsurgency operations in Afghanistan (Rohde 2007). This created another firestorm that resulted in the formation of an AAA Commission on Engagement of Anthropology with the U.S. Security and Intelligence Communities that met for four years and issued a final report focusing heavily on the HTTs. The deaths of three nonanthropologist social scientists added potentially deadly consequences to the matter of engagement as well as raising compelling moral and ethical issues. Left inadequately addressed in the final report was any assessment of the HTTs on the ground and the degree to which they may be responsible for harm reduction through the use of applied cultural knowledge. The view that trained anthropologists working in these operations may be more appropriately labeled intelligence workers

14_107-Bernard.indb 133

6/9/14 10:02 AM

134

Carolyn Fluehr-Lobban

does not fully satisfy the moral and ethical question as to whether anthropologists should or should not engage in such defense and intelligence work. Nonetheless, the Vietnam War era was the most important turning point for anthropology and other social science disciplines. It caused anthropologists to take matters of professional ethics seriously, to develop a code of conduct, and to provide a mechanism for enforcement. Once again, allegations that anthropologists had engaged in counterinsurgency research, this time in Thailand and southeast Asia, precipitated a crisis in the field. Anthropologist was pitted against anthropologist, just as the entire American society was torn apart by issues of politics and morality in the Vietnam War. In 1971, after heated confrontations in print and in public meetings, a Committee on Ethics was formed and the first code of ethics—the Principles of Professional Responsibility—was drafted and promulgated by the AAA (AAA 1971). In this code, a strong first principle was enunciated that lasted for a generation: In research, the anthropologist’s paramount responsibility is to those he studies. When there is a conflict of interest, these individuals must come first. Today, with more than half of anthropologists being female, the reference to “he” is unacceptable, but this language illustrates that much has changed in ethics as well as gender since 1971. Other strong principles reflected the tenor of the times: 1) that no secret or clandestine research be carried out; that even the appearance of conducting clandestine research hurts the good reputation of anthropology and its practitioners; 2) that in relations with one’s own and host governments anthropologists should not be required to compromise ethics as a condition for the conduct of research 3) when an anthropologist by his[her] actions jeopardizes peoples studied, colleagues, students or others the Association may inquire into the propriety of these actions and take appropriate measures. (AAA 1971 [as amended through November 1976]; reprinted in FluehrLobban 1991b, 247–56)

As a marked reflection of the strength of the idea of ethical principles over prescriptions of dos and don’ts, the most recent revision of AAA code of ethics in 2011 is based entirely on principles of ethics from which good professional practice is derived. These principles are concise and easily understood and remembered tenets of ethical research: (1) do no harm; (2) be open and honest regarding your work; (3) obtain informed consent and necessary permissions; (4) balance ethical obligations to collaborators and affected parties; (5) make your results accessible; (6) protect and preserve your records; and (7) maintain respectful and ethical professional relationships (AAA 2012). Simple guiding principles are preferable to more complicated codes or unsuccessful efforts to enforce standards (Fluehr-Lobban 2009) and should enhance the larger and most important mission of ethics education at every stage of anthropological study and practice. HISTORICAL TENSION BETWEEN APPLIED AND ACADEMIC ANTHROPOLOGISTS

In the decades after the end of the Vietnam War, the immediacy and intensity of these selected core principles faded dramatically as contract or proprietary research for governmental and nongovernmental agencies became more common and tended to neutralize the issue of secret/clandestine research for governments. This coincided with the

14_107-Bernard.indb 134

6/9/14 10:02 AM

4: Ethics

135

post–Vietnam War economic downturn. Many professionally trained anthropologists were unable to secure academic employment, and some turned to professional work in applied fields. This demographic transformation meant that in 1986, for the first time since the founding of the AAA in 1902, more anthropologists (51%) were employed outside academia than within it (Fluehr-Lobban 1991a, 5). Throughout the 1980s, practicing anthropologists grew in such numbers so as to constitute something of a lobbying force within the AAA, and some formed a new professional organization, the National Association of Practicing Anthropologists (NAPA). It sought to revise the PPR to make it more accommodating to the interests and needs of professional, nonacademic anthropologists. In particular, members of NAPA were concerned that the first responsibility of the anthropologist to the people studied could not always be met under the terms of contract research. They attempted unsuccessfully to revise the PPR in 1984, again precipitating charged debates within the profession. They eventually created their own set of ethical guidelines that were more suited to the complexities of contract and proprietary research. Many academic anthropologists, recalling the heady days in which the PPR was created, were reluctant to see any modification of the principle that the first responsibility in research is to the people studied. I have summarized some of the tensions between academic and applied anthropologists expressed during this time (Fluehr-Lobban 1991b), suggesting that the differences were significant enough to challenge the traditional holism of anthropology itself. Discussions of differences, real and imagined, between academic and practicing anthropologists continued throughout the 1980s. They were carried out in various forums, including symposia at the annual AAA meetings, several of which I organized and chaired. These debates and discussions evoked the subtitle of my 1991 edited volume on ethics and the profession of anthropology, Dialogue for a New Era (Fluehr-Lobban 1991b).

Darkness in El Dorado Controversy Informed consent lies at the core of ethical principle and practice in Western biomedical and social and behavioral science. There was no specific informed consent language in anthropological codes of ethics until 1998. There is an anecdotal history of the resistance and even refusal of some anthropologists to come before their university institutional review boards (IRBs) and defend their non-application of the essential components of informed consent in their research (Fluehr-Lobban 1994). It is no longer possible for anthropologists to resist or oppose the concept of informed consent in research, as their research is subject to ethical scrutiny by their peers from outside the discipline and profession. The AAA code of ethics acknowledged in 1998 that, “It is the quality of the consent, not the format, that is relevant. Further, it is understood that the informed consent process is dynamic and continuous; the process should be initiated in the project design and continue through implementation by way of dialogue and negotiation with those studied” (AAA CoE 1998, 3.4). The controversy over the publication of Darkness in El Dorado, How Scientists and Journalists Devastated the Amazon (2000) by Patrick Tierney highlighted the historical lack of informed consent. The book charged anthropologist Napoleon Chagnon and others with inciting warfare, introducing or not treating a measles epidemic,

14_107-Bernard.indb 135

6/9/14 10:02 AM

136

Carolyn Fluehr-Lobban

staging films, and falsifying data on the Yanamomö, portraying them as a “fierce people,” which drew the attention of scientists and journalists worldwide, leading to devastating results for this indigenous people. This was all done during fieldwork in the 1960s, without any consideration of the knowledge or consent of the Yanomami. The story broke in cyberspace in September 2000, a month before the book was due to be released (Fluehr-Lobban 2000). The AAA formed an El Dorado Task Force to determine whether the allegations made in the book warranted formal investigation and promised to review its code of ethics in light of the issues raised. The government of Venezuela imposed a moratorium on all research in indigenous areas. The Final Report of the Task Force did not make any significant finding of ethical standards violations by Chagnon or any other American anthropologist, but faulted Tierney for violating journalistic ethics. It called on anthropologists to seize on this opportunity to consider the ethics of the anthropological enterprise and the plight of the Yanomami and other indigenous peoples, and to use this case to reflect on the ways they conduct research (AAA 2002). After this crisis, the old order of unregulated, laissez-faire research passed away. Tension between practicing and academic anthropologists surfaced during the 1980s, when jobs for anthropologists were scarce. Academic anthropologists may have not fully understood or were skeptical about the types of contract-based research with which practicing anthropologists were involved. And applied anthropologists may have experienced some isolation from the traditionally academically oriented discipline as they pioneered new venues for applying and using anthropological research. Some academic anthropologists asserted that their research was “pure,” whereas applied research is compromised by the client–researcher relationship. These tensions have eased as the job market improved somewhat and more pressing matters of military and security anthropology penetrated anthropology in the two wars that ensued in the post–9/11 era of the new millennium. The codes of ethics of 1998 and 2011 make no distinction whatever between so-called pure and so-called applied research. Practicing anthropologists have contributed ideas and their perspective to the new code. In the end, research is research, period. REVISION OF THE PPR, A CODE FOR ALL ANTHROPOLOGISTS

Beyond the tension between academic and practicing anthropologists, other weaknesses in the PPR were revealed. For example, a complex grievance procedure was provided for the AAA Committee on Ethics (CoE). The procedure included receiving and screening cases, with the Executive Board being responsible for the final judgment and any action taken. However, not a single anthropologist was censured after its adoption in 1976. The CoE was primarily called on to mediate disputes between colleagues over matters of plagiarism, employment, and personal differences. Based on her experience as former member and chair of the AAA CoE, Janet Levy concluded that the grievance mechanism should be reviewed and possibly eliminated because of the complex procedures involving multiple levels of review and the slowness and general lack of resolution of intraprofessional disputes (Levy 1994). In 1994, the AAA decided to systematically reexamine the underlying principles and practice of professional ethics within the discipline. A Commission to Review the AAA

14_107-Bernard.indb 136

6/9/14 10:02 AM

4: Ethics

137

Statements on Ethics was formed in 1994 and chaired by James Peacock, then president of the AAA. The commission included AAA members from archaeology and physical anthropology as well as applied and cultural anthropologists who have traditionally been at the center of ethics discourse. I was a member, along with James Peacock, Barbara Frankel, Janet Levy, Murray Wax, and Kathleen Gibson. The commission discussed changes in the PPR and reached consensus on a number of fundamental issues. After two years, the PPR was revised and a new code of ethics was drafted. The most important change was the shift from a code with a grievance procedure to one that emphasized ethics education. Ethics education was preferred so that an ongoing discussion of ethical issues could take place as the profession develops and adapts to changing circumstances for anthropological research. A major development was that, for the first time, the code of the AAA embraced a four-field definition of American anthropology: “Anthropological researchers have primary ethical responsibility to the people, species, and materials they study and to the people with whom they work.” This is a modification of the former “first responsibility” of the anthropologist to the people studied, a cornerstone principle of the 1971 PPR. Thus, the anthropologist was now enjoined to do no harm, to respect the well-being of humans and nonhuman primates, and to conserve the archaeological, fossil, and historical record. Without being explicit, the code reinforced a holistic approach that has been the hallmark of American anthropology. It also reflected a reaction to the proliferation of codes of different anthropological groups. These codes had developed in response to their realization or complaint that the AAA code did not address their subgroups’ concerns. For example, by 1995, there were codes of professional ethics for the AAA, the PPR, for the Society for Applied Anthropology, the NAPA, the Society of Professional Archaeologists, and the code of the Archaeological Institute of America (see Fluehr-Lobban 1991b). The general admonition in the new code to “do no harm” is a basic principle in many professional codes and is purposefully broad. However, in commenting on the draft code of the mid-1990s, NAPA judged this phrasing to be too simplistic. Practicing anthropologists, they argue, are involved in many types of research, frequently affecting individuals and groups with diverse and sometimes conflicting interests. In such cases of potential conflict, no absolute rule can be observed, but the individual practitioner must make carefully considered ethical choices and be prepared to make clear the assumptions, facts, and issues on which those choices are based (Jordan 1996, 18). The issue of secret or clandestine research, which evoked such passion in the past, gradually evolved to concerns over proprietary research, where anthropologists contract their services to an agency for a specified period of time and amount of funding. Clandestine research has been consistently criticized in previous codes and statements on ethics; proprietary research has been an area where applied and practicing anthropologists have urged their colleagues to be aware of possible conflicts between the interests of funders of research and the people studied. Presumably, anthropologists conducting pure research were not constrained by their funders in the same ways that anthropologists conducting proprietary research are. The revised code extinguished distinctions between pure and applied research: Quite simply, research is research, and

14_107-Bernard.indb 137

6/9/14 10:02 AM

138

Carolyn Fluehr-Lobban

standards of ethical conduct in research are the same, irrespective of private or public funding. The new code does call for the open and accessible publication of research results within a reasonable period of time, which might be negotiated in advance, to ensure that secret research is not sanctioned. The 1998 code contained specific language regarding informed consent for the first time, which means that anthropology has situated itself within the broader scientific and professional community of scholars and scientists in the United States and increasingly in the world. The informed consent language is intended to encompass both the ethical and legal intent of this broad principle. This will be discussed in detail below. The issue of advocacy in anthropology was addressed as to whether or not this is an ethical responsibility. It was agreed that advocacy is very much a moral choice that may be made by individual anthropologists but that it is not a professional duty. When anthropologists choose advocacy, they should be cautious about the selective use of their data for whatever good cause is being advocated. In the present era, in which concerns about anthropologists engaging with the military and defense communities have arisen in conjunction with involvement of social scientists in the wars in Iraq and Afghanistan, a new round of review and revision of the AAA ethics code has taken place. This much simplified code offers core principles that govern all research and that should, theoretically, apply to all contexts and environments of research. More important is the discussion of each of these principles in the draft code that can guide thoughtful practitioners in their own decision making. The shift in professional responsibility to the anthropological researcher as bearing the weight of decision making is critical to the use of the code itself as a device, not a litany of commandments. Friends of the Committee on Ethics was formed in 2008 as an ad hoc consultative body comprised of former chairs of the Ethics Committee who serve as informal advisers to anthropologists seeking professional advice about ethical choices and decision making. The Friends committee can be contacted at: http://www. aaanet.org/cmte4s/ethics/Ethics-Resources.cfm. AN ETHICALLY CONSCIOUS MODEL OF ANTHROPOLOGICAL RESEARCH

The above discussion of the history of ethical discourse and concern within anthropology should demonstrate to the relative newcomer to the field that anthropologists and their professional associations take ethics seriously. Ethics should be considered first, not last, in anthropological fieldwork. Thinking about ethics should be a part of every research plan from beginning to end, from initial design, to pursuit of funding, through actual research, to eventual publication of results. There are many regulations that students and researchers become aware of only when conducting research, but a little advance preparation might allow them to make more informed and better decisions. The basic guidelines for the ethical conduct of research in anthropology, or any field, are openness and disclosure. The present draft code of the AAA is clear and unambiguous on this fundamental idea: “In both proposing and carrying out research, anthropologists must be open about the purpose(s), potential impacts, and source(s) of support for research projects with funders, colleagues, persons studied or providing information, and with relevant parties affected by the research” (Draft Code 1995:A.41).

14_107-Bernard.indb 138

6/9/14 10:02 AM

4: Ethics

139

CONSIDERATION OF ETHICS AT EVERY STAGE OF RESEARCH AND PUBLICATION

Anthropologists, and probably other researchers, used to think that they had complete freedom in the conduct of their research. “Just come up with a good idea and go and study it” might have been the advice of anthropologist mentors in past generations. However, the complexities and sensitivities of conducting research in the postcolonial, non-Western world, not to mention studying American populations, make this former absolutist position untenable today.

Potential Research Projects From the initial concept of a research project, an anthropologist should ask: Can I carry out the project without violating ethical standards or compromising professional standards? If there are questions about this, can I modify the research to accommodate the legitimate interests of the group to be studied? Is negotiating the terms of research a possibility or, among some groups, a necessity? Have I discussed the ethical issues or questions with other anthropologists and mentors, with members of the Institutional Review Board or the ethics committee of my institution, or with some representatives of the group I plan to study, if possible? Have I acquainted myself with the existence, guidelines, and procedures of tribal, ethnic, or national review boards? If ethical considerations are sufficiently serious, should I consider abandoning the project? Sources of Funding In the next phase, the anthropologist should ask the following questions: Have I considered various sources of funding appropriate to the research and their potential regulatory role over the conduct of my research? Have I considered the conditions that certain government or private funding sources may impose on my research? If I understand that, should I disclose the source(s) of my funding to the people/group(s) studied? If the research is being funded by a nongovernmental organization, or if the funding is slated for proprietary research, have I discussed with the funder the degree to which the research results will remain private or eventually become a part of the public domain? Have I satisfied myself that the research is proprietary and not secret (i.e., not ever intended for public distribution)? In the last phase, the anthropologist should ask him- or herself: Have I considered the potential use(s) to which my research will be put, and have I resolved, to my own satisfaction and through consultation with others, potential conflicts of interest between the client-funder and the people-culture researched? Using Data from Research The questions raised here are not simple ones, but if you ask and answer them satisfactorily before you begin your research, you will save yourself a great of anxiety and discomfort in the field. Such questions and answers might also be the proper doses of prevention that avoid disaster when communication and trust break down because openness and disclosure were not practiced. Most people, irrespective of cultural difference, operate in good faith when they have the necessary information to make informed choices. If you are open with informants, they have the right to say no to your requests for information.

14_107-Bernard.indb 139

6/9/14 10:02 AM

140

Carolyn Fluehr-Lobban

PROPRIETARY AND SECRET RESEARCH

Students and researchers may be confused by the difference between proprietary and secret research. This is an especially sensitive subject for anthropology since allegations that secret research had been conducted in Southeast Asia led to the crisis that produced the first AAA code of ethics. The issue reemerged in the controversies over applications of anthropology in the military in the war efforts in Iraq and Afghanistan. Proprietary research is normally negotiated between client and researcher for a specific project, length of time for research, and terms of publication. Proprietary research is normally not “owned” by the researcher but is research conducted for the client. Often, proprietary research is conducted for nongovernmental organizations, such as the World Bank, or nonprofit groups, such as Save the Children. Usually, access to published results of the research is not restricted. In secret research, openness and disclosure of research intentions, funding, and outcomes is not practiced with the people studied. Further, publication of research results is usually restricted and not available to the general public through usual avenues of distribution. Professional codes of ethics within the AAA, from 1971 to the present revisions, have stood firmly and consistently against anthropologists engaging in secret research of any kind. The present code says: “The researchers must have the intention and expectation to disseminate publicly results of the research within a reasonable period of time” (Draft Code 1995). RESEARCH REVIEW BOARDS

All universities and colleges are mandated by federal guidelines to have IRBs that are charged with the responsibility to review all research proposals that involve the conduct of research with human subjects.1 The guidelines were originally established for research in the biomedical fields, but anthropologists are also bound by these regulations. Some anthropological research may be evaluated as being of minimal risk to research participants, such as the observation of public behavior. Little anthropological research remains solely in the public domain. It often moves to a personal, closer relationship with human beings. Physical anthropology research may be classified as a type of biomedical research requiring informed consent. Archaeologists, though they work primarily with material culture, are nonetheless responsible to various “stakeholders,” who may be representatives of indigenous peoples, state or national historic commissions, and/or their sponsors, each of whom has a special and sometimes conflicting interest with the other. Ideally in this situation, or a like one for cultural research, the anthropologist can play a constructive role as broker, negotiating and achieving compromise among the various interested parties. There is no subfield of anthropology, or any part of the practice of anthropology, that is free of ethical responsibility. There has been some resistance by anthropologists to being called before review boards, having what we may view as our special field methods subjected to scrutiny. Perhaps because our research is comparative and most often conducted outside the United States, there has been a tendency to believe that the rules do not apply to us. A parallel tendency sees anthropological research as unique and therefore not related to

14_107-Bernard.indb 140

6/9/14 10:02 AM

4: Ethics

141

ideas or ethical practices that developed outside our discipline. This view mistakenly adopts an exceptionalist view of anthropological research as being somehow fundamentally different from other kinds of research. Again, research is research, and the rights of research participants do not vary, whether they are studied by psychologists, cardiologists, or anthropologists. Nor do the responsibilities of the researchers vary substantially. Indeed, today much anthropological research is conducted in an interdisciplinary environment where ethical standards must be uniform. Review boards themselves may vary considerably in their composition and practice. Some may take a strict or narrow interpretation of informed consent, requiring explanations of why anthropological research does not lend itself readily to the use of consent forms. IRBs may be sensitive or insensitive to issues in cross-cultural research, depending on the experiences of board members. Anthropologists would do well to consult with their IRB and perhaps open a dialogue with members about the nature and extent of federal and institutional regulation of their proposed research; an initial informal inquiry might make the formal request go more smoothly. The researcher might inquire how the research proposal would best address the expected standards of research practice while bringing up some of the particular cultural or linguistic issues that are usually part of anthropological research. A significant amount of anthropological research today is carried out in the United States, and federal guidelines apply because the anthropologists are recipients of federal funding. Researchers from outside the United States should acquaint themselves with comparable national or institutional review boards. My own experience as a member, chair, and grant recipient of my home institution’s Committee on Research with Human Participants is that review boards are user-friendly places that act more in an advisory capacity than as “courts of law,” where researchers are given a thumbs-up or -down approval or rejection. IRBs in large private or public research institutions can be overwhelmed with large numbers of proposals for review and the management of diverse projects from the biomedical fields to ethnographic research in education as well as anthropology. Usually, in my experience, a research design that is flawed or insufficiently addresses issues of ethics can be discussed, negotiated, redesigned, improved, and eventually approved. The dialogue is what is important; it is part of taking a proactive approach to ethics, not one based on the fear of alleged misconduct or wrongdoing. Don’t be shy or intimidated about educating members of IRBs, or their equivalents, about the nature of anthropological research and the methods we use, and do negotiate issues or areas of possible disagreement with committee members. LOW-RISK, UNOBTRUSIVE RESEARCH

Much anthropological or social scientific research may appear to be of low or minimal risk, with insignificant or no potential harm to participants. As such, federal regulations and institutional review may be exempt in cases of low or no risk to those studied. The Federal Policy for the Protection of Human Subjects, known as the Common Rule (www.hhs.gov/ohrp/humansubjects), adopted the view that consent need not be obtained for research characterized by the following conditions:

14_107-Bernard.indb 141

6/9/14 10:02 AM

142

Carolyn Fluehr-Lobban

1. the observation of behavior in public places where questions of privacy do not exist; 2. review of publicly available information, including personal identity information; 3. research using low-risk methods, such as questionnaires, interviews, or tests in which agreement to participate effectively constitutes consent.

The first two might be considered unobtrusive research, where no observable risk to those studied can be shown. In the last case, however, anthropologists should not exempt themselves from practicing full disclosure simply because they are using low-risk methods, such as using questionnaires or interviews. In administering tests, the potentially coercive relationship between test-giver and test-taker should be recognized as one where a power differential exists. Using students to fill out questionnaires within the classroom setting, for example, has a coercive dimension, and anthropologists should instead look for venues for research outside of the classroom. INFORMED CONSENT

Informed consent has become a virtual canon of ethics and research in all fields, whether the research is biomedical, behavioral, or social scientific (see Fluehr-Lobban 1994). The legal doctrine of informed consent grew out of the post–World War II Western world that was shaken to its moral and ethical core, not only by the revelation of Nazi atrocities but by the collaboration of science with immoral and unethical practices. Originally developed for medical research, informed consent expanded to include research with humans across the scientific–social scientific spectrum, with psychology having provided the historic bridge between the two. In 1995, the principle of informed consent was added to the new code of ethics of the American Anthropological Association (AAA 2009). Few anthropologists know this history or appreciate its relevance to their own research. Some may view informed consent in a rather mechanistic fashion because, frankly, in medical research, much of it has devolved to signing a form for legal purposes. This is unfortunate because the genesis of informed consent and its guiding spirit is that of openness and disclosure in research practice. By extension, in social science research, using the spirit of informed consent means that the researcher discusses the methods and likely anticipated research outcomes with the participants; thus, the researcher and the studied develop an open relationship. Ideally, informed consent opens up a two-way channel of communication that, once opened, allows for a continuous flow of information and ideas. This is the spirit of informed consent, rather than the mechanistic application of a form designed more to protect the researcher than the research participant. In Human Organization (Fluehr-Lobban 1994), I argued for the acceptance of the spirit of informed consent, combined with behavior that is appropriate to our methods and our traditionally close relationship with the peoples we research. For purposes of federal regulation and enforcement, informed consent has been defined as “the knowing consent of an individual, or a legally authorized representative, able to exercise free power of choice without undue inducement or any element of force, fraud, deceit, duress, or other form of constraint or coercion” (Protection of Human Subjects 1990). For federally regulated agencies that sponsor research, the documentation of informed consent is demonstrated in one of the following ways: (1) written consent

14_107-Bernard.indb 142

6/9/14 10:02 AM

4: Ethics

143

document, to be signed by the subject or authorized agent; (2) a short form summary document ensuring that informed consent requirements have been met; and (3) modification of 1 and 2, where minimal risk is demonstrated or where obtaining informed consent would invalidate the objectives of research. In the latter case, the responsibility of the external review committee or IRB increases proportionately to the diminished informed consent requirement. The 1998 code of the AAA was the first to make reference to informed consent: Anthropological researchers must obtain in advance the informed consent of persons being studied, providing information, owning or controlling access to material being studied or otherwise identified as having interests which might be impacted by the research. It is understood that the degree and breadth of informed consent will depend upon the nature of the project. . . . Further it is understood that the informed consent process may be dynamic. . . . Informed consent does not necessarily imply or require a particular written or signed form. It is the quality of the consent, not the format, that is relevant.

Much anthropological research has not been subjected to federal regulation because it is small scale (i.e., relatively few anthropologists have been funded by the National Science Foundation or National Institute of Health). Also, as noted, many anthropologists view their own research as relatively benign, posing little risk to research informants, and perhaps they believe they can protect their subjects from harm. However, much contemporary anthropological research is occurring in complex areas of applied social science and in public sector, community-based projects with protected Americans and indigenous groups considered vulnerable, such as certain South American Indians or my own recent research with persons in Sudanese Internally Displaced camps (IDPs). Or the anthropological researcher is often a part of a multidisciplinary team working in the United States or in development-related projects overseas. Questions of the universality of informed consent have been posed. Informed consent is now enshrined in American anthropology, as it is in all research involving humans in the United States. European social anthropologists may have been reluctant to apply informed consent as an ethical or legal concept because of its perceived specific connection to the United States. However, the intent of informed consent—to be open and clear about the goals and funding for research and to obtain the consent of those studied—has become more universally recognized as a significant ethical principle that transcends nation, culture, and language. Working with this history, we may ask whether non-American research participants are entitled to the same protections afforded by informed consent as are Americans. I hope that the answer would be that humans, irrespective of culture or nation, are entitled to the same rights and protections, but I fear that there may be, in practice, multiple double standards. Does informed consent translate across languages and cultures? Arguments against using informed consent in anthropology have drawn on the cultural relativist theme. Some have argued that informed consent is a U.S. or Western concept that cannot be fully explained in cross-cultural research. Honesty about one’s research methods, goals, and sources of funding, however, is not a Western concept. Although cultures vary in their modes of communication, no culture endorses dishonesty or deception.

14_107-Bernard.indb 143

6/9/14 10:02 AM

144

Carolyn Fluehr-Lobban

Experienced anthropological researchers know of the benefits of an open, mutual relationship with research participants. Some even think in terms of a “covenantal” relationship that anthropologists have with their research collaborators (Wax 1995). I draw on my own seven years of anthropological research, spanning nearly four decades and three different African countries, to inform this discussion. I began my doctoral research when the first code of anthropological ethics was promulgated, and those heated debates among anthropologists over the politics of research were my training ground. In the Sudan in 1970, people were suspicious as to why two American researchers had come all that way to learn Arabic and study their culture. The CIA was mentioned as the agent behind us, rather than an American university fellowship, and my husband (a fellow anthropologist) and I felt awkward and untrusted. The only antidote to this public perception was honesty, openness, and full disclosure of who we were and what we intended to do during our 18 months in the country. During subsequent research in the Sudan (1979–1980), Egypt (1982–1984), and Tunisia (1990), we developed greater familiarity with the language, with research participants, and cultural sensibilities about research and privacy issues. I came to understand that being open about research is a way to keep open the lines of communication throughout the course of research, to allow negotiation of terms of research as it progresses through its various stages. Informed consent does not translate literally and directly into Arabic, or I suspect, many other languages. But the spirit of informed consent, making honest attempts to talk about the nature, duration, and research funding and allowing a relationship to unfold that may even permit a participant’s withdrawal from research, is more ethical and probably results in better research outcomes. In my fieldwork, once people understood the nature of my inquiry into some admittedly sensitive family law cases and issues, some asked that certain information not be used or published. Sometimes, they withheld information, understanding that certain individuals or families might be compromised by public knowledge of their personal conflicts. In retrospect, this selective review of my fieldwork experiences is not without its own dose of self-criticism. I am acutely aware that higher-class families were more sensitive about protecting their respectability and privacy than were lower-class, less well-educated families, and I may have used a double standard in this respect. In another vein, I became increasingly cognizant of highly sensitive research subject matter, such as the rise of Islamist movements in Egypt, Sudan, and Tunisia during later stages of my research career. I struggled with the ethical dilemma of being open with national research boards, disguising my interest in Islamism with other research objectives or resolving the dilemma by not conducting research. Between 2005 and 2009, I resumed research in Sudan as it was on the verge of separation into two countries and there were entirely new challenges of research. After the Islamist regime of Omar al-Bashir (1989–present) imprisoned and tortured all opponents, including university professors and human rights activists, academic researchers from the West stopped doing research, amounting to an informal boycott. This continued until 2004–2005, when the regime began peace talks to end the 22 years of civil war between north and south, leading eventually to separation of the country into two countries.

14_107-Bernard.indb 144

6/9/14 10:02 AM

4: Ethics

145

Although the Darfur crisis was mounting, the promise of peace created a more favorable environment for the return of researchers. I returned, with a handful of others, in 2005, funded by two European university grants and the European Union, and subsequently in 2007–2009 by the U.S. Institute of Peace. Due to the latter’s U.S. government funding, I was required to comply with U.S. sanctions against Sudan through the Office of Foreign Assets Control. Some of my colleagues criticized my return as well as my openness to the Sudanese media, who were understandably interested in an Arabic-speaking “expert” on their country who might provide insight on their isolation from the world. If there were ever a place that anthropologists would be advised to avoid researching, it is Sudan. My study involved the sensitive subject of the withdrawal of Shari’a law after a historic peace agreement from the insurgent, non-Muslim south and from southerners in the Muslim north. Common sense reveals that good people live under bad governance, and the existential question of what is to be gained by withdrawing from research must be satisfactorily answered. An unwillingness to document and analyze current reality goes against the fundamental scientific principle of knowledge acquisition, leaving, for me, only ethical and moral reservations to be weighed. The rules of engagement that I developed grew out of both ethical consciousness and pragmatism. They were also discussed intensively with my home institution IRB and a research protocol to which I adhered was prepared. In this environment of research, I accept the potential restrictions on my freedom to observe events and meet with individuals in the society. I have been granted visas, which are by no means automatic, to conduct research by successive military governments. I am always transparent about my research project, its sponsors, and intended outcome(s) and have often decided not to pursue a line of research or was refused permission to carry out research as a direct result. In practical terms, this meant being turned away from observing certain Shari’a courts while being given full access to others and being granted permission to conduct research in IDP camps and to interview victims of Shari’a harassment and punishments. In the field, I observed, but never participated in, public demonstrations or protests against the lack of freedoms in the military regimes. I do, of course, write about these. I “respect” the reality of the general lack of freedoms of speech, assembly, and so on and the potential harm it represents for the participants who choose to collaborate in my research. Vulnerable populations (e.g., human rights victims) are of particular concern, although the most vulnerable with whom I worked in the IDP camps wanted their identities to be known and recorded, expressing both defiance and cynicism after years in the camps. I work almost completely in Arabic and two of my books have been translated into Arabic and subjected to local scrutiny and debate. I strive to be objective in my public interviews and commentaries. Openness and negotiation of the known risks and benefits of research with informed individuals within an otherwise restricted environment has been my modus operandi. Outside of Sudan I freely criticize, for the public record, the lack of freedoms and restrictions of human rights for citizens of Sudan. I am frequently an expert witness in political asylum cases where the conditions of abuse under chronic militarism are at issue.

14_107-Bernard.indb 145

6/9/14 10:02 AM

146

Carolyn Fluehr-Lobban

The ethical and moral ambiguity of a case like that of Sudan is not atypical. The post–cold war, multi-polar world of the twenty-first century presents international researchers with a global environment not only of irregular wars but also a rapidly shifting balance of power. Such hard cases need exploration, debate, and clarity of common ground and differences, and that requires dialogue. For my generation of anthropologists trained in the late 1960s and 1970s, and for those who preceded awareness of the importance of theoretical and practical ethics, the training we received in graduate school was inadequate to carry out informed and balanced ethical decision making. With the shift from an adjudicative code to an educational one, education in professional ethics and active discussion of ethical dilemmas has improved but is still not an integral part of undergraduate and graduate education in anthropology. INFORMED CONSENT WITHOUT FORMS: ACKNOWLEDGING THE RISK OF PATERNALISM

Almost invariably, when the subject of informed consent is brought up with anthropologists, the first objection they raise is the use of a consent form. However, informed consent does not require forms. Indeed, much of anthropological research, utilizing methods of participant observation, would mitigate against using forms. But researchers should be aware that they may be required by their funding agency or home institution to use informed consent forms. Most, if not all, medical–biological research that interfaces with physical or cultural anthropology requires informed consent forms. Informed consent, when mechanically applied using a form or some verbal formula, may become more of a protection for the researcher than the researched. Informed consent obtained in this way is unilateral rather than bilateral and protects the researcher against charges from participants that they did not understand fully the intent or outcome of research. In a research relationship with non-Western, often relatively powerless participants, the Western researcher may feel awkward asking for a signature on a consent form. The person being studied may not be literate or the official appearance of the form may be intimidating. Our “subjects” are not necessarily passive. The value of conducting research in an open, collaborative manner is that informed consent becomes a natural part of the development of the research project and relationship with those you study. You can obtain informed consent without using forms by raising relevant issues that inform and thereby empower the participant. It is true that anthropological research can be highly personal, with intimate relations of confidence and trust developing as a result of long-term residence with people. It is also true that research is often conducted in the local language, which conveys a greater intimacy than would occur in translation. That intimacy is a powerful instrument that must be used with care so as not to violate the trust established nor abuse the confidence that has been given. An unconscious or unspoken paternalism/maternalism may have kept some of these issues from being fully aired in anthropological discourse. This is what used to be called the “My Tribe” syndrome, in which anthropologists might believe that they did what was best for “their” people. In social or cultural anthropology, under historical

14_107-Bernard.indb 146

6/9/14 10:02 AM

4: Ethics

147

conditions of actual colonialism or perceived colonial-like agencies like the American Bureau of Indian Affairs, paternalism was not uncommon and may have characterized human relations between anthropologist and subject (Asad 1973, 16). Indigenous peoples of America, indeed throughout the postcolonial world, are actively restructuring their relationships with states, external institutions, and foreign researchers. Most U.S. and Canadian tribal councils have autonomous research review boards to which anthropologists and other researchers must apply before receiving approval to carry out their projects. Typically, through the process of requesting permission to conduct research, the terms and conditions of the research plan are negotiated; in effect, this is an official airing of the proposal where the required openness and disclosure amount to informed consent. Controversy over ethical and legal issues arose during the Human Genome Diversity Project (HGDP) centered on the potential abuse of a relationship between powerful Western scientific bodies and relatively powerless indigenous populations whose DNA had become invaluable. It may have been that proper communication amounting to informed consent was lacking in the celebrated cases where “theft” or appropriation of human genetic material was alleged. As the HGDP progressed, it engaged in more open dialogue as to the low risks and greater benefits of the research with the research participants. This is a clear and positive trend in anthropological research—one that is increasingly recognized as constituting not only better ethics in research but as producing better research results. Much of the history of informed consent in the West has been linked to protecting the rights of individuals. This reflects the norms of American life. Anthropologists and non-Western scholars have been the leaders in pointing out the greater value placed on collective rights in non-Western societies. This should not present an insurmountable obstacle to obtaining informed consent, for negotiations with representative and responsible agents of tribal and ethnic groups can be combined with individual consent. In some cases, where large community-based studies are proposed and negotiated, public meetings of potential participants are scheduled, during which opposing viewpoints can be openly expressed. These can and should be regularized as an ongoing part of research projects, ensuring active community involvement and monitoring. Some researchers may feel uncomfortable with the degree of openness discussed here, but retreating from openness results in some form of deception. This is why anthropologists must carefully consider the ethical implications of every phase of their research project. I have had two purposes in this rather lengthy discussion of informed consent. First, the application of the ethical and legal principle of informed consent has been a relatively recent development in anthropological research that is in conformity with developments in sociocultural research as a whole. Therefore, the importance and utility of informed consent may need some explanation and justification. Second, the spirit of informed consent has potency as a summary concept for research, irrespective of discipline or subdiscipline. Informed consent, in its fullest interpretation, means openness and disclosure with participants, and models of research that are collaborative, rather than hierarchical and, relatively, nonparticipatory. This may be reflected in a detectable change in terminology of social research, with fewer references to informants and

14_107-Bernard.indb 147

6/9/14 10:02 AM

148

Carolyn Fluehr-Lobban

subjects and more reference to “collaborators” and “participants.” When the spirit of informed consent is implemented, it results in better researchers and better research. COLLABORATIVE RESEARCH

Since 1991, I have argued that collaborative research is “ethically conscious” research (1991b, 2003). Not only is collaborative research ethical and thus morally preferable to historical models of research, it is better research because of its methodology emphasizing multiple, polyphonic perspectives, leaving a richer heritage of ethnography to subsequent generations of ethically conscious researchers. Collaborative research involves the people studied in an active way, as individuals or groups having vested interests in the project through their participation in research design, execution, publication, and outcomes potentially related to community or individual improvement of well-being. Collaborative studies can potentially inform or affect social policy. Often, jointly directed and jointly authored projects replace the older, more hierarchical model of research planned, executed, and published by the anthropologist alone. Community or individual collaboration in research—with partnership incorporated in every phase of research—becomes a condition for its success, not simply a fortuitous by-product of work with communities. This newer model of research presumes, for the most part, a literate, socially conscious set of partners who not only participate in research but read and critique drafts of publishable results. However, literacy among research participants is not essential to its viability or success, as openness and mutual exchange of research ideas and outcomes can be communicated without the ability to read or sign informed consent forms. Collaborative research stands in dramatic contrast with historical models of Boasian anthropology—with its general emphasis on informants, ethnographic subjects, and a central objective of data collection—and with European social anthropology methods with its linkages to colonialism and latter-day postmodernism. Uniquely, the subfields of applied anthropology in the United States and development anthropology in Europe have recognized and embraced the value of collaboration in research as it is necessarily attached to applications of anthropology to institutions and agencies—governmental and nongovernmental—whose mission is to promote the well-being of humans. It is this mission of research designed to promote human wellbeing that has led critics to view collaborative research as advocacy, confusing anthropology with social work (Gross and Plattner 2002). Others view collaborative research through the lens of the same history of anthropological research and would argue that the approach reflects an increasing decolonization of the discipline. CONCLUDING REMARKS

The preamble to the new AAA code of ethics (AAA 2009) states that the anthropological researcher, scholar, or teacher is a member of many different communities, each with its own set of contextual obligations. He or she is, for example, a member of a family and of a community, is a public or private employee, has many simultaneous statuses, and plays many simultaneous roles. Anthropological research can place the researcher/scholar in complex situations, where competing but legitimate ethical claims can arise.

14_107-Bernard.indb 148

6/9/14 10:02 AM

4: Ethics

149

The informed and ethically conscious researcher will recognize the multiple layers of responsibility that can obtain in research. These might include ethical responsibility to the people studied in their complex relations with one another and to their state and other communities and institutions that impact their lives; responsibility to those into whose confidence you were taken; responsibility to the truth, science, and one’s discipline; and responsibility to the client or funder of research. In the context of this complex assessment, the anthropologist must be guided by the following general ethical principles of conduct: 1. To do no harm or wrong, understanding that the development of knowledge can lead to change that may be positive or negative for some people. Weighing the kinds, degrees, duration, and probability of goods to be gained and harms to be avoided is the task of the informed and ethically conscious researcher. 2. Avoiding deception or knowing misrepresentation of one’s research goals, methods, or funding with the persons/communities studied is a sine qua none of social research. This extends, naturally, to any fabrication of evidence, falsification of data, or plagiarism in the processing and publication of the results of research. 3. Acting impartially, such that all persons affected by our research are treated in the same manner, in so far as this is possible under the conditions of research.

The most recent revision of the AAA Code of Ethics adopts general principles that should guide all research (see www.aanet.org, code of ethics). Studying these principles, and discussing their application in various research settings, should be part of undergraduate and graduate education and training in anthropology, and reference to them should be a part of every research proposal. Developing awareness of the fundamental importance of ethics as a key component of professionalism and as a necessary adjunct to science and research is an essential task of the next generation of anthropologists. Ethics education, and more importantly, constructive and engaged dialogue about the ethics of social research, should be more central to our work and to our profession than it is at the moment. Anthropologists, irrespective of national origin or country of research, face common challenges and similar responsibilities when they conduct research in our evershrinking global context. Whether as academic researchers, development workers in governmental or nongovernmental agencies, or applied anthropologists, researchers need to consider their ethical choices and professional responsibilities in ever more complex social environments. AIDS-related studies in cross-cultural research, participation in development projects, and the study of ethnic minorities in nation-states are just a few examples of contemporary research projects that raise ethical dilemmas and require dialogue and negotiation of the terms and conditions of research within funding agencies and the nation-states where the research is to be carried out. An international discussion of cases and comparison of research experiences would be a welcome addition to broadening the discourse about ethics in anthropology globally. NOTE

1. Historically, research participants have been referred to as “subjects” in biomedical and psychological research. Anthropologists–sociologists have used the term “informants.” However,

14_107-Bernard.indb 149

6/9/14 10:02 AM

150

Carolyn Fluehr-Lobban

“participants” or “collaborators” is gradually replacing subjects and informants as more egalitarian ideals of the relationship between researcher and researched are evolving. REFERENCES

American Anthropological Association (AAA). 1971. American Anthropological Association statement on ethics. Principles of Professional Responsibility, Adopted by the Council of the American Anthropological Association, May, Washington, DC. American Anthropological Association Code of Ethics. 1998. www.aaanet.org/committee/ ethicscode/htm (accessed March 6, 2013). American Anthropological Association Code of Ethics. 2009. www.aaanet.org/committee/ ethicscode/htm (accessed March 6, 2013). Asad, T., ed. 1973. Anthropology and the colonial encounter. Atlantic Highlands, NJ: Humanities Press. Boas, F. 1919. Correspondence: Scientists as spies. The Nation 109: 279. Draft Code. 1995. Final report of the commission to review the AAA statements on ethics; Draft AAA Code of Ethics. Anthropology Newsletter (April): 13–16. Fluehr-Lobban, C. 1967. Statement on problems of anthropological research and ethics. Reprinted in Ethics and the profession of anthropology: Dialogue for a new era. 1991. Philadelphia: University of Press. Fluehr-Lobban, C. 1991a. Professional ethics and anthropology: Tensions between its academic and applied branches. Business and Professional Ethics 10: 57–68. Fluehr-Lobban, C., ed. 1991b. Ethics and the profession of anthropology: Dialogue for a new era. Philadelphia: University of Pennsylvania Press. Fluehr-Lobban, C. 1994. Informed consent in anthropological research: We are not exempt. Human Organization 53: 1–10. Fluehr-Lobban, C. 2000. How anthropologists should respond to an ethical crisis. The Chronicle of Higher Education, October 6, B24. Fluehr-Lobban, C. 2009. Guiding principles over detailed codes. Anthropology News (Sept.): 8–9. Gross, D., and S. Plattner. 2002. Anthropology as social work: Collaborative models of anthropological research. Anthropology News, November: 4. Jordan, A. 1996. Review of the AAA Code of Ethics. Anthropology Newsletter (April): 17–18. Levy, J. 1994. Report to the American Anthropological Association. Anthropology Newsletter (February): 1–5. Protection of Human Subjects, Code of Federal Regulations. 1990. www.hhs.gov/regulations (accessed March 6, 2012). Rohde, D. 2007. Army enlists anthropology in war zones. New York Times (Oct. 5). http://query .nytimes.com/gst/fullpage.html?res=9d04e3d81130f936a35753c1a9619c8b63&pagewanted= all (accessed February 29, 2012). Tierney, P. 2000. Darkness in El Dorado: How scientists and journalists devastated the Amazon. New York: W.W. Norton. Wax, M. 1995. Informed consent in applied research: A comment. Human Organization 54: 330–331.

14_107-Bernard.indb 150

6/9/14 10:02 AM

CHAPTER 5

Feminist Methods Christine Ward Gailey

Anthropology as a field lends itself to feminist methods. First, at least in its North American version, because of the core concern with cultural dynamics and shifting terrains of meaning, it flourishes on the borderland between the social sciences and humanities. This interdisciplinary space is one valued in feminist research in general. Second, anthropology is comparative and historical in scope, permitting those who wish to seek patterns to do so while remaining skeptical of unified theories based on nationalist assumptions or images of human nature rooted in particular societies. Third, there is a historical tendency in anthropology to focus on the local, the everyday, or the marginal. At its worst, anthropology can ignore global structures of meaning or transform peoples into exotic objects for consumption by more powerful audiences. At its best, it becomes the witness for human practices that resonate with, but are not mere reflexes of, the processes of global capitalism or earlier colonialism, worlds of meaning in the making. Anthropology can value the local, the subordinated, and the creation of meaning among those who are not powerful in the reductive sense of wealth or political control or even social prestige. Perhaps it is this tendency that contributed to the formative influence that anthropologists had in the early 1970s in the development of what would become international or transnational feminisms by the twenty-first century. Taken narrowly as techniques of investigation, the multiple methods of feminist anthropology—its use of participant observation, interviews, surveys, and quantitative data development—are not distinct from those of other critical methods in anthropology. Even the coupling of theory and method found in this chapter, although typical of feminist methods in anthropology, can be found in other forms of critical anthropology: Theory informs method and method shapes theory. What is distinctive is how theory and method are related: The bridges between the two are ontology (the set of assumptions that underlies one’s research design) and epistemology (how one knows what one thinks one knows, the basis for knowledge claims). Feminist anthropology, then, is distinguished by the ontological and epistemological dimensions of method. We shall see that there is no one feminist method in anthropology or, for that matter, in any other discipline. Instead, epistemological and ontological debates emerge within feminism. These include, in various forms, what we want to know, what we can assume, who “we” are, and the implications of our research for social engagement. There are disparate positions taken in each of these debates, and periodizing feminist methods depends on their contours. 151

14_107-Bernard.indb 151

6/9/14 10:02 AM

152

Christine Ward Gailey

FEMINIST ONTOLOGY

The bedrock assumption in feminist research is that gender is a dimension of societies and cultures that is key to understanding human relations. Correlatively, studies that ignore gender presume that gender always means men and women, assume gender is always a fixed identity versus a performance or process, view gender as always isomorphic with sex differences, or subsume it without investigation into family life or sex roles do serious disservice to the production of knowledge about social and cultural dynamics. Feminists disagree whether gender is always constructed as a hierarchy but agree that differential social power or authority usually frame gender scripts. Feminists in anthropology share a concern that power inherently shapes the kinds of understanding people from stratified societies develop of our own and other societies and cultures. But feminists differ in how we conceptualize power and whether we see it as grounded in social relations and structures or as more amorphously associated with conceptual schemas and language practices. Some of us see these as dialectically related. Despite such theoretical differences, feminist anthropologists concur that: (1) power differences shape the lived experiences of people in most if not all societies as well as the meanings that people attribute to those experiences; (2) in state societies, gender is one of several intersecting variables that together produce and embody differences in power among sets of people; and (3) gender is a major organizing principle that shapes the reproduction of culture and, consequently, the transformation of cultural practices. Another distinctive aspect of feminist anthropology is the vantage point taken in inquiry. Sometimes gender is the lens through which the researcher views a society, but often the viewpoint constructed is that of a woman or some category of women (e.g., middle-class women, professional women, battered women, etc.). This perspective is necessarily partial and doesn’t claim to encompass all possible perspectives on a particular society or cultural setting. Instead, it derives from a methodological starting point Cynthia Enloe has called “feminist curiosity”: posing questions about the gendering of sociocultural, political, and economic processes and their gendered consequences that (1) cease taking these complex articulations for granted and (2) consider the gendered effects or dynamics “to deserve public attention and public resolution” (Enloe 2007, 9). Such concerns mark feminist anthropology as eschewing political neutrality: It demands a stance or standpoint (see Harding 1987, 1991; see also Wylie 2004). Feminist anthropology shares with other critical methodologies in anthropology the position that objectivity is impossible because we are cultured beings. The position argues further that claims of objectivity in any set of research approaches or policies disguise political claims or, more often, support the status quo. Instead, researchers acknowledge their perspectives and scholarly traditions and argue from there, making audiences aware of the standpoint taken. Feminist writers take seriously the obligation to let our audiences know the basis for our claims. As such, abandoning claims to objectivity helps one avoid bias in research, for bias differs from standpoint or situating oneself. A biased researcher makes assumptions that remain implicit or, in many cases, are so deeply held that even the researcher does not recognize them as presumptions. In feminist methodology, researchers reflect on and examine their set of assumptions and convey these to audi-

14_107-Bernard.indb 152

6/9/14 10:02 AM

5: Feminist Methods

153

ences, thus avoiding bias. Such self-reflection and intellectual forthrightness help the audience situate and evaluate claims within a wider field of argument. Cross-checking for bias, for feminist anthropologists, lies in reflexivity, in scrutinizing the links that we make or do not make among our assumptions, why and how we design our projects and conduct research, and the conclusions we draw. Acknowledging that inquiry, including scientific inquiry, takes place in a social world and is conducted by people embedded in cultural frameworks that shape their thinking should not translate into a rejection of quantitative techniques or an assumption that somehow these are inherently masculinist or antifeminist (on feminist quantitative research, see Reinharz 1992). Indeed, techniques can be viewed as neutral, and our goal may be to reduce bias by making our assumptions and epistemological claims clear to our audiences. Some feminist anthropologists are empiricists—building theory through what can be observed—and some are committed to the goals of positivist science. Within feminist inquiry in anthropology, the claims of positivism demand scrutiny and gender analysis, as do the ontological and epistemological claims of other traditions (functionalist, structuralist, poststructuralist, Marxist, postmodernist): What is the context for these claims? Who is the audience? How is expertise established and for whom? What counts as evidence? What is truth in this framework and how does the author build a case? Most notably, how does this understanding further feminist practice? Feminist researchers argue that research should inform our own social engagement. Feminist research, as distinct from gender research in general, seeks to link theoretical understandings with practical activity, human agency as a matter of course. The knowledge we produce regarding gender configurations and women’s struggles in whatever contexts should, routinely, guide our own efforts, and our own everyday lives (see Lamphere et al. 1997). Feminist ontology thus produces a reflexive methodology, one that considers the researcher’s own social engagement as implicated in and by the research, and the researcher accountable to the people who make the research possible. These are not “research subjects” or “informants” who the researcher studies but knowers in their own right and facilitators and co-producers of the research. Another assumption made by almost all feminist anthropologists is that, as a differentiation process and in most cases a hierarchical one, gender is historically and socially constructed. With the exception of those sociobiologists who ascribe gender to non-symboling primate species, feminist anthropologists argue that gender is distinct from sex differences. Gender categories may be isomorphic with the categories of sex difference in particular societies, but neither the cultural script of “doing gender” nor the structures of gender hierarchy can be reduced to female and male roles in biological reproduction. The ethnohistorical, archaeological, and ethnographic records amply demonstrate that multiple genders exist (see, e.g., Davies 2007). Moreover, recent feminist research on intersex and human embryonic sex differentiation calls the rigid definitions of “female” and “male” as biological distinctions into question (Bardoni et al. 1994; Fausto-Sterling 1993). On the basis of many studies from a range of subfields and disciplines, feminist anthropologists assume that gender varies culturally and across time periods and social strata. Indeed, one of the hallmarks of feminist theory in the twenty-first century is a

14_107-Bernard.indb 153

6/9/14 10:02 AM

154

Christine Ward Gailey

mistrust of the assertion that there is a concept of “woman” apart from or independent of national and cultural agendas (see Siapno 2002). Gender ideologies and hierarchies are cross-cut by other hierarchies and ideologies. The intersection of these hierarchies, whether mutually constraining or synergistic, produces different experiences of gender and other dimensions of identity—age, ethnicity, class, race, sexual orientation, religion—that may operate in a society and, at the same time, provide complex grounds for acting, or standpoints (see Hafez 2012). Listening to the words and observing the actions of people whose everyday lives are riven by conflicting stresses of race, gender, and class help us dismantle widely held notions that there is an essential nature to women or men (see Stack 1997). The call for integrating feminism into one’s relations in daily life also distinguishes feminism from other critical methodologies. As Kaufman (1996, 165) writes, feminist methodology is “not just a reconceptualization of a way of ‘doing research,’ but also a way of living in the world. It openly advocates a decentered nonauthoritarian approach to all human relationships and to thinking about those relationships.” Other critical methodologies don’t insist on such efforts. Typically, calls for action and engagement are at the public domain level—addressing national or international policies, clearly defining one’s research agenda to one’s informants, or helping shape the policies of one’s professional organizations, universities, or local communities. The activist aspect usually stops at arenas deemed private, particularly the researcher’s family and personal relationships of colleagues. The feminist rubric that “the personal is political” disallows such compartmentalization. The problem, always, is setting priorities. More than other critical methodologies, then, feminist anthropology makes research and the researcher—no matter how elaborately theorized the research or how seemingly abstract or micro level the concerns of the researcher herself—accountable to a range of audiences. The audiences to whom the researcher is accountable include not only colleagues in the social sciences and the humanities but also the subjects of the research, those at the receiving end of policies, and the researcher’s family and community members. The connecting thread in this set of audiences and engagements is the call to help reduce gender inequities wherever these occur and in collaboration with local initiatives. Contradictions become acute when someone calling herself a feminist anthropologist investigates the intricacies of how gender hierarchies operate in a particular society while expecting a pregnant or new mother colleague to carry on as if the new state of being had no effect on her energy level or overall workload. It isn’t sufficient to investigate the relationship between gender and sexuality while reproducing gender stereotypes or practicing homophobia or dismissal of emerging categories of sexualities in one’s own milieu. It isn’t enough to teach about gender difference while ignoring student complaints about sexual harassment by one’s departmental colleague or shrugging off a teacher’s gender stereotyping in the classroom because your son benefits from it in terms of attention time. The creative discomfort or tension that results from the adoption of such responsibility is part of one’s state of being, because ethics are intrinsic to the methodology (Jaggar 1991). With this outline of ethical, practical, and theoretical issues, I now turn to epistemological dimensions of feminist methodologies in anthropology.

14_107-Bernard.indb 154

6/9/14 10:02 AM

5: Feminist Methods

155

FEMINIST EPISTEMOLOGIES

How do feminist researchers know what they know? On what basis do feminist scholars make knowledge claims? What are data in feminist research? What is the relation of observed reality to evidence? What constitutes proof? There are debates within feminism regarding each of these issues. These debates provide a wider latitude for what is considered feminist research than detractors might think. There is no orthodoxy, but there are types of feminist research that mark periods in feminist scholarship from the 1970s to the present. These can be characterized as: gynocentrism, standpoint theory, and intersectional analysis. They represent different degrees, depths, and facets of an overarching feminist critique of science, even as many feminist researchers remain committed to scientific inquiry (Harding 1991).

Beginnings: How Does Focusing on Women Change Anthropological Inquiry? Feminist empiricism, or gynocentrism, was most prevalent in the earliest phase of feminist anthropology, roughly 1970–1980. Whatever the disagreements among feminists in this period, most shared an underlying commitment to modernist orientations, and most feminist social scientists, because of their training, had a commitment to positivism. The first step in the process of creating a feminist anthropology was uncovering male bias in the discipline. How did the central questions of anthropology embody assumptions that what men did was more important than what women did? What difference does it make to have research done by women? How were women’s sources of authority, dignity, and value ignored or denied in the way research was conducted (see Rapp [Reiter] 1975; Rosaldo and Lamphere 1974)? Later, feminist scholars dubbed this the “add women and mix” period (Harding 1987), but placing women at the center of analysis—or just including them—led to posing new research questions, research approaches, and the assessment of women’s status and roles viewed comparatively and historically. In this form of feminist research, the researcher first reviews existing literature on the topic, being attentive to indications of masculinist bias in assumptions made, research design, and development of evidence, data analysis, and conclusions. Retrieval of earlier studies, ignored or merely forgotten in the academy, is necessary in this phase. Contributions by earlier generations of women anthropologists or ethnographers require reexamination and acknowledgment even in an academy that seems to credit the newest wrinkle, the ahistorical claims of each new generation as indicative of innovation (see Gacs et al. 1988; Golde 1970; Lurie 1966). Appreciating the efforts made by pioneers gives us a better sense of why feminist anthropology came to be shaped the way it is. Moreover, situating one’s arguments in a kind of contextual lineage across national boundaries is more in keeping with anthropology as a field that claims to value difference. Anthropological literature in and before the 1950s presented marriage as the central institution of kinship, and references to women were clustered in kinship studies. So we can expect that feminist critiques of anthropology would begin by integrating women’s work and sources of authority through kin connections into the analysis, as Eleanor Leacock’s (1954) research on the Innu of Labrador did, and, simultaneously, by reconsidering how marriage was defined. Kathleen Gough’s (1959) research on the

14_107-Bernard.indb 155

6/9/14 10:02 AM

156

Christine Ward Gailey

Nayars of Kerala, India, challenged the ways that anthropologists had defined marriage. Although not read as feminist at the time, this early work would inspire later, more explicitly feminist research. The next step was to design research that would correct these distortions, usually within the same theoretical tradition. Researchers corrected the biases inherent in: (1) selecting only male informants; (2) assuming that whatever men did was more important that what women did; (3) analytically demeaning the contributions of women; (4) discounting research topics that would be associated with women in the researcher’s own culture; and (5) assuming that women did the same things as men. The basic questions most research designs posed were: “What does this topic look like when women are included?” and “How do portrayals of this society change when women are the focus or are included in the analysis?” Based on these early investigations, questions emerged regarding how one establishes what women’s status is; whether women were universally subordinated; whether sex differences were the same as gender differences between people labeled “women” and those labeled “men” in various societies; why women would have more or less power in different types of societies; and so on. Approaches to feminist anthropological research were diverse, following major theoretical orientations in the discipline among functionalists, structuralists, and Marxists. These approaches can be seen in early anthologies, which stimulated cross-cultural analyses of gender roles and of women’s authority patterns through time and place (see Jacobs 1971; Paulmé 1971; [Reiter] Rapp 1975; Rosaldo and Lamphere 1974; Schlegel 1977). Alongside the anthologies came a series of theoretical and ethnographic studies, such as Leacock’s commentaries (1972) and Ernestine Friedl’s (1975) and Peggy Sanday’s (1981) comparative research, among others. By the mid-1970s, a series of review essays examined the proliferation of research on women within anthropology (Lamphere 1977; Quinn 1977; Rapp [Reiter] 1978). By focusing on what women were doing, and designing research that required observing and talking with women as well as with men (Strathern 1972), researchers provided insights into cultural dynamics that had been assumed before to exclude women from important social decision-making. In a series of historical and contemporary case studies, researchers identified diverse sources of authority available to women at various stages in their life, arenas where some women could become authoritative, and ways in which various societies or processes such as capitalist colonialism accorded social recognition or erasure of contributions made by woman (see, e.g., Van Allen 1972). On a practical level, the effort led to important changes in state policies toward local ethnic groups (Stack 1974), a critique of gendered effects of capitalist development or at least shifts in how agencies and processes conceptualized “development,” how projects, when implemented, affected gender relations (Leacock 1977b; Okeyo 1980), and sometimes how local women organized to ameliorate their own conditions (Bell and Ditton 1980; Omvedt 1977). Within the academy, these approaches led to a profound reevaluation of kinship studies, economic anthropology, and political anthropology (Vincent 1990), among other subfields. The concomitant push for affirmative action by the ongoing women’s movement saw important changes in hiring and, to a lesser extent, tenuring and promotion of women in the academy (Sanjek 1978).

14_107-Bernard.indb 156

6/9/14 10:02 AM

5: Feminist Methods

157

Most of the early studies (1970–1980) assumed that gender differences paralleled sex differences or uncritically accepted sex differences as unmediated by culture. Yet, even in this early period, some understood that sex differences were not the same as gender (Oboler 1980) and that gendering changed historically and in relation to racial stratification, even if few used the term “gender” at the time (see Leacock and Nash 1977; Stack 1974). The effort to retrieve earlier, submerged ethnographies that did deal with women in important ways, such as Eileen and J. D. Krige’s The Realm of a Rain Queen (1943), challenged assumptions about the presumed parallel of sex and gender: Prominent Lovedu females could be simultaneously wives and mothers and husbands and fathers, for example. Other women scholars, reflecting on field experiences, noted their being situated locally as categorical men, so the contention that this first spate of feminist anthropology didn’t distinguish sex and gender is inaccurate: Some anthropologists did, even if the terms used then weren’t the same as the distinctions used now. Many researchers, however, identified gender roles as sex roles; later feminist scholars would point out the limitations inherent in such analyses, particularly when claims were being made about “women’s status” (Amadiume 1987). Focusing on “what the women are doing” remains a viable research strategy in arenas of social research that retain a masculinist set of assumptions and social relations. In those arenas, it is still unknown what women contribute in terms of productive and reproductive activities and how this work and, indeed, women’s lives have cultural and social value. It generally remains a starting point for ethnographic research, although focusing on gendering will often change initial impressions. Including other dimensions of social hierarchy (class, ethnicity, racialization as a process, nationality, sometimes regional or religious affiliation) prevents “women” from becoming a monolithic category. Most feminist researchers now appreciate the importance of analyzing how gendering occurs in a specific setting, how people perform gender, and how gender categories and performances can shift in relation to local, national, and international political and economic processes as well as in relation to life cycle and kin dynamics. A more radical departure from the early studies entails questioning the “naturalness” of sex differences; some argue that the split between sex and gender is itself a Cartesian dichotomy (Leacock 1977a). The argument is not that there are not physiological distinctions between females and males related to biological reproduction, but that, on the one hand, what constitutes sex difference has changed with new technologies and, on the other hand, people’s perceptions of what “counts” as a natural difference are culturally shaped. Conducting this kind of corrective research raised three other sets of issues for feminist anthropologists: (l) how funding for field research occurs and consequently how power practices in fields differentially value research topics (see Hanks 2005, 72–75); (2) how the audience and discipline value research and researchers; and (3) how the standpoint of the researcher shapes the relationship of researcher and those studied. The first two sets of issues exposed key areas of masculinist bias in anthropology and related disciplines: Research by or about women was not comparably valued. These issues became painfully obvious when the first round of tenure

14_107-Bernard.indb 157

6/9/14 10:02 AM

158

Christine Ward Gailey

decisions of newly included women scholars publishing their research on gender saw a flurry of rejections (see Valian 1998). Here the social engagement relationship of theory and practice came into play. As researchers were uncovering dynamics in other societies that valued women’s work, or how social hierarchies eclipsed or demeaned women’s contributions, researchers came to realize how similar dynamics operated or did not exist in their own milieus. The sharing of research results in the emerging cohort of feminist anthropologists led to national campaigns to redress male bias, including orchestrated action to affect university policies and funding priorities in private and public foundations. Alongside a large number of individual researchers’ case studies, Friedl’s (1975), Leacock’s (1972), and Karen Brodkin Sack’s (Brodkin [Sacks] 1979) comparative studies suggested that women in less-stratified settings enjoyed greater social authority. Researchers differed, however, about whether gender hierarchies existed in societies not encompassed by states (Leacock 1981). Debates ensued regarding: variations in the meaning of marriage (Etienne 1980); how to assess women’s status (Brodkin [Sacks] 1976; Edholm et al. 1978; Ortner 1974); whether public/private realms existed in all societies (Lamphere 1974, 1977; see also Lamphere 1993); whether a division of labor by gender inherently embedded gender hierarchy (Brown 1975; Leacock 1972); whether state formation involved the historical development of consistent gender hierarchy (Leacock 1983; Rapp 1977; Silverblatt 1988); and whether other forms of gender hierarchy existed besides patriarchy (Rapp 1977; Gailey 1980, 1987a, 1987b; Ortner 1981; Silverblatt 1978). Feminist anthropology in the 1970s problematized the study of women in a number of dimensions and legitimated the line of inquiry, at least as a “ghetto” most departments came to deem necessary to include, although not emphasize. The works were empirically grounded (i.e., based on field or archival research). Some were also empiricist (i.e., they presumed a tangible, measurable real world existing independent of interpretation). Some were also positivist, aiming at general social laws and achievement of a gradually perfected truth. The influence of the flood of studies and theoretical writings led researchers to question whether it was acceptable to talk about women as if they were a homogeneous group, especially but not exclusively, in stratified societies. This realization helped move the discourse of feminist anthropology to discuss gender as a configuration of cultural expectations of people, a script that might not be isomorphic with the categories of sex difference.

From the 1980s to the End of the Century: Problematizing Gender, Complicating Women’s Identities and Oppressions By the early 1980s, researchers who had not already done so were moving away from discussing “woman” as a monolithic category or assuming that writers could present women as having a modal experience even in one culture. Correlatively, feminist anthropologists eroded notions that “the household” was a monolithic unit vis-à-vis outside power relations, presenting kin networks and housing arrangements as fraught with internal, gendered power dynamics (Stack 1974). The 1980s in the United States also saw demands for inclusion and diversification within feminism by women of color (Lorde 1990), which brought new concerns on both practice and research fronts to

14_107-Bernard.indb 158

6/9/14 10:02 AM

5: Feminist Methods

159

feminist anthropology. The differential effects of race, class, ethnicity, nationality, and sexual orientation on the lives of women and men made feminist anthropologists move toward analyzing the intersections of social hierarchies that produced distinctive perceptions of the world (see, e.g., Morgen 1989; Mullings 1986). The evolutionism that marked much of the earlier work gave way to research focusing on configurations of political economy and sociocultural dynamics that shaped diverse sources of oppression and authority within societies at any given point in time. Moreover, the 1980s saw the emergence of a series of studies that complicated the category “women” with both historical and comparative variation and that began to examine the gendering of analytical categories commonplace in anthropology (MacCormack and Strathern 1987 [1980]). Others analyzed the transformation of gender through imposed political and economic processes and resistance to those processes as a vehicle for understanding the range of feminist practices in the work. Etienne and Leacock’s Women and Colonization (1980) was a series of case studies analyzing dimensions of colonization (conquest, economic domination, state formation, ethnic and class formation, racism, religious conversion), as each involved efforts to impose new notions of gender or to ordain aspects of local gender systems that “fit” the colonizing agents’ notions of propriety. The case studies also investigate how these colonial agendas were sometimes at cross-purposes and often triggered profound (sometimes successful) local resistance. Key in these analyses were: (1) how gender ideologies and practices in the colonizing group affected labor relations in colonized areas; (2) the collision and collusion of gender systems, depending on the degree of stratification before colonization; and (3) the emergence of gender as an arena of struggle in both colonial and postcolonial periods. The argument was that gender was a historical, changing set of beliefs that were related to political and economic changes but not reducible to them. Gendering was a process that articulated disparate structures of racial, ethnic, and religious domination in colonialism, but depended also on women’s and men’s actions, or agency (a point explored later in Stoler 1991). It also introduced ways of documenting historical and contemporary resistance to the imposition of gender hierarchies and the uneven development of gender hierarchies, even under conditions of capitalist development (Leacock and Safa 1986; Nash 1985; see also Gailey 1987b). The activist demands of gays and lesbians within the United States opened a new terrain for scholarly inquiry, legitimating the study of sexualities that long had been limited, with very few exceptions, to heterosexual and clinical studies, Gayle Rubin’s early contribution (1975) led the way by challenging the naturalness of heterosexuality in sex/gender systems and calling attention to the heterosexual bias in the anthropological study of kinship. An anthology by Whitehead and Ortner (1981) sought to discuss the relationship between sexuality and gender as cultural constructs. Some of the contributors approached the issue of sexuality in a cultural context that included historical change. In keeping with the overall structuralist perspective of the volume, however, gender was assumed to be inherently hierarchical, although taking different forms in different circumstances. The volume’s linkage of sexuality with gender systems anticipated the move among many feminist anthropologists in the 1980s toward Foucault’s version of poststructuralism (Yanagisako and Delaney 1995), particularly

14_107-Bernard.indb 159

6/9/14 10:02 AM

160

Christine Ward Gailey

the assumption that difference is rooted in conceptual schemas in which the terms may shift as part of a configuration of cultural practices, including the ways in which the issues are framed by those deemed experts or powerful. Because mainstream feminist movement practices in the 1970s and early 1980s tended to ignore or otherwise bracket or marginalize working-class women, lesbians, and women of color, by the mid-1980s these previously submerged or ignored people demanded not just inclusion but transformation of feminist organizations. Researchers had to abandon more naive assumptions that simply “being a woman” gave a privileged lens through which to view power relations in a setting. This led a wider range of feminist anthropologists to examine critically the relationship of structural oppression to types of women’s activities and activism. Women’s identities were more complex— cross-cut and embedded in other kinds of sociocultural hierarchies—and this complexity required careful investigation. The forms of social hierarchy that interwove and collided, producing multiple identities for women and men, also created arenas where action was possible, where agency mattered (Lorde 1990; Mullings 1997). Attention to structure and agency, on the one hand, and, on the other hand, the relationship of researcher to questions that could be posed or imagined, moved most of the terrain of debate from positivism toward what became known as standpoint theory. STANDPOINT APPROACHES

The early feminist approaches raised another set of methodological issues: How does the social category and position of the researcher influence a relationship with those who make the study possible, the so-called subjects? In earlier studies, it was thought that because the researcher was in a marginalized category—women—as were the people among whom the researcher lived and observed, there would be a kinship that might even provide a privileged lens for analysis (see, e.g., Rohrlich-Leavitt et al. 1975). This position paralleled that of cultural nationalist strands of the civil rights movement in the United States: The researcher from a disempowered category has an edge in conduct in research on other disempowered people. Even though she is privileged in being able to conduct the research at all (national and class privilege), she carried with her the consciousness of her own oppression within her home society. Thus, the researcher in the marginalized category develops a kind of “double consciousness” that permits insight into the kinds of power relations that might be oppressing the people whom she studies. They might relate to her better because, on some level, she is one of them, too. The position raised two questions: On what basis do we make claims and to which audiences? How do we reduce the chasm of power between anthropologist and informants/subjects? This argument remains controversial. Researchers in the 1980s and 1990s developed a number of strategies that addressed the problem of power in anthropological field research: advocacy, witnessing, conducting research on those more powerful, challenging the notions of “informant” and “subjects” to the point of moving toward co-production, and so on. Scholars held one central technique of field research, participant observation, open to critical examination for an inherent evasion of the question of researcher commitment and its reproduction of “othering”—that is, rendering the subjects of research into objects, at its worst, voyeurism by the more powerful.

14_107-Bernard.indb 160

6/9/14 10:02 AM

5: Feminist Methods

161

Demands for inclusion from oppressed groups—primarily women of color—within industrial capitalist countries and the postcolonial world moved from insistence on being heard, to shaping the research agenda, and exposed the pretense of “shared fate” in feminist fieldwork as another form of imperial arrogance (Narayan 1995). Some called for research to be done only by women from the same culture, but as many pointed out, the class, ethnicity, race, and religion of outsiders might make them better witnesses than someone whose gender or national identity was the same (see Bell et al. 1993; Whitehead and Conaway 1986). What emerged from this ongoing controversy are different feminist strategies for developing sharing and collaboration with participants in a research process. At the same time, the reflexive aspect of one line of anthropological research (recalling Montaigne) pushed for more modest claims, working with people to develop critiques of both local dynamics and those in the researcher’s society. Some proposed a kind of resolution to this by having the researcher’s configuration of social categories be somewhat aligned with those of the subjects. The problem is, of course, which categories matter more for the research in question. The researcher might consider that because there is a match with informants in one or two dimensions there is privileged access or understanding. But other dimensions could be more important to local participants in the project, possibly alienating them from the researcher who is unable or unwilling to recognize these other dimensions. Indeed, for U.S. feminists, class differences within a shared ethnicity pose particular problems (see Zavella 1996), particularly since class analysis isn’t encouraged within the U.S. social sciences. Acknowledging power differences is necessary in feminist research, but how to integrate one’s knowledge of them into the research design and process remains a matter of contention (Warren and Hackney 2000). TRANSLATION, TESTIMONIES, AND TRANSNATIONAL QUESTIONS OF POWER

Creating “situated knowledges”—a series of clearly interested arguments—has been suggested as a means of making clear that knowledge is socially produced and a historically shifting terrain of contentious meanings. Situated knowledge, based on the meanings created by the encounter between people who are different in various ways, depends on understanding that social hierarchies make shared meanings unlikely. It is a logical outgrowth from what is called “standpoint theory” and a resolution of some of its problems (Wylie 2004). Harding characterized standpoint as feminist analysis that “insists the inquirer him/herself be placed on the same critical plane as the overt subject matter, thereby recovering the entire research process for scrutiny in the results of research” (Harding 1987, 9). Harding elaborates that standpoint entails explicitly acknowledging for the audience/reader the role played by the researcher’s “class, race, culture, and gender assumptions, beliefs, and behavior” in shaping the research design, process, and results. The researcher in this stance cannot remain an anonymous expert, a hidden authoritative voice. This standpoint is sometimes seen as a vehicle for achieving authorial responsibility and for insisting that the audience has a right to know that knowledge is socially produced. This concern with empowering the reader to join the evaluation process by making the author accountable for the text has become

14_107-Bernard.indb 161

6/9/14 10:02 AM

162

Christine Ward Gailey

widespread. The question of accountability enters into feminist anthropology in the mid-1980s and articulates in an important way with an implicit contradiction between feminist practice and cultural relativism. One of the major debates continuing to the present day challenges cultural relativism where violence against women is concerned. A number of researchers have argued that the particular expertise of feminist anthropology can help stop violence against women and can facilitate creating fulfilling lives beyond violation (Anglin 1998; Hoff 1990; Wies and Haldane 2011). Studies of gang and acquaintance rape (Sanday 1990, 1995, 1996), wife battering (Counts et al. 1992), female genital mutilation (Gruenbaum 1993 [1982]), and legal outcomes of women seeking redress or protection from violent partners call into question anthropology’s historical commitment to relativism where the health and safety of marginalized people are involved. But the issues involved also underscore the relationship of the researcher’s relative social power to the kind of research undertaken. Let us consider one continuing controversy. Beginning in the 1980s, there was an outcry among several anthropologists from the United States and other Euro-American countries calling for an international ban on female circumcision, clitoridectomy, and infibulation in specific countries in Africa. (See Walley [1997] for a nuanced discussion of the issues involved.) Feminist anthropologists and feminists in general from countries where female circumcision and infibulation are routine responded to this call in ways that rocked feminist circles across the globe. Feminists from Egypt, Nigeria, Sudan, and other relevant countries had denounced the practices and called for their abandonment before feminists from other parts of the world took notice of the issue (e.g., El Saadawi 1980; Thiam 1986 [1978]). These African activist scholars and public health workers have called activists from the United States—both white and black—to task for reinforcing colonial images of Africa and for focusing more on claiming leadership of some kind of “global feminism” rather than on showing needed international solidarity. The question was not whether female genital mutilation was tolerable—no one involved in the debates defended the practice. But local activists insisted that others consider: (1) What happens to girls—ranging from sexual harassment to sexual assault and relegation to the sex trade—if the practice is legally banned without providing alternative ways to establish respectability; (2) What venues for activism might be more appropriate for activists from other countries—such as placing pressure on their own countries to grant asylum or addressing the murders and mutilations of women on a daily basis within the activists’ own countries; and (3) What happens to the effectiveness of local feminist interventions when perceived leadership is too closely identified with intellectuals from former or present colonial powers? Ensuing discussions revealed the power dynamics inherent in cultural relativism: The notion of a homogeneous culture does not fit the lived realities of women and men in societies differently affected by class, gender, ethnic, racial, or religious hierarchies. Whose version of the culture is privileged when researchers argue that a particular practice is “traditional” and therefore inviolate? Whose interests are served by recourse to legal rather than community-based actions? Does a top-down strategy necessarily mimic masculinist models of social action?

14_107-Bernard.indb 162

6/9/14 10:02 AM

5: Feminist Methods

163

At the same time, the debate has challenged feminist researchers within more powerful countries to examine the responsibility of outside investigators to local women and national-level activists. The power dynamics in the range of international feminisms are intricate, complicated by lines of national status, class, and color (see Bolles 1996; Susser 2009). They demand that the researcher examine how collaborative the planned project is and how it articulates with work being done by local feminist activists and organizations (Stephen 1997). One positive result of the debates has been a proliferation of research on the role of U.S. medical, legal–judicial, and social institutions in facilitating or reproducing violence against women (e.g., Sanday 1995; Schultz and Mullings 2006). Another has been a call to address the multiple needs of women who have been violated or face risks associated with chronic, structural violence, particularly where these women come from diverse backgrounds (Connell 1997; Richie and Kanuha 1997). Many of these recent studies situate the anthropologist more clearly and involve local people in research design and analysis in unprecedented ways. Calls for arguing from a stated position are efforts to make inquiry accountable. Although clearly articulated standpoints make evaluation easier and place debates on a straightforward footing, the power problem remains: Some knowledges are better situated for inclusion in global debates and policy than others. How do we situate ourselves as ethnographers, for instance, without either overwhelming the voices and perspectives of local people or irritating readers with self-indulgent forms of reflection? The strategies taken to address the question of femininist solidarity versus intersecting hierarchies that create power differences depend largely on the political stance of the researchers in relation to the women who make the research possible (Brodkin [Sacks] 1988). The concern with standpoint approaches in feminist anthropology continues. It also has given rise to an offshoot, born of the efforts to make standpoints clear and to produce an anthropology that forefronts the unequal negotiating process inherent in field research. Feminist anthropologists engaged in field research where the anthropologist’s identity overlaps those of her informants in key ways have emphasized that shared dimensions of identity are no guarantor of either access or resolution of questions of expertise or power. For example, Kath Weston’s (1997) research in lesbian and gay communities in the United States and Ellen Lewin’s (1993) work on the implications of lesbian mothering for American kinship have stressed the need for negotiating and developing consensual strategies for communication between the ethnographer and her informants where both are aware of the implications of the research for an entire community, even as the author remains accountable to both community and wider audiences for her arguments. Field research involving vulnerable populations, such as battered women or breast cancer activists, highlights the need to not revictimize as well as the vagaries of participant observation where empathy, power, deployment of data, and confidentiality are so clearly implicated in women’s safety and well-being (Anglin 1997; Connell 1997; Hoff 1990). The new global division of labor draws on women and men differently and intensifies conditions leading to international migration (Boehm 2012; Nash and Fernandez Kelley 1983), creating a condition of transnationalism. On an unprecedented scale, people’s identities and affiliations are challenged and hybridized through experiences

14_107-Bernard.indb 163

6/9/14 10:02 AM

164

Christine Ward Gailey

of migration, living “in between” as well as in two or more countries (and gender systems), and tapping cultural resources from a variety of sources (Segura and Zavella 2007). Transnationalism has made concerns about feminist practice in field research even more charged (Warren and Hackney 2000). The shift in communications technology, literacy, and migration have meant that informants are better informed, more skeptical of outside expertise, and are not invisibly and silently ensconced far from the researcher’s home (Mascia-Lees 2010). Feminist anthropologists tackling the thorny issue of how not to impose a form of feminism embedded in a particular configuration of gender, race, class, and cultural tradition have explored what methodology means in these new transnational, linked but power-laden spaces (see Brodkin 2000; Grewal and Caplan 1994). Reflexive approaches are one response to the problems raised (see Enslin 1994). In these approaches, the researcher assumes that the entire research process—from design to audience—reveals as much, if not more, about the society sending the researcher and the researcher’s own concerns as it does about the people who are the targets of inquiry. Most feminist anthropologists would argue that, in important ways, issues in their own society and/or the concerns of the people with whom they work spark, shape, or inform the problems they research (Lamphere et al. 1997; Zavella 1996). The degree to which reflexivity characterizes a feminist text varies, but all seek to make the act of writing and filtering experience—of creating knowledge claims—apparent to the reader (see, e.g., Malkki 2007). For those most committed to the approach, the task of the author is as much to unmask and lay bare the artificiality, the edifice, of the text as it is to make any other claims (Behar 1993; Behar and Gordon 1995). Many feminist anthropologists from the postcolonial regions of the world, however, tend to deploy reflexivity as a strategy to reveal division of class and other forms of social hierarchy in complex social settings where precolonial, colonial, and postcolonial labor forms and gender ideologies collide. Piya Chatterjee, for example, challenges received knowledge about caste, class, and gender in the lives of tea plantation workers in India through calling into question the many-layered textures of power shaping how and when these women mobilize for improving their conditions in communities and fields (Chatterjee 2001). In the course of discussions with the women about labor, literacy, citizenship, and agency, she and the research participants call into question her role as ethnographer and as an educated and mobile woman from a higher class and caste. She shares the local women’s views with the reading audience and thereby illustrates dimensions of feminist fieldwork far beyond the limitations of identity politics. Some researchers write about experience as if it and the production of knowledge were the same. The multiplicity and often amorphous swirl of perceptions, meanings, and emotions accompanying a significant event or process must be related to an audience, a problem, a reference to become knowledge; the form of writing and sense of audience shape the knowledge produced (Wolf 1992). Experiential knowledge is possible, but it is not automatic. Writing about experience without these intervening concerns is memoir. Some feminist ethnographies done in the name of reflexivity convert the researcher into the main, or a main, object of research. This is not a problem for memoir as a genre, but it is a problem for ethnography (see Wagner-Martin 1994). Politically, the field memoir can create a new

14_107-Bernard.indb 164

6/9/14 10:02 AM

5: Feminist Methods

165

form of silencing. To redress the imbalance of power that originally makes translation of one people’s experiences for an audience in another culture, according to the Latin American saying, into betrayal (Alarcón 1994), the researcher uses the words of the informant as a foil for her own exploration of self. The risk here is collapsing the certainly problematic anthropological encounter into a kind of psychological memoir or, at worst, a travelogue. As Mascia-Lees et al. (1989) point out, research in this vein risks audience alienation. It also, I might add, risks favoring a form of selfhood that may be at odds with local constructions of self-in-community. In the testimonial of Maria Teresa Tula (1994), done in dialogue with anthropologist Lynn Stephen, Stephen argues that collaborative encounters are necessarily flawed in a global setting where power accrues to the researcher but that this is no excuse to withdraw from feminist practice. She argues that feminist ethnographers can have witnessing and facilitating roles, furthering the agendas of the informants and adding to the subjects’ sense of their own experiences her privileged lens onto global capitalist processes, state dynamics, and the like. This kind of collaboration involves sharing lessons from both parties’ efforts and bringing the collective wisdom to bear on issues in each society, informing both collaborators’ practices. In another volume on women-centered social movements in Latin America, Stephen argues that working-women’s movements often develop out of concerns for the well-being of family and community (Stephen 1997). In addressing these issues, participants come to define what can be seen as a feminist agenda and adopt strategies of empowerment that define a distinctively indigenous and often class-based feminist practice (McClaurin 2001; Montoya et al. 2002). As discussed above, a range of techniques of investigation can be used in feminist anthropological research. The techniques used depend largely on the questions posed and what can be developed as data in the research context. Ethnographic work on labor and survival strategies among women differentially affected by offshore corporate investment, sweatshop labor, and transformations of national markets have deployed a range of methodological approaches (e.g., Heiman and Freeman 2012). Gracia Clark’s book on Ghanaian market women makes use of life histories to address the intricacies of structural constraints and individual and collective agency in the context of state policies, gender ideologies and hierarchies, and international markets (Clark 2010). Jane Collins uses comparative, multi-sited approaches in her examination of gendered labor and power dynamics in garment assembly sweatshops (Collins 2003). Ida Susser compares women’s self-reported histories of collective organization and various degrees and forms of gender and racialized ethnic hierarchies in contemporary South African communities to analyze how different groups of women once conflated as “Blacks” under apartheid confront, contest, and make demands of state agents and policies concerned with HIV/AIDS (Susser 2009). Also relevant are the audiences for whom the research is intended. If feminist ethnographers are trying to influence policy, for example, the research conducted has to make sense to policymakers: Quantitative techniques may be necessary in such circles. The same researchers could also use the same research project to reach decidedly different audiences—social scientists or undergraduate students, for example. They would therefore develop data and communicate it in different ways to relate to those

14_107-Bernard.indb 165

6/9/14 10:02 AM

166

Christine Ward Gailey

people. What mark the project as feminist are the focus on gender as a salient analytical category and the purpose in contributing to an overall effort to dismantle gendered forms of oppression and exploitation. Generally, these twin aspects make the research interdisciplinary as well. IMPACT ON ARCHAEOLOGY, BIOLOGICAL/PHYSICAL, AND LINGUISTIC SUBFIELDS

Today, American anthropology’s canonical four fields are a terrain of struggle—some call for hiving off cultural anthropology from the other subfields; many others call for reinvigorating and fully integrating the four in novel ways. A number of anthropologists defend arenas where addressing questions about human beings in general remains meaningful—our symboling capacity, our sense of our own physiological makeup or what some call “corporeality,” our sociocultural structures and universes of meaning, and our sense of our past. Feminist anthropology both merges some of those terrains and operates within them. Yet feminist anthropology has developed unevenly in the subfields of anthropology, a fact that reflects in large measure the different proportions of women to men in the subfields. Wylie argues, for instance, that feminist methods in archaeology have been later to develop than those within ethnology (Wylie 1991). Patterson (1995) holds that in archaeology this is due to both social exclusion of women and hermetic concerns with empiricism (e.g., Flannery and Marcus 1994). Indeed, the first feminist anthology in ethnology was published 20 years before the first one in archaeology (Gero and Conkey 1991), and the uneven development of the concept of gender in these initial articles reflects some of the scholars’ discomfort with the notion of gender rather than simply women. But there are excellent examples of how gendering archaeology can provide vital new insights into sociocultural dynamics (see Wright 1996). Elizabeth Brumfiel’s and Rosemary Joyce’s research show how gender can be central to the analysis of state dynamics, a concern that has dominated the subfield since its inception (Brumfiel 1996; Joyce 2000). Using standard field techniques and a theoretical orientation informed by debates regarding gender from a number of disciplines, Brumfiel challenged the prevailing ideology in the archaeology of state formation that attention to gender is best limited to the division of labor and the succession disputes or alliance strategies of ruling elites. This is not to say that gender does not matter in these arenas, only that women are more than producers and vehicles of political alliance among the “real” movers and shakers—that is, men (see Wright 1996). Joyce’s studies of sex and gender in Mayan archaeological contexts shows that archaeological data need not limit the extent to which sex can be analytically separated from gender, but that theorizing gender can help us develop new data (Joyce 2008; see also Conkey and Williams 1991). Using stylistic variations from central and more marginalized areas of the empire, Brumfiel argued that, through gendered symbols, local material culture can literally embody resistance to a dominant state ideology of gender. This moves the argument beyond the gender division of labor—although that remains vital—to mark and provide the contours of cultural struggle.

14_107-Bernard.indb 166

6/9/14 10:02 AM

5: Feminist Methods

167

In biological anthropology, pioneering examinations in the 1970s by Lancaster (1973) and Leibowitz (1983 [1975]) of male bias in primate studies and in human origins work produced novel theorizations that changed the “story” of human origins permanently. It was one of the most important contributions during the more disciplinary, “corrective” phase of feminist anthropology. During the 1970s, widely accepted “man the hunter” arguments that stressed the centrality of purportedly all-male hunting groups in the development of language, food sharing, and cooperation were countered by “woman the gatherer” ones, which stressed the centrality of gathering and mother–infant bonding (Slocum 1975; Zihlman and Tanner 1978). Only Leibowitz questioned why feminists would assume that any gender division of labor existed from the outset of humanity and offered a revisionist interpretation of the fossil and material culture remains that proposed an initial age-based technical division of labor; a gender division of labor, she argued, only emerged during Homo erectus times (Leibowitz 1983 [1975]). Critiques of basic assumptions that gender is derived from sex differences, however, are slower to emerge in physical anthropology circles (see Hrdy 1981; Zihlman 1985), in part because feminist women and men whose work does not fit into the dominant paradigms in the field experience professional marginalization with fewer opportunities for funding and deep suspicion on the part of those most committed to prevailing models (see Lancaster [1991] for an extensive review of these persistent issues). Donna Haraway’s Primate Visions was pathbreaking in its exposé of the gendering of primatology and human origins research (Haraway 1989). Most of the analyses of how science has objectified women’s bodies and biological processes and how women’s diverse experiences can lead us to revise such iconographic visions of physiology have come from cultural anthropologists (Franklin 1993a; Ginsburg and Rapp 1995; Gottlieb 1988; Martin 1992). In recent years, however, the work of Dean Falk, Lori Hager, and Rachel Caspari in paleoanthropology have facilitated understandings across the subfields regarding how deep patterns of difference by sex and gender might be (see Caspari 2011; Falk 1999; Hager 1999). The particular burden of feminist biological anthropologists is the resurgence in the past five years particularly of popular studies that reduce women to a biologically scripted category expressing natural predilections in their psychology and social interaction. Having to respond to such ill-grounded but attention-grabbing claims is a distraction, albeit a politically necessary move (see Zihlman 1999). Linguistics is itself a discipline, so justice can hardly be done to the exciting work on gender and language in this small space. Feminist ethnographies of speaking and linguistic anthropology have focused from the outset on analyzing how various language practices are gendered and how that gendering, in turn, shapes ways of thinking about the social world and ordering social relationships (Bonvillain 1993; Tannen 1994). In the 1980s, feminist anthropological linguists investigated gendered languages of specific workplaces and their implications for how technology, warfare, and so on enter into societal discourse (e.g., Cohn 1988). Exciting research is ongoing into the ways that gender affects classroom discourse and the marginalization of girls and women (Tannen 1996). Others have analyzed gender dynamics in primary and secondary language acquisition (e.g., Pica et al. 1991) and how gender bias enters classrooms through language use in naming, claiming, and

14_107-Bernard.indb 167

6/9/14 10:02 AM

168

Christine Ward Gailey

distancing (e.g., Wellhouse and Yin 1997). Feminist linguistic anthropologists also have been active in documenting the ways that creoles and pidgins, born of colonial encounters, are gendered in ways that may hamper women’s actions in postcolonial and decolonization efforts. Related research focuses on how, in linguistically diverse postcolonial settings, women’s access to learning national languages may be blocked by educational structures and local practices. Other arenas include the gendering of nationalist discourses or the deployment of gendered languages in the shaping or dissolution of community in transnational settings (see Urciuoli 1995). FEMINIST METHODS IN ACTION

By the 1990s, feminist anthropology had become a legitimate specialization. But despite efforts to integrate gender and feminist epistemologies in the anthropology curriculum (Ewick 1994; Morgen 1989), canonical texts in anthropology—including the so-called new ethnography that discusses the power dynamics of fieldwork and representing other people through writing (e.g., Clifford and Marcus 1986)—continued to ignore or give short shrift to the influences of feminist anthropologists (Lutz 1990). Hindered by their androcentric scope, the irony is that some of these same sources unwittingly appropriated concepts developed years before by feminist anthropologists. The eclipse did not stop the methodological and theoretical debates in feminism and feminist anthropology (see Folbre 1993). The 1990s found feminist anthropologists reexamining central anthropological concepts, such as kinship, whenever gender is deployed as a lens (Collier et al. 1993; Collier and Yanagisako 1987). Similarly, intersectional studies explored ever deeper the ways that gender, race, and class are mutually constructive (Brodkin 1996, 2000; Mullings 1997), and how gender provides a novel way of understanding the dynamics within ethnic and working-class communities (Brodkin 1993; Lamphere 1987). But even without these re-envisionings of classic anthropological terrains, we can see from the discussion above that there was little in the concerns of the “new ethnography” that feminist anthropologists did not discuss and debate much earlier. Feminist researchers developed conceptual tools for investigations embedded now in anthropological discourse: intersectionality, standpoint, authorial accountability, collaborative design, situated knowledge, constrained agency, and so on (Sandstrom and Opsal 2013). Indeed, the interdisciplinarity characteristic of feminist anthropology mark it as a model for the debates regarding textuality, the writer as expert, the question of reflexivity in social science inquiry, and the relationship of theory and practice. Perhaps because industrial capitalist societies that still produce most anthropologists barrage women and especially women of color with naturalizing and reductive images of their being, feminist anthropologists have been earlier and more consistent than proponents of “writing culture or the “new ethnography” in their criticism of the objectifying and reductive tendencies in social research. Let us examine some actual studies to see how the methods work in practice and how theoretical inquiry and practice can be mutually instructive. Research on gender and power in the 1990s informed efforts to call attention to and change policies directed at women both in the global north and south, but with new attention to ways that other vectors of social power affected the research process and new strategies for redressing the imbalance and producing participatory research (see

14_107-Bernard.indb 168

6/9/14 10:02 AM

5: Feminist Methods

169

Jarrett 1994; Wang et al. 1996). For example, Ida Susser’s ongoing work with Denise Oliver, Maritza Williams, and other anthropologists on homeless families in New York City in the 1990s paid keen attention to intersections of identities in the development of research strategies and in outreach work. In a series of articles Susser examined, through narratives from homeless women “talking to” social workers, how dire poverty shapes women’s lives as mothers, wives, sexual partners, and intermittent wage workers (Susser 1991, 1993, 1996; Susser and Gonzalez 1992). The work anticipated later ethnographic explorations of women’s agency. Susser and her team showed how women were active on behalf of themselves and their children, yet were also structurally and therefore personally vulnerable. Gracia Clark would explore this in her 2010 work on market women in Ghana; Leith Mullings and Alaka Wali analyzed constrained agency and the everyday violence of poverty conditions as constituting stress-related health risk factors affecting women in central Harlem (Mullings and Wali 2001). The Susser team studies present a comprehensive view of gender relations across generations, including men and children as actors, and in relation to the agencies and gatekeepers the people, framed in gendered ways, must engage to survive. The composition of the research team drew on diversity as a resource in anthropological scholarship: The interviewers’ own racial and ethnic backgrounds provided useful differences in how the range of people interviewed might deploy information, implementing how the team and the people who facilitated the study problematized perceptions of relative power between interviewer and interviewee. Although difficult to arrange given the transient living conditions of the people contacted, the team conducted multiple interviews. The team also interviewed social workers because, as Susser explains, they often were the only outsiders with any sustained connection to these families and because they simultaneously hold power over homeless families but also can channel resources no one else can (Susser 1998). The multiple predicaments faced by these women and their families are placed in the political–economic and historical context of a rapidly restructuring New York City (Susser 1996). This research has clearly drawn implications for health care, social welfare, and domestic violence services. Thus, the study addresses the power relations of anthropologists and informants structurally by diversifying the team and making policy arguments at key public events in the city without agonizing about the personal identity of the researchers. In tandem with this research into the relationship of gender to the political and economic aspects of oppression in late capitalism within the United States, internationally and transnationally (di Leonardo 1993), two other arenas are of growing importance in feminist anthropology today: the politics of sexualities and their relationship to gender and other aspects of social hierarchy and the cultural and political–economic aspects of reproduction. The research conducted on sexuality and culture demands that we reconsider how closely or obliquely sexual orientation and practices are linked with gender systems and identities (Wilson 2002). Works such as Claire Riley’s and Kath Weston’s different treatments of American kinship focused on lesbian families (Riley 1988; Weston 1991), and Ellen Lewin’s (1993) argument about reproduction and parenting in lesbian communities encouraged reexamination of what kinship means and how people create kinship.

14_107-Bernard.indb 169

6/9/14 10:02 AM

170

Christine Ward Gailey

Today, because of the invention and marketing of technologies permitting transgendering that may include surgical reconfiguration, new questions have come to the fore. Such technologies challenge the naturalness of sex differences and make apparent that what is perceived as nature is actually a cultural encoding of physicality. In the United States, for example, female-to-male operations vary from breast removal plus testosterone treatment to radical hysterectomies and constructions of penises. The variation places emphasis on gender as performative in a range of ways, both for the transgender or transsexual person and the intimate partners (see Ward 2010). The variation reveals the widespread association of maleness in the United States with having a penis and more testosterone than estrogen production, but at the same time calls into question the “naturalization” of masculinity as associated with ejaculation and sperm production. This arena of inquiry pushes the frontiers of feminist anthropology past the dichotomies of earlier formulations toward fully social constructionist appreciation of both sex and gender. Feminist ethnographic research on the gendering of reproduction, similarly, draws on the technologies invented and deployed in metropolitan centers of late capitalism to reexamine what is meant by sex differences, on the one hand, and familial kin roles, on the other (Franklin 1988). The class, race, and gender issues surrounding surrogacy, in-vitro fertilization, egg transplantation, sperm donation, and even cloning invite reevaluation of kinship categories and appreciation of class and race in relation to gender (Franklin 1993b, 1997; Ginsburg and Rapp 1991, 1995; Ragoné and Twine 2000; Strathern 1992), at least in the social locations where these technologies have their greatest consumption. Correlatively, the expansion of transnational and transracial adoption in many industrialized capitalist countries, particularly where these adoptions tend to favor one gender or the other, creates new arenas of complicated gender and racial as well as class and national identities (Gailey 1998). These adoptions call on anthropologists to develop new models of kin formation (Modell 1994) and acknowledge the primacy of women in kin formation, including, but not limited to, domestic and international adoption in racialized and class-sensitive contexts (Gailey 2010). Feminist research on new technologies and ideologies of reproduction in the new world order articulates with research on cultural variations in how human status and social personhood accrue to fetuses and infants (Morgan 1993 [1989]), research that can help readers approach debates over reproductive rights in innovative ways. Feminist anthropologists, by engaging in research that demands that they problematize the participation end of participant observation and develop more explicitly collaborative studies, are at the forefront of these new studies. From the mid-1990s to the present, several convergences in world politics have drawn feminist researchers to reconsider citizenship, nationalism, and militarism (see Nagengast 1994). The formation of new states in the wake of the collapse of the Soviet bloc, the imposition by the International Monetary Fund and other multilateral lending agencies of austerity programs that create increasing economic insecurity for women and ethnic minorities, repressive immigration policies in a number of countries, genocidal militarism in the Balkans, Rwanda, Darfur, and still in eastern Congo sparked feminist research into gender and genocide, gender and militarism, and

14_107-Bernard.indb 170

6/9/14 10:02 AM

5: Feminist Methods

171

the heavily compromised notion of citizenship for women, especially ethno-racially marked minority women. Susan Gal and Gail Kligman, for example, outlined approaches to examine the problematics of gender and citizenship in postsocialist Europe apparent in persistent and novel patriarchal ideologies and practices, loss of worker rights, increasing economic vulnerability, and shifting reproductive policies (Gal and Kligman 2000). Jacqueline Siapno investigated the reluctance of women in Aceh to participate in the nationalist movement aimed at independence from Indonesia, even as they took collective action against the militarism and gender hierarchies of both Indonesian and anti-colonial movements (Siapno 2002). Denise Segura and Patricia Zavella brought together a range of scholars exploring various facets of the impact of changing U.S. immigration policy and practices on Mexican women (Segura and Zavella 2007), and Deborah Boehm’s ethnography parses the gendered impact of undocumented status for Mexican women in the United States in relation to their intimate partners, children, and extended families (Boehm 2012). The international dimensions of gender in sexual markets have sparked feminist ethnographies and analyses of sexual orientation in comparative perspective, shifting cultural contours of desire and class and gender facets of commercialized sex (see, e.g., Brennan 2004; Cabezas 2009, Padilla et al. 2007). Recent ethnographic studies by Sherine Hafez of Islamic women’s organizing in Egypt (Hafez 2011) and of challenges posed by women in the uprising that toppled the Mubarak regime (Hafez 2012) privilege the participants’ views and rationales for their actions and how they characterize the interstices of gender hierarchy operating in daily life. These gaps and structural contradictions form socially legitimized spaces for women’s claims on the state and on a particular patriarchal order. Through her informants, Hafez provides alternative views of women’s forms of agency in patriarchal conditions, on the one hand, and, on the other hand, of the mutual construction of secularism and religious rationales for social engagement. The result challenges the “received wisdom” of much feminist discourse from within and outside Muslim societies regarding women and religious motivation. The local women’s simultaneous adherence to religious values and subversion of patriarchal and state repression requires feminist and other readers to rethink the dichotomies of secular/sacred in their own frames of reference. Feminist anthropologists continue to work with people throughout the world to understand how class and race, ethnicity, nationality, and regimes of belief shape the experience of gender and strategies for transforming oppressive relations and structures within our own societies and elsewhere (Brodkin [Sacks] 1988; Harrison 1991; Mascia-Lees 2010; McClaurin 2001). Class formation in association with offshore capitalist operations fuels research by Carla Freeman among “pink-collar” workers in the Caribbean (Freeman 2000). This concern with “intersections” cannot be dismissed as political correctness at home, for if we make claims about ourselves through the inevitably partial knowing gleaned from participants, we must understand how these important aspects of social differentiation, often but not always hierarchies, shape the lives of the people who make our studies possible (see Morgen 1989). Moreover, we need to see how shifting and intersecting social and cultural identities affect the questions we can imagine and pose and how participants and audiences perceive us.

14_107-Bernard.indb 171

6/9/14 10:02 AM

172

Christine Ward Gailey

A number of feminist anthropologists are exploring the worlds of popular culture and consumerism in relation to gender, race, and class. Elizabeth Chin’s consideration of race and gender in the consumption of rigidly marked gendering in mass-marketed toys (Chin 2001) and Purnima Mankekar’s study of gender and television consumption in postcolonial India (Mankekar 1999) anticipate some of the ongoing discussions of intersections of gender in social media (Hobson 2012). Trin Minh-Ha’s consideration of feminist positionality in relation to postcolonial and transnational gendering (Minh-Ha 2009) and her film that takes up feminist methodological problems in postcolonial contexts (Minh-Ha 2005) have influenced a growing number of filmmakers, particularly those associated with Women Make Movies, which distributes a range of feminist ethnographic works. Faye Ginsberg and her associates have taken up issues surrounding collaborative ethnographic film efforts and social media by and about women and gendering (Ginsburg et al. 2002). For the burgeoning feminist research on social media and gender transformations, interventions, and resistances, Karen Ross’s recent compendium promises to be a valuable resource, both theoretically and methodologically (Ross 2012). CHALLENGES OLD AND NEW

One of the challenges facing feminists in the twenty-first century has called particularly on feminist anthropologists concerned with the gendering of scientific discourse and the power of gendered language (see Gal 2001; Warren and Kay 2000). One arena has been the resurgence in popular science arenas and a number of scientific journals of the term “man” to refer to human beings in general. The slippage has gone largely unaddressed outside of feminist scholarly circles, but it has accompanied and is a symptom of the rightward move politically in English-speaking arenas. Another symptom of antifeminist backlash has been the return—after feminist challenges about 15 years ago—of the use of “gender” to refer to nonhuman sex roles. Most sociobiologists muddy the waters by using gender to describe sex roles and sex-related behaviors of primates and other mammalian species that do not have the capacity to create symbolic scripts around or in tangential relation to sex differences. These researchers either use sex and gender interchangeably or ascribe male/female interactions to gender. This does disservice to an arena of feminist anthropology that has documented multiple genders in some societies (beginning with Nanda 1993). These studies underscore the differences between sex and gender, on the one hand, and, on the other hand, the separation of gendering as a process of socialization, culturally created scripts, and performance from socially perceived sex differences. The overlapping of sexes and genders in some societies, while they are oblique or lack correspondence in others, makes the ascription of gender to non-symboling primates (and in some sociobiological circles, all sexually reproducing organisms) decidedly inaccurate. Feminist anthropology has drawn on studies by feminist geneticists and biological anthropologists of the range of intersex outcomes in embryonic and fetal development (e.g., Bardoni et al. 1994), some of which may not be sociologically apparent or culturally elaborated if detected. The cultural construction of gender in the United States at least persists in ascribing two basic gender scripts, regardless of such diversity in terms of sexes (Fausto-Sterling

14_107-Bernard.indb 172

6/9/14 10:02 AM

5: Feminist Methods

173

1993). If we add to this the methodologically sensitive field research that feminist anthropologists such as Serena Nanda, Sharyn Davies, and Ara Wilson have conducted to document gender diversity in a range of societies (Davies 2007; Nanda 2000; Wilson 2004), the cultural construction of multiple genders is undeniable. Together, better understanding of the genetic and hormonal processes that can produce a number of intersex outcomes in humans, coupled with studies of cultures supporting three or more genders (that usually do not map onto intersex variations) make the conflation of sex and gender indefensible. Ascribing gender to nonhuman primates is particularly egregious because it creates an image of human gendering as naturally dualistic. In terms of the scholarly and activist mission of feminist anthropology, the political implications of naturalizing gender subvert both scientific understandings of sex differences, on the one hand, and, on the other hand, lend support to a range of patriarchal policies regarding gender and sexual rights. Philosophically and linguistically, the ascription of gender to a wide range of nonhuman beings, coupled with tendency to consider gender as relating primarily to women, harkens back to the patriarchal trope that reduces women to nature and nature, in patriarchal reasoning, should be controlled by (“real”) men. Still another major challenge facing feminist anthropology in this era of economic crisis and the societal anxieties of cutbacks, little hiring, and job loss is the reduction in the number of quality Women’s Studies departments and programs in U.S. colleges and universities. There are two related assumptions, albeit rarely voiced out of Title IX concerns by university administrators. One is that Women’s Studies—often a “home” for feminist anthropologists in U.S. higher education—is too expensive, meaning there are fewer majors than in traditional disciplines, although enrollment numbers tend to be heavy. The other assumption is more insidious, namely, that the field has, as one colleague commented, “run its course,” meaning it is intellectually passé. The latter is easily dismissed by those scholars who know the administration has very little sense of the intellectual integrity of feminist methodologies despite efforts at remedial education, and deep suspicion that interdisciplinary methods related to gender are merely political and not rigorous. When we consider the attention paid to critical pedagogies, debates within feminist scholarship, support provided by the curricula and feminist mentors for students that include more diversity than almost all disciplinary departments, and national political attacks on reproductive rights and health care differentially affecting women, racialized, and sexual minorities, targeting these programs attacks one of the most important epistemological contributions of feminist anthropology, that is, intersectionality. Moreover, considering the dearth of women in the senior professoriate, downplaying the ongoing contributions of feminist methodologies provides a rationale for abandoning much-needed efforts to continue to diversify the academy. By buying into the conservative and neoliberal arguments that result in cloning “safe” methods and disciplines that marginalize feminist research, we court reinforcing existing racism, sexism, and ongoing refusal to acknowledge that working-class standpoints matter in the construction of research that can help address the multifaceted crises of the twenty-first century.

14_107-Bernard.indb 173

6/9/14 10:02 AM

174

Christine Ward Gailey

CONCLUSIONS

Feminist anthropology can‘t escape the inherent production of Others so long as power-laden differences are reproduced through the normal operations of capitalist and related/articulated political economies and attendant ideological practices (Brodkin 2000; Mascia-Lees 2010). But the process of Othering and frequently attendant racism can be held up to scrutiny (McClaurin 2001; Warren and Hackney 2001; Zavella 2006). Beyond research practices, the process of Othering includes ways cutbacks in the academy further marginalize or threaten the continuation of feminist scholarship as an acceptable intellectual arena. In the 1990s, as feminist anthropology became institutionalized as a specialization, scholars witnessed a lamentable if typical process associated with the formation of academic disciplines: the defense of boundaries signaled by efforts to produce canonic, feminist texts (Behar and Gordon 1995) and to inscribe canonical histories (Viswesrayan 1997). In due course, other feminist anthropologists held these efforts toward orthodoxy up to critical evaluation (e.g., Andersen and Collins 2009; di Leonardo 1991). Indeed, the view of feminist methodologies presented in this handbook is in part a warning about the creative loss involved in orthodoxies and a claim for appreciating the insights and diversity of approaches and theories that has informed feminist practice in anthropology from the outset. The vitality of feminist methodologies lies in the linkage of practice and social engagement with theorizing and collaborating with others to produce useful knowledges (see, e.g., Wies and Haldane 2011). We need to guard against viewing our own scholarship and engagement as inherently feminist simply because we define ourselves as feminist or are seen that way. In the resistance against orthodoxy, we can find the best argument against assertions that “feminist anthropology has run its course.” Keeping vigorous the linkage of theory and practice, however, helps researchers and their collaborators pose meaningful questions and design techniques to probe for answers—and solutions to social ills. It is not so much an issue of giving voice as it is of hearing, listening for meanings, and helping project voices: those who have been disenfranchised, marginalized, or silenced because of their gender, race, class, sexualities, nationality, and other social positioning. The global impact of industrial and postindustrial capitalism and the crises that accompany its normal operation call on us to listen, to engage, and to convey to our own communities and other audiences, the gendering of struggles throughout the world. Local variations in what constitutes gendering, the relative importance of gender in social structures and cultural practices, and how gendering shapes people’s perceptions of possible pasts, presents, and futures matter more now than ever if we are not to be consumed and fragmented beyond all ability to construct meaningful lives. Such an anthropology—by speaking from our various and shifting standpoints, and mostly by standing alongside and opening audiences for previously silenced or muted voices—offers much to the development of a wide range of feminist agendas and practices throughout the world.

14_107-Bernard.indb 174

6/9/14 10:02 AM

5: Feminist Methods

175

ACKNOWLEDGMENTS

I am grateful to Karen Brodkin and H. Russell Bernard for very helpful criticisms and suggestions regarding this chapter. I also thank Sherine Hafez, Mary Anglin, Thomas Patterson, and anonymous reviewers for their comments. REFERENCES

Alarcón, N. 1994. Traddutora, traditora: A paradigmatic figure of Chicana feminism. In Scattered hegemonies, ed. I. Grewal and P. Caplan, 110–33. Minneapolis: University of Minnesota Press. Amadiume, I. 1987. Male daughters, female husbands: Gender and sex in an African society. London: Zed. Andersen, M., and P. Hill Collins, eds. 2009. Race, class, and gender: An anthology. Belmont, CA: Wadsworth. Anglin, M. 1997. Working from the inside out: Implications of breast cancer activism for biomedical policies and practices. Social Science and Medicine 44: 1403–15. Anglin, M. 1998. Feminist perspectives on structural violence. Identities 5: 145–51. Bardoni, B., E. Zenaria, S. Guioli et al. 1994. A dosage sensitive locus at chromosome Xp21 is involved in male to female sex reversal. Nature Genetics 7: 497–501. Behar, R. 1993. Translated woman. Boston: Beacon Press. Behar, R., and D. Gordon, eds. 1995. Women writing culture. Berkeley: University of California Press. Bell, D., P. Caplan, and W. Jahan Karim. 1993. Gendered fields: Women, men, and ethnography. London: Routledge. Bell, D., and P. Ditton. 1980. Law: The old and the new: Aboriginal women in central Australia speak out. Canberra: Aboriginal History. Boehm, D. 2012. Intimate migrations: Gender, family, and illegality among transnational Mexicans. New York: New York University Press. Bolles, A. L. 1996. Sister Jamaica: A study of women, work, and home in Kingston. Lanham, MD: University Press of America. Bonvillain, N. 1993. Language, culture, and communication: The meaning of messages. Englewood Cliffs, NJ: Prentice Hall. Brennan, D. 2004. What’s love got to do with it? Transnational desires and sex tourism in the Dominican Republic. Durham, NC: Duke University Press. Brodkin [Sacks], K. 1976. State bias and women’s status. American Anthropologist 78: 565–69. Brodkin [Sacks], K. 1979. Sisters and wives: The past and future of sexual equality. Westport, CT: Greenwood Press. Brodkin, K. 1988. Toward a unified theory of class, race, and gender. American Ethnologist 16: 534–50. Brodkin, K. 1993. Euro-ethnic working-class women’s community culture. Frontiers 14: 1ff. Brodkin, K. 1996. Race and gender in the construction of class. Science and Society 60: 471–78. Brodkin, K. 2000. Global capitalism: What’s race got to do with it? American Ethnologist 27: 237–56. Brown, J. K. 1975. Iroquois women: An ethnohistoric note. In Toward an anthropology of women, ed. R. Reiter [Rapp], 235–51. New York: Monthly Review. Brumfiel, E. 1996. Figurines and the Aztec state: Testing the effectiveness of ideological domination. In Gender and archaeology, ed. R. Wright, 143–66. Philadelphia: University of Pennsylvania Press.

14_107-Bernard.indb 175

6/9/14 10:02 AM

176

Christine Ward Gailey

Cabezas, A. 2009. Economies of desire: Sex and tourism in Cuba and the Dominican Republic. Philadelphia: Temple University Press. Caspari, R. 2011. The evolution of grandparents. Scientific American Audio Book. August 11. Chatterjee, P. 2001. A time for tea: Women, labor, and post/colonial politics on an Indian plantation. Durham, NC: Duke University Press. Chin, E. 2001. Purchasing power: Black kids and American consumer culture. Minneapolis: University of Minnesota Press. Clark, G. 2010. African market women: Seven life stories from Ghana. Bloomington: Indiana University Press. Clifford, J., and G. Marcus. 1986. Writing culture: The poetics and politics of ethnography. Berkeley: University of California Press. Cohn, C. 1988. Sex and death and the rational world of defense intellectuals. In Women on war, ed. D. Gioseffi, 84–99. New York: Simon and Schuster. Collier, J., M. Rosaldo, and S. Yanagisako. 1993. Is there a family? New anthropological views. In Talking about people: Readings in contemporary anthropology, ed. W. Haviland and R. Gordon, 151–58. Mountain View, CA: Mayfield. Collier, J., and S. Yanagisako, 1987. Toward a unified analysis of gender and kinship. In Gender and kinship, ed. J. Collier and S. Yanagisako, 14–50. Stanford: Stanford University Press. Collins, J. 2003. Threads: Gender, labor, and power in the global apparel industry. Chicago: University of Chicago Press. Conkey, M., and S. Williams. 1991. Original narratives: The political economy of gender in archaeology. In Gender at the crossroads of knowledge: Feminist anthropology in the postmodern era, ed. M. di Leonardo, 102–39. Berkeley: University of California Press. Connell, P. 1997. Understanding victimization and agency: Considerations of race, class, and gender. Special Issue: Gender, Violence, and the Law. Political and Legal Anthropology Review 20: 115–43. Counts, D., J. K. Brown, and J. Campbell, eds. 1992. Sanctions and sanctuaries: Cultural perspectives on the beating of wives. Boulder, CO: Westview. Davies, S. G. 2007. Challenging gender norms: Five genders among Bugis in Indonesia. Belmont, CA: Thomason Wadsworth. di Leonardo, M., ed. 1991. Gender at the crossroads of knowledge: Feminist anthropology in the postmodern era. Berkeley: University of California Press. di Leonardo, M. 1993. What a difference political economy makes: Feminist anthropology in the postmodern era. Anthropological Quarterly 66: 76–81. Edholm, F., O. Harris, and K. Young. 1978. Conceptualising women. Critique of Anthropology 3: 103–14. El Saadawi, N. 1980. The hidden face of Eve: Women in the Arab world. London: Zed. Enloe, C. 2007. Crafting a global “feminist curiosity.” In Globalization and militarism: Feminists make the link, ed. C. Enloe, 1–18. New York: Rowman & Littlefield. Enslin, E. 1994. Beyond writing: Feminist practice and the limitations of ethnography. Cultural Anthropology 9: 537–68. Etienne, M. 1980. Women, men, cloth and colonization: The transformation of productiondistribution relations among the Baulé (Ivory Coast). In Women and colonization, ed. M. Etienne and E. B. Leacock, 214–38. New York: Praeger. Etienne, M., and E. B. Leacock. 1980. Introduction: Women and anthropology, conceptual problems. In Women and colonization, ed. M. Etienne and E. B. Leacock, 1–23. New York: Praeger. Ewick, P. 1994. Integrating feminist epistemologies in undergraduate research methods. Gender and Society 2: 92–108.

14_107-Bernard.indb 176

6/9/14 10:02 AM

5: Feminist Methods

177

Falk, D. 1999. Brain evolution in females. In Women in human evolution, ed. L. Hager, 114–46. London: Routledge. Fausto-Sterling, A. 1993. The five sexes: Why male and female are not enough. The Sciences 33: 20–25. Flannery, K., and J. Marcus. 1994. On the perils of “politically correct” archeology. Current Anthropology 35: 441–42. Folbre, N. 1993. How does she know? Feminist theories of gender bias. History of Political Economy 25: 167ff. Franklin, S. 1988. Reproductive futures: Recent literature and current feminist debates on reproductive technologies. Feminist Studies 14: 545–61. Franklin, S. 1993a. Essentialism, which essentialism? Some implications of reproductive and genetic techno-science. Journal of Homosexuality 3: 27ff. Franklin, S. 1993b. Postmodern procreation. Science as Culture 17: 522–42. Franklin, S. 1997. Dolly: A new form of transgenic breedwealth. Environmental Values 6: 427ff. Freeman, C. 2000. High tech and high heels in the global economy: Women, work, and pink-collar identities in the Caribbean. Durham, NC: Duke University Press. Friedl, E. 1975. Women and men: An anthropologist’s view. New York: Holt, Rinehart and Winston. Gacs, U., A. Khan, J. Mcintyre, and R. Weinberg, eds. 1988. Women anthropologists. Westport, CT: Greenwood Press. Gailey, C. Ward. 1980. Putting down sisters and wives: Tongan women and colonization. In Women and colonization, ed. M. Etienne and E. B. Leacock, 294–322. New York: Praeger. Gailey, C. Ward. 1987a. Kinship to kingship: Gender hierarchy and state formation in the Tongan Islands. Austin: University of Texas Press. Gailey, C. Ward. 1987b. Evolutionary perspectives on gender hierarchy. In Analyzing gender, ed. B. Hess and M. Ferree, 32–67. Thousand Oaks, CA: Sage. Gailey, C. Ward. 1998. Making kinship in the wake of history: Gendered violence in U.S. older child adoption. Identities: Global Studies in Culture and Power 5: 249–92. Gailey, C. Ward. 2010. Blue ribbon babies and labors of love: Race, class, and gender in U.S. adoption practice. Austin: University of Texas Press. Gal, S. 2001. Language, gender, and power: An anthropological review. In Linguistic anthropology: A reader, ed. A. Duranti, 420–30. Oxford: Blackwell Publishers. Gal, S., and G. Kligman. 2000. The politics of gender after socialism. Princeton, NJ: Princeton University Press. Gero, J., and M. Conkey, eds. 1991. Engendering archaeology. Cambridge: Basil Blackwell. Ginsburg, F., L. Abu-Lughod, and B. Larkin, eds. 2002. Media worlds: Anthropology on new terrain. Berkeley: University of California Press. Ginsburg, F., and R. Rapp. 1991. The politics of reproduction. Annual Review in Anthropology 20: 311–43. Ginsburg, F., and R. Rapp, eds. 1995. Conceiving the new world order. Berkeley: University of California Press. Golde, P., ed. 1970. Women in the field: Anthropological experiences. Chicago: Aldine. Gottlieb, A. 1988. Blood magic: The anthropology of menstruation. Berkeley: University of California Press. Gough, K. 1959. The Nayars and the definition of marriage. Journal of the Royal Anthropological Institute 89: 23–34. Grewal, I., and P. Caplan, eds. 1994. Scattered hegemonies. Minneapolis: University of Minnesota Press.

14_107-Bernard.indb 177

6/9/14 10:02 AM

178

Christine Ward Gailey

Gruenbaum, E. 1982 [1993]. The movement against clitoridectomy and infibulation in Sudan: Public health policy and the women’s movement. Medical Anthropology Newsletter 13(2). Reprinted in 1993 in Gender in cross-cultural perspective, ed. C. Brettell and C. Sargent, 411–22. Englewood Cliffs, NJ: Prentice Hall. Hafez, S. 2011. An Islam of her own: Reconsidering religion and secularism in women’s Islamic movements. New York: NYU Press. Hafez, S. 2012. No longer a bargain: Women, masculinity, and the Egyptian uprising. American Ethnologist 39: 32–37. Hager, L. 1999. Sex and gender in paleoanthropology. In Women in human evolution, ed. L. Hager, 1–28. London: Routledge. Hanks, W. 2005. Pierre Bourdieu and the practices of language. Annual Review of Anthropology 34: 67–83. Haraway, D. 1989. Primate visions: Gender, race, and nature in the world of modern science. New York: Routledge. Harding, S. 1987. Is there a feminist method? In Feminism and methodology, ed. S. Harding, 1–14. Bloomington: Indiana University Press. Harding, S. 1991. Whose science? Whose knowledge? Ithaca, NY: Cornell University Press. Harrison, F. 1991. Decolonizing anthropology. Washington, DC: Association of Black Anthropologists/American Anthropological Association. Heiman, R., and C. Freeman, eds. 2012. The global middle classes: Theorizing through ethnography. Santa Fe: School for Advanced Research Press. Hobson, J. 2012. Body as evidence: Mediating race, globalizing gender. Albany: SUNY Press. Hoff, L. A. 1990. Theoretical and methodological issues. In Battered women as survivors, ed., 243–55. New York: Routledge. Hrdy, S. Blaffer. 1981. The woman that never evolved. Cambridge, MA: Harvard University Press. Jacobs, S.-E., ed. 1971. Women in perspective. Urbana: University of Illinois Press. Jaggar, A. 1991. Feminist ethics: Projects, problems, prospects. In Feminist ethics, ed. C. Card, 78–106. Lawrence: University Press of Kansas. Jarrett, R. 1994. Living poor: Family life among single parent, African-American women. Social Problems 41: 30–49. Joyce, R. 2000. Gender and power in Prehispanic Mesoamerica. Austin: University of Texas Press. Joyce, R. 2008. Ancient bodies, ancient lives: Sex, gender, and archaeology. New York: Thames and Hudson. Kaufman, D. R. 1996. Rethinking, reflecting, rewriting: Teaching feminist methodology. The Review of Education/Pedagogy/Cultural Studies J8: 165–74. Krige, E. Jenson, and J. D. Krige. 1943. The realm of a rain queen. London: Oxford University Press. Lamphere, L. 1974. Strategies, cooperation, and conflict among women in domestic groups. In Women, culture, and society, ed. M. Rosaldo and L. Lamphere, 97–112. Stanford: Stanford University Press. Lamphere, L. 1977. Review essay: Anthropology. Signs 2: 612–27. Lamphere, L. 1987. From working daughters to working mothers: Women in a New England industrial community. Ithaca, NY: Cornell University Press. Lamphere, L. 1993. The domestic sphere of women and the public world of men: The strengths and limitations of an anthropological dichotomy. In Gender in cross-cultural perspective, ed. C. Brettell and C. Sargent, 67–76. Englewood Cliffs, NJ: Prentice Hall. Lamphere, L., H. Ragoné, and P. Zavella, eds. 1997. Situated lives. New York: Routledge. Lancaster, J. 1973. In praise of the achieving female monkey. Psychology Today 7: 30–36ff.

14_107-Bernard.indb 178

6/9/14 10:02 AM

5: Feminist Methods

179

Lancaster, J. 1991. A feminist and evolutionary biologist takes a look at women. Yearbook of Physical Anthropology 43: 1–11. Leacock, E. B. 1954. The Montagnais’ “hunting territory” and the fur trade. Washington, DC: American Anthropological Association Memoirs 78. Leacock, E. B. 1972. Introduction. In Frederick Engels: The origins of the family, private property, and the state, ed. E. B. Leacock, 7–67. New York: International Publishers. Leacock, E. B. 1977a. The changing family and Lévi-Strauss, or whatever happened to fathers? Social Research 44: 235–59. Leacock, E. B. 1977b. Women, development, and anthropological facts and fictions. Latin American Perspectives 4: 8–17. Leacock, E. B. 1981. Myths of male dominance. New York: Monthly Review. Leacock, E. B. 1983. Interpreting the origins of gender inequality: Conceptual and historical problems. Dialectical Anthropology 7: 263–84. Leacock, E. B., and J. Nash. 1977. Ideologies of sex: Archetypes and stereotypes. In Annals of the New York Academy of Sciences 285: Issues in cross-cultural research, ed. L. Adler, 618–45. New York: New York Academy of Sciences. Leacock, E. B., and H. Safa, eds. 1986. Women’s work. South Hadley, MA: Bergin and Garvey. Leibowitz, L. 1983[1975]. Origins of the sexual division of labor. In Woman’s nature, ed. M. Lowe and R. Hubbard, 123–47. Elmsford, NY: Pergamon Press. Lewin, E. 1993. Lesbian mothers: Accounts of gender in America. Ithaca, NY: Cornell University Press. Lorde, A. 1990. Age, race, class, and sex: Women redefining difference. In Out there: Marginalization and contemporary culture, ed. R. Ferguson, M. Gever, T. Minh-Ha, and C. West, 281–87. Cambridge, MA: MIT Press. Lurie, N. 1966. Women in early American anthropology. In Pioneers of American anthropology: The uses of biography, ed. J. Helm, 29–83. Seattle: University of Washington Press. Lutz, C. 1990. The erasure of women’s writing in sociocultural anthropology. American Ethnologist 17: 611–27. MacCormack, C., and M. Strathern, eds. 1987 [1980]. Nature, culture, and gender. Cambridge: Cambridge University Press. Malkki, L. 2007. Tradition and improvisation in ethnographic fieldwork. In Improvising theory: Process and temporality in ethnographic fieldwork, ed. A. Cerwonka and L. Malkki, 162–88. Chicago: University of Chicago Press. Mankekar, P. 1999. Screening culture, viewing politics: An ethnography of television, womanhood, and nation in postcolonial India. Durham, NC: Duke University Press. Martin, E. 1992. The woman in the body. Boston: Beacon Press. Mascia-Lees, F. 2010. Gender and difference in a globalizing world: Twenty-first-century anthropology. Long Grove, IL: Waveland. Mascia-Lees, F., P. Sharpe, and C. Ballerino Cohen. 1989. The postmodernist turn in anthropology: Cautions from a feminist perspective. Signs 15: 7–33. McClaurin, I., ed. 2001. Black feminist anthropology: Theory, politics, praxis, and poetics. New Brunswick, NJ: Rutgers University Press. Minh-Ha, T. 2005. Surname Viêt, given name Nam. New York: Women Make Movies. Minh-Ha, T. 2009. Woman, native, other: Writing postcoloniality and feminism. Bloomington: University of Indiana Press. Modell, J. 1994. Kinship with strangers: Adoption and interpretations of kinship in American culture. Berkeley: University of California Press. Montoya, R., L. J. Frazier, and J. Hurtig, eds. 2002. Gender’s place: Feminist anthropologies in Latin America. New York: Palgrave Macmillan.

14_107-Bernard.indb 179

6/9/14 10:02 AM

180

Christine Ward Gailey

Morgan, L. 1993 [1989]. When does life begin? A cross-cultural perspective on the personhood of fetuses and young children. In Talking about people: Readings in contemporary anthropology, ed. W. Haviland and R. Gordon, 28–38. Mountain View, CA: Mayfield. Morgen, S. 1989. Introduction. In Gender and anthropology: Critical reviews for research and teaching, ed. S. Morgen, 1–20. Washington, DC: American Anthropological Association. Mullings, L. 1986. Uneven development: Class, race, and gender in the United States before 1900. In Women’s work, ed. E. B. Leacock and H. Safa, 41–56. South Hadley, MA: Bergin and Garvey. Mullings, L. 1997. On our own terms: Race, class, and gender in the lives of African American women. London: Routledge. Mullings, L., and A. Wali. 2001. Stress and resilience: The social construction of reproduction in central Harlem. New York: Kluwer Academic Publishers. Nagengast, C. 1994. Violence, terror, and the crisis of the state. Annual Review in Anthropology 23: 109–36. Nanda, S. 2000. Gender diversity: Crosscultural variations. Prospect Heights, IL: Waveland. Narayan, K. 1995. Participant observation. In Women writing culture, ed. R. Behar and D. Gordon, 33–48. Berkeley: University of California Press. Nash, J., ed. 1985. Women and change in Latin America. South Hadley, MA: Bergin and Garvey. Nash, J., and M. P. Fernandez-Kelley, eds. 1983. Women and men and the international division of labor. Albany: SUNY Press. Oboler, R. Smith. 1980. Is the female husband a man? Woman/woman marriage among the Nandi of Kenya. Ethnology 19: 69–88. Okeyo, A. P. 1980. Daughters of the lakes and rivers: Colonization and the land rights of Luo women. In Women and colonization, ed. M. Etienne and E. B. Leacock, 186–213. New York: Praeger. Omvedt, G. 1977. On the participant study of women’s movements: Methodological, definitional, and action considerations. In The politics of anthropology: From colonialism and sexism toward a view from below, ed. G. Huizer and B. Mannheim, 373–93. The Hague: Mouton. Ortner, S. 1974. Is female to male as nature is to culture? In Woman, culture, and society, ed. M. Rosaldo and L. Lamphere, 67–88. Stanford: Stanford University Press. Ortner, S. 1981. Gender and sexuality in hierarchical societies. In Sexual meanings: The cultural construction of gender and sexuality, ed. H. Whitehead and S. Ortner, 359–409. Cambridge: Cambridge University Press. Padilla, M., J. Hirsch, M. Munoz-Laboy et al., eds. 2007. Love and globalization: Transformations of intimacy in the contemporary world. Nashville: Vanderbilt University Press. Patterson, T. 1995. Toward a social history of archaeology in the United States. Fort Worth, TX: Harcourt Brace. Paulmé, D, ed. 1971. Women of tropical Africa. Berkeley: University of California Press. Pica, T., L. Holliday, and N. Lewis. 1991. Language learning through interaction: What role does gender play? Studies in Second Language Acquisition 13: 343–61. Quinn, N. 1977. Anthropological studies on the status of women. Annual Review in Anthropology 6: 181–225. Ragoné, H., and F. Twine, eds. 2000. Ideologies and technologies of motherhood. New York: Routledge. Rapp, R. [Reiter], ed. 1975. Toward an anthropology of women. New York: Monthly Review. Rapp, R. [Reiter]. 1977. The search for origins: Unraveling the threads of gender hierarchy. Critique of Anthropology 3: 5–24. Rapp, R. [Reiter]. 1978. Review essay: Anthropology. Signs 4: 497–513. Reinharz, S., ed. 1992. Feminist methods in social research. New York: Oxford University Press.

14_107-Bernard.indb 180

6/9/14 10:02 AM

5: Feminist Methods

181

Reiter, R. See Rapp, R. Richie, B., and V. Kanuha. 1997. Battered women of color in public health care systems: Racism, sexism, and violence. In Through the prism of difference, ed. M. Baca Zinn, P. HondagneuSotelo, and M. Messner, 121–29. Boston: Allyn and Bacon. Riley, C. 1988. American kinship: A lesbian account. Feminist Issues 8: 75–94. Rohrlich Leavitt, R., B. Sykes, and E. Weatherford. 1975. Aboriginal woman: Male and female perspectives. In Toward an anthropology of women, ed. R. Reiter (Rapp), 110–26. New York: Monthly Review. Rosaldo, M., and L. Lamphere, eds. 1974. Women, culture, and society. Stanford: Stanford University Press. Ross, K., ed. 2012. The handbook of gender, sex, and media. Malden, MA: Wiley-Blackwell. Rubin, G. 1975. The traffic in women. In Toward an anthropology of women, ed. R. Reiter (Rapp), 157–210. New York: Monthly Review. Sacks, K. See Brodkin, K. Sanday, P. Reeves. 1981. Female power and male dominance: On the origins of inequality. Cambridge: Cambridge University Press. Sanday, P. Reeves, 1990. Fraternity gang rape. New York: NYU Press. Sanday, P. Reeves. 1995. Pulling train. In Race, class, and gender in the United States, ed. P. Rothenberg, 396–402. New York: St. Martin’s Press. Sanday, P. Reeves. 1996. A woman scorned: Acquaintance rape on trial. New York: Doubleday. Sandstrom, K., and T. Opsal. 2013. Grappling with the dilemmas of feminist ethnography. Journal of Contemporary Ethnography 42: 383–93. Sanjek, R. 1978. The position of women in the major departments of anthropology, 1967–1976. American Anthropologist 80: 894–904. Schlegel, A., ed. 1977. Sexual stratification: A cross-cultural view. New York: Columbia University Press. Schultz, A., and L. Mullings, eds. 2006. Gender, race, class, and health: Intersectional approaches. San Francisco: Jossey-Bass. Segura, D., and P. Zavella, eds. 2007. Women and migration in the U.S-Mexico borderlands. Durham, NC: Duke University Press. Siapno, J. Aquino. 2002. Gender, Islam, nationalism, and the state in Aceh: The paradox of power, co-optation, and resistance. London: Routledge Curzon. Silverblatt, I. 1978. Andean women in the Inca empire. Feminist Studies 4: 37–61. Silverblatt, I. 1988. Women in states. Annual Review in Anthropology 17: 427–60. Slocum, S. 1975. Woman the gatherer: Male bias in anthropology. In Toward an anthropology of women, ed. R. Reiter (Rapp), 36–50. New York: Monthly Review. Stack, C. 1974. All our kin. New York: Harper and Row. Stack, C. 1997. Different voices, different visions: Gender, culture, and moral reasoning. In Through the prism of difference, ed. M. Baca Zinn, P. Hondagneu-Sotelo, and M. Messner, 51–57. Boston: Allyn and Bacon. Stephen, L. 1997. Women and social movements in Latin America: Power from below. Austin: University of Texas Press. Stoler, A. 1991. Carnal knowledge and imperial power: Gender, race, and morality in colonial Asia. In Gender at the crossroads of knowledge, ed. M. di Leonardo, 51–102. Berkeley: University of California Press. Strathern, M. 1972. Women in between: Female roles in a male world: Mount Hagen. New Guinea. London: Seminar Press. Strathern, M. 1992. Reproducing the future: Essays on anthropology and the new reproductive technologies. New York: Routledge.

14_107-Bernard.indb 181

6/9/14 10:02 AM

182

Christine Ward Gailey

Susser, I. 1991. The separation of mothers and children. In The dual city, ed. J. Mollenkopf and M. Castells, 207–25. New York: Russell Sage Foundation. Susser, I. 1993. Creating family forms: The exclusion of men and teenage boys from families in the New York City shelter system, 1987–91. Critique of Anthropology 3: 267–85. Susser, I. 1996. The construction of poverty and homelessness in U.S. cities. Annual Review in Anthropology 25: 411–35. Susser, I. 1998. Inequality, violence, and gender relations in a global city: New York, 1986–1996. Identities: Global Studies in Culture and Power 5: 219–47. Susser, I. 2009. AIDS, sex, and culture: Global politics and survival in southern Africa. Malden, MA: Wiley-Blackwell. Susser, I., and M. Gonzalez. 1992. Sex, drugs, and videotape: The prevention of AIDS in a New York City shelter for homeless men. Medical Anthropology 14: 307–32. Tannen, D. 1994. Gender and discourse. New York: Oxford University Press. Tannen, D. 1996. Gender in research on language: Researching gender patterns in classroom discourse. Tesol Quarterly 30: 341ff. Thiam, A. 1986 [1978]. Clitoridectomy and infibulation. In Speak out, black sisters: Feminism and oppression in black Africa, trans. D. Blair, 57–87. London: Pluto Press. Tula, M. T., and L. Stephen. 1994. Hear my testimony: Maria Teresa Tula, human rights activist from El Salvador. Austin: University of Texas Press. Urciuoli, B. 1995. Language and borders. Annual Review in Anthropology 24: 525–47. Valian, V. 1998. Why so slow? The advancement of women. Cambridge, MA: MIT Press. Van Allen, J. 1972. “Sitting on a man”: Colonialism and the lost political institutions of Igbo women. Canadian Journal of African Studies 6: 165–81. Vincent, J. 1990. Politics and anthropology. Tucson: University of Arizona Press. Visweswaran, K. 1997. Feminist ethnographies. Annual Review in Anthropology 26: 591–621. Wagner-Martin, L. 1994. Telling women’s lives: The new biography. New Brunswick, NJ: Rutgers University Press. Walley, C. 1997. Searching for “voices”: Feminism, anthropology, and the debate over female genital mutilation. Cultural Anthropology 12: 405–38. Wang, C., M. A. Burris, and X. Yue Ping. 1996. Chinese village women as visual anthropologists: A participatory approach to reaching policymakers. Social Science and Medicine 42: 1391–400. Ward, J. 2010. Gender labor: Transmen, femmes, and the collective work of transgression. Sexualities 13: 236–54. Warren, C., and J. K. Hackney. 2000. Gender issues in ethnography. Thousand Oaks, CA: Sage. Wellhouse, K., and Z. Yin. 1997. “Peter Pan isn’t a girl’s part”: An investigation of bias in a kindergarten classroom. Women and Language 20: 35ff. Weston, K. 1991. Families we choose: Lesbians, gays, kinship. New York: Columbia University Press. Weston, K. 1997. Fieldwork in gay and lesbian communities. In Through the prism of difference, ed. M. Baca Zinn, P. Hondagneu-Sotelo, and M. Messner, 79–85. Boston: Allyn and Bacon. Whitehead, H., and S. Ortner, eds. 1981. Sexual meanings: The cultural construction of gender and sexuality. Cambridge: Cambridge University Press. Whitehead, T., and M. E. Conaway, eds. 1986. Self, sex and gender in cross-cultural fieldwork. Urbana: University of Illinois Press. Wies, J., and H. Haldane, eds. 2011. Anthropology at the front lines of gender-based violence. Nashville: Vanderbilt University Press. Wilson, A. 2002. The transnational geography of sexual rights. In Truth claims: Representation and human rights, ed. M. P. Bradley and P. Petro, 251–65. New Brunswick, NJ: Rutgers University Press.

14_107-Bernard.indb 182

6/9/14 10:02 AM

5: Feminist Methods

183

Wilson, A. 2004. The intimate economies of Bangkok: Tomboys, tycoons, and Avon ladies in the global city. Berkeley: University of California Press. Wolf, M. 1992. A thrice-told tale: Feminism, postmodernism, and ethnographic responsibility. Stanford: Stanford University Press. Wright, R., ed. 1996. Gender and archaeology. Philadelphia: University of Pennsylvania Press. Wylie, A. 1991. Gender theory and the archaeological record: Why there is no archaeology of gender. In Engendering archaeology, ed. J. Gero and M. Conkey, 31–56. Cambridge: Basil Blackwell. Wylie, A. 2004. Why standpoint matters. In The feminist standpoint theory reader: Intellectual and political controversies, ed. S. Harding, 339–52. London: Routledge. Yanagisako, S., and C. Delaney, eds. 1995. Naturalizing power. New York: Routledge. Zavella, P. 1996. Feminist insider dilemmas: Constructing ethnic identities with Chicana informants. In Feminist dilemmas in fieldwork, ed. D. Wolf, 138–59. Boulder, CO: Westview. Zihlman, A. 1985. Gathering stories for hunting human nature. Feminist Studies 11: 364–77. Zihlman, A. 1999. The Paleolithic glass ceiling: Women in human evolution. In Women in human evolution, ed. L. Hager, 91–113. London: Routledge. Zihlman, A., and N. Tanner. 1978. Gathering and the hominid adaptation. In Female hierarchies, ed. L. Tiger and H. Fowler, 163–94. Chicago: Beresford.

14_107-Bernard.indb 183

6/9/14 10:02 AM

14_107-Bernard.indb 184

6/9/14 10:02 AM

CHAPTER 6

Participatory Methods and Community-Based Collaborations Stephen L. Schensul, Jean J. Schensul, Merrill Singer, Margaret Weeks, and Marie Brault This chapter is written from the perspective of the authors’ commitment to an anthropology that seeks to reduce the power imbalance for underserved and marginalized populations in society (J. Schensul 2013; S. Schensul 1973; Singer 1990). We strive to change the model of research from solely scientist directed and implemented to approaches involving members of underrepresented communities and groups as research collaborators. We argue that the science of anthropology lies not in theory or methodology alone but in the application of research results to change processes that validate or modify anthropological theory (S. Schensul 1985) and that both theory development and results must involve the participation of representatives of the affected populations as our partners (Brydon-Miller et al. 2008; J. Schensul and LeCompte 1999, 2002; Singer 2003). In doing so, we disagree with the charge of some that by contributing to the goals and development of underserved communities we may lose our scientific objectivity (S. Schensul 1985). Instead, we claim that the demands for accurate results needed to address real-world problems and improve the human condition parallels and may, at times, far exceed those that are discipline based (J. Schensul et al. 1999). We believe that anthropology, through participant observation and face-to-face interviewing, is rooted in local communities and that place-based research can provide both a grounded perspective on the impact of social and structural factors on community groups and residents and a “ground up” perspective on the world drawn from members of underresourced, marginalized, or otherwise unrepresented communities. We do not claim that the methods required for this work are unique; many are drawn from classic cultural anthropology as well as other social science fields. We do argue that consideration of structural factors and power differentials leads to the formulation of questions developed from the perspective of engaged groups and communities, making results of this kind of research available to them as they seek to advocate for needs and required resources (J. Schensul 2006). We have been involved in conducting applied anthropology in domestic and international communities for many years. Stephen L. Schensul began his applied work in the Mexican American community in Chicago (from 1968 to 1976), continued that work in the Puerto Rican and African American communities in Hartford, Connecticut, and has conducted applied anthropological research in Mauritius, Sri Lanka, and for the last decade in India. Jean J. Schensul began her work in 1974 on educational issues in Chicago, directed research at the Hispanic Health Council from 1979 to 1987, and was director of the Institute for Community Research from 1987 to 2004, continuing as 185

14_107-Bernard.indb 185

6/9/14 10:02 AM

186

Stephen L. Schensul et al.

full-time and senior scientist from 2003 to the present. Merrill Singer was associate director for Research for the Hispanic Health Council from 1987 to 2008 and is currently professor of anthropology at the University of Connecticut. Margaret Weeks joined the Institute for Community Research in 1990 and assumed the directorship in 2007. Marie Brault is an applied researcher-in-training, conducting field research in economically marginal communities in Mumbai, India, as a part of the Ph.D. program in anthropology at the University of Connecticut. Many of the examples in this chapter are drawn from our own experience, both individually and in collaboration, and have been drawn from models developed in Chicago (Van Willigen 1993) and Hartford (J. Schensul 2006, 2013; Singer and Weeks 2005). Many terms, with often overlapping meanings, are used to categorize the approach to anthropology that we describe here, and we have used most of them in publications describing our own work as well as the work of others. The first is “applied anthropology,” a generic term with a long, well-established, and well-criticized history in anthropology (see Agrawal 1981; Rylko-Bauer et al. 2006; S. Schensul 1974; Tax 1958, 1975). Other terms seek to distance the work from applied anthropology, while at the same time attempting to identify the unique focus of each endeavor. Such terms are “action anthropology” (Tax 1958), “engaged anthropology” (Low and Merry 2010), “public anthropology” (Beck 2009; Borofsky 2000; Checker 2009), “advocacy anthropology” (S. Schensul and J. Schensul 1978), “collaborative anthropology” (Austin 2004; Fluehr-Lobban 2008; Geilhufe 1978; Gonzalez 2010; J. Schensul and S. Schensul 1992; J. Schensul and Stern 1985), and terms that have counterparts in other fields, such as “participatory action research” (Berg and J. Schensul 2004) and “community-based, participatory research” (Israel et al. 1998). Below, we describe some of these approaches in more detail. APPLIED ANTHROPOLOGY

“Applied” is the omnibus term for those who use anthropological concepts and methods to contribute directly or indirectly to policy, administration, and intervention in larger systems and in communities. From this perspective, all of the above approaches and terms can be seen as subsets of applied anthropology. The association of applied anthropology with colonization, Japanese relocation, colonial administration, armed conflict, and other forms of working with and in the power structure in the twentieth century resulted in intense criticism within and beyond anthropology (Bennett 1996; Low and Merry 2010; Rylko-Bauer et al. 2006). At the same time, many universitybased anthropologists saw themselves as involved in basic research and resisted the concept of the discipline as forgoing its objectivity and participating in the engineering of change. Anthropologists committed to involvement in societal change sought new labels that would distance them from the negative critiques of applied anthropology and justify the legitimacy of their involvement as academically valid and socially useful for marginalized peoples. ACTION ANTHROPOLOGY

Action anthropology was developed by Sol Tax and his students (1958, 1975) in an effort to overcome the post–World War II critique of applied anthropology as pater-

14_107-Bernard.indb 186

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

187

nalistic. Tax (1975) developed action anthropology with two goals: to help a group of individuals solve social problems and to learn something that could be anthropologically and scientifically useful through the process. To achieve these goals, Tax proposed that anthropologists, particularly those with an academic base, volunteer their services to communities and encourage local residents to choose both the research issues and development directions. Tax and his students in the Fox Project started out to do traditional ethnological research, but the work “conspired to turn us into actionists” (Gearing et al. 1960). The project facilitated the development of a cooperative farming effort, an education project, a scholarship program, and traditional crafts using a method of non-directed facilitation. Tax also sought to engage action anthropologists with policymakers at county, state, and national levels in seeking solutions. His critics, both within his University of Chicago department and in the discipline, argued that Tax’s approach to action anthropology was more like a type of social work (Gross 2002), that it lacked a coherent set of methodologies, and that it did not develop or build on anthropological theory (Bennett 1996). Others questioned whether and to what degree the efforts of Tax and his students were participatory, from the standpoint of indigenous Mesquaki people (Foley 1999). Nonetheless, the terms “action anthropology” and “action research” (see J. Schensul 2010; S. Schensul and Bakszysz 1975) have been used to describe research that supports community advocacy and development (see Action Anthropology blog at: http://openanthcoop .ning.com/group/actionanthropology). PUBLIC ANTHROPOLOGY

Borofsky (2000, 2011) criticizes anthropology as a field no longer interested in producing research relevant to the general public. He has called for a “public anthropology” that researches and disseminates accessible studies that relate to public concerns. Borofsky states that public anthropology should not be concerned with distinguishing between application and theory but should incorporate both application and theory into work relevant for the general society. Singer (2000a), responding to Borofsky, is critical of the contention that anthropologists fail to engage the public and points to the wide variety of work conducted by anthropologists that spans topics ranging from HIV in minority communities to environmental issues to land reform. The call for public anthropology seems more about topical foci and the dissemination of research results to the wider public than new methodologies or orientation to the needs of a local group or community. In other words, the public of concern is not those with whom anthropologists may work to help address their pressing needs but those publics who consume books and other media that thoughtfully address contemporary social issues without necessarily acting on them. Beck and Maida (2013) argue for a more critical definition and approach to public anthropology through its application to problems of inequity and social, economic, and ecological injustice (see also Eriksen, this volume). ENGAGED ANTHROPOLOGY

Low and Merry (2010) use the term “engaged anthropology” to identify various forms of engagement including:

14_107-Bernard.indb 187

6/9/14 10:02 AM

188

Stephen L. Schensul et al.

• sharing and support that emerge from close working and professional relationships that anthropologists develop through participant observation; • teaching and public education; • social critique, using anthropological methods and theories to uncover power relations and structures of inequality; • collaboration, ranging from working with local organizations to co-constructing ethnographies or sharing leadership in research and intervention projects; • advocacy, in which anthropologists work to support the objectives of groups seeking to redress inequities; and • activism, in which anthropologists work side by side with members of groups seeking to address inequities.

Our own work illustrates these different forms of engagement. During field research in Uganda, for example, S. Schensul used his Toyota Land Cruiser to transport women with complicated deliveries to the regional hospital. This ancillary contribution is a common a form of sharing and support that anthropologists provide. Singer’s classes at the University of Connecticut, which present a critical perspective on global health, illustrate the role of teaching and public education in engaged anthropology. So, too, does J. Schensul’s training of inner-city youth to conduct research on topics affecting their lives and conditions they wish to change. Singer’s (2009) perspective on “syndemics” also presents engaged anthropology as social critique. The work of Weeks and colleagues (2010) is collaborative, as they built relationships with the Peking Union Medical College and the Hainan Centers for Disease Control (CDC) to develop and evaluate interventions that protect women in the service industry. S. Schensul and Borrero (1984) are engaged in advocacy to generate a societal response to Hispanic health needs. And the work of the Institute for Community Research and of S. Schensul, J. Schensul, and M. Singer in the Puerto Rican community involves activism to develop social change around health, education, and other issues. This broad definition of engaged anthropology is inclusive, and most anthropologists would see themselves as engaged in some manner. As such, the term is a less useful rubric for classifying an anthropology seeking direct involvement and collaboration for transformative change in local communities. ACTION RESEARCH

In action research, the action is meant to be transformative—that is, to reverse oppressive power structures, organizational hierarchies, or national systems that intentionally or unintentionally raise barriers for advocacy, voice, or decision making. This form of action research has a long tradition in the Americas and elsewhere. In the United States, it is associated with Kurt Lewin (1946; Lewin and Cartwright 1952) and has roots in the civil rights movements of the 1960s and 1970s. Later, action research turned to more focused issues such the environment, housing, and health movements of the 1980s and 1990s. In the late 1960s and early 1970s, S. Schensul and J. Schensul (1979, 1982) mounted an action research and community development approach from Schensul’s base in the Community Mental Health Program of the Westside Medical Complex in Chicago. The approach eventually brought anthropologists and Chicano community activists together

14_107-Bernard.indb 188

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

189

to develop alternative service and cultural institutions during a time when federal resources were becoming available for the development of ethnically targeted programs. The Schensuls and their family lived in the community, became part of community life, and engaged in many activities with community activists to further the visibility and the economic, cultural, and political development of the 18th and 26th Street Chicano neighborhoods. This joint approach merged local knowledge and the tools of ethnographic research to build neighborhood-level social, cultural, and health infrastructures created and administered by professionals based in the neighborhood. Alternative culturally based drug and mental health programs were developed and funded, mental health paraprofessionals were trained in a unique program that was rooted in the diversities of Mexican health and cultural histories in the area, and local groups built community educational institutions, a bilingual public education program, and a women’s health and leadership center (S. Schensul 1974, 1980). The work of these early Chicano leaders, bolstered by formative research and program development, produced a new generation of activists that included elected public officials, a Chicano-oriented public high school, and a well-recognized art museum. The Chicago model established the importance of a long-term commitment and residence in a community in developing new ideas within ongoing community processes (e.g., introducing interventions into ongoing systems). It provided a means of culturally situating interventions and training efforts that could be conducted from a community base with community resources. It demonstrated that institutions and organizations that produced, taught, and promoted community cultures were important in anchoring a community to its history and future development. By drawing anthropology and social science students into action research, it showed that this form of development work could provide an important field school setting, but only if faculty senior anthropologists were involved to make the link between academic training and local program development and activism. Through linkages created with the National Institute of Mental Health, federal research funding could be channeled to a community organization and conducted by teams of community researchers and anthropologists. This model has been adapted elsewhere (Hyland 2005; Li et al. 2003; J. Schensul 2005, 2010; S. Schensul 1980; Singer and Weeks 2005). Many communities are now quite sophisticated in thinking about their need for information and are requesting specific kinds of relationships with researchers. This approach is enhanced by federal efforts to promote community engagement in scientific research through centers for translational science, urban research, prevention research, and the Carnegie initiative to promote engaged scholarship. For example, some HIV and translational research centers fund pilot collaborative studies directly to community organizations that enhance their capacity to take leadership in research that improves services to their constituencies. Further, organizations that accept funding to implement evidencebased interventions are sometimes required to conduct their own evaluations and seek the scientific capacity to carry out these evaluations. Community organizations that conduct research, such as the Institute for Community Research and the Hispanic Health Council, have established their own institutional review boards to review ethical practices in studies.

14_107-Bernard.indb 189

6/9/14 10:02 AM

190

Stephen L. Schensul et al.

COMMUNITY-BASED, PARTICIPATORY RESEARCH

Community-based, participatory research (CBPR) is a term coined by public health researchers to refer to research in which researchers, community organizations, leaders, and individuals living in the community form partnerships that minimize power inequities between researchers and research participants and use research to affect change at the community level (Israel et al. 1998). CBPR in public health, which developed well after action research in anthropology, draws from many of the same sources of inspiration. Acceptance of this approach in public health has been demonstrated by federal agencies including the Agency for Healthcare Research and Quality, which recommended increased use of CBPR in health care research in the United States (AHRQ 2003). CBPR developed out of a growing recognition that traditional population-based biomedical research methodology lacked authentic community roots and often alienated the community from both research and researchers. In certain instances, communities have come to view researchers not as natural allies but as self-interested exploiters of communities. Consequently, in CBPR-guided projects, the community is expected to participate actively in setting the research agenda. In this approach, communities are guided by their experience-based concerns about specific health-related problems and the ways community or local knowledge can be put to use in addressing those health concerns. As the concept of CBPR has developed and its value has been recognized, various institutional efforts have been made to establish adoptable guidelines for successful and mutually satisfying participatory research initiatives (Cook 2008). Like action research projects initiated by anthropologists, CBPR projects use the concept of a local community as the unit of research (Cornwall and Jewkes 1995; Israel et al. 1998). Further, CBPR supports and builds collaborative partnerships for the duration of the project with organizations and individuals within the community. Once these relationships have been established, the researcher must ensure that the interests of the community are represented in the research questions. A primary goal of CBPR is to integrate knowledge and action for the benefit of all partners involved. Information is gathered to inform subsequent action. Not every project will have an intervention component, but all CBPR projects strive to affect social change. CBPR also promotes a co-learning and empowering process that facilitates transfer of information and skills between all parties involved. This principle recognizes that marginalized groups have often not had a voice in the research in which they have participated. CBPR-oriented researchers acknowledge the inequalities that may exist between researchers and community participants and attempt to ameliorate these inequalities through an emphasis on information, resource, and skill sharing. CBPR also strives to be iterative and reflexive throughout the research process. As information and data are collected, researchers and community members refine methodologies and determine future actions. Research findings are disseminated to all collaborators in language that is understandable and accessible to all collaborators. This principle includes discussion of materials and findings with research participants prior to submission for publication and acknowledgment of the contributions of community collaborators. Research findings are to be disseminated to all collaborators, often in public forums or other public arenas (Cornwall and Jewkes 1995; Israel et al. 1998).

14_107-Bernard.indb 190

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

191

Both the process of CBPR, or any participatory research, and the dissemination of results may involve disagreements and require conflict negotiation. For example, as Louis Marcelin and colleagues’ work in Haiti showed (Marcelin et al. 2008), unaligned sectors of involved communities were in significant disagreement with their study of attitudes about community development in Cité Soleil. When the researchers first presented their findings in a community forum, participants were sharply splintered along preexisting lines. In the end, however, seeing the usefulness of the research and the extensive efforts to involve the community at every step in the process, participants met together to plan a strategy for community advancement. There are many examples of CBPR in the literature; most of them, although collaborative, are missing several main features of anthropologically driven CBPR work. First, many are short- rather than long-term efforts, lasting for the duration of a project but not over long periods of time. There are, however, examples of long-lasting university–community collaborations (Israel et al. 2001a, 2001b; Minkler 2005; Minkler et al. 1997; Minkler and Wallerstein 2011; Parker et al. 2003; Radda and Schensul 2011; Schensul, Radda et al. 2009; Schensul, Robison et al. 2006; Wallerstein and Duran 2006) in public health participatory research. PARTICIPATORY ACTION RESEARCH

Participatory action research (PAR; Hagey 1997) is the systematic data-gathering process involving those affected by a social or health problem in a collaborative effort to conduct research that is then directed toward actions to alleviate the root causes of the problem. Thus, PAR is generally thought of as being transformative in all respects: It is research directed toward structural, policy, or other higher order changes; it is conducted by affected groups; and in the process, it brings about change in individual-level capacity, power, influence, and voice. PAR becomes the basis for the analysis of political structures impeding access of marginalized and oppressed groups, indigenous people, and impoverished and exploited workers, and that analysis is used to spearhead radical political actions. Analysis of oppressive structures stems in large part from the earlier work of Paulo Freire (1970), who used visual images to help rural farmers understand inequitable distribution of resources and ask about the root causes of these inequities. The work was continued by Fals Borda (1979, 1987) and others who used PAR to assist in the mobilization of indigenous populations in Colombia. We can identify several specific approaches to PAR. First, it has been used as a form of critical pedagogy. In this approach, adults and children whose voices have been silenced learn from and with researchers to use ethnographic methods, along with other group and capacity-building techniques, to identify issues that affect them directly, to investigate those issues with peers and others who may hold power over them, and to use the results to advocate for transformational change (Berg et al. 2009; Cammarota 2008; J. Schensul and M. Berg 2004; J. Schensul et al. 2008; Sydlo et al. 2000). This form of PAR, which also addresses concerns about open discussion of research findings, is a way of democratizing science (i.e., conceptualizing theory for action and making it available to those who need information to bring about social change). At the same time, like other forms of PAR, a critical pedagogy approach transforms the researchers, creating opportunities for improving communication and critical analysis skills.

14_107-Bernard.indb 191

6/9/14 10:02 AM

192

Stephen L. Schensul et al.

Because it is always conducted by groups, PAR also forges a group voice, which increases potential for power and influence. A critical pedagogy approach to PAR resembles the political agenda framed by Fals Borda (1987), but at times it places more emphasis on developing critical consciousness through collective efficacy, which may or may not entail specific political action. Second, PAR can focus on community development. This approach has been advanced in the United States and internationally by the work of William Foote Whyte and colleagues (1991; Greenwood et al. 1993). It engages university-based intellectuals or politicized scholars working collaboratively with marginalized or underserved communities on issues of infrastructure (housing, economic development, and education), political inclusion, and voice. They explore ways of transferring research technology such that community organizations have more control over the research process. These researchers are empirical in their approach, using research technology to address community questions over time, and direct their attention to communities experiencing inequities. By collaboration, they mean integrating local community knowledge into policy through PAR (Davis and Reid 1999). Third, feminist PAR is rooted in social psychology, phenomenology, empowerment theory, and the social construction of knowledge. It focuses on the co-construction of knowledge based on lived experience. Most feminist PAR scholar activists work with relatively small groups of women to assist them, using reflexive research techniques, and autoethnography, including performance ethnography (performing their lives), to help them reflect together on barriers to self-efficacy, personal and group achievement, and external structural and political dynamics that restrict their voices (Gatenby and Humphries 2000; Goodman et al. 2007; Maguire 2006; Reid et al. 2006). Many, though not all forms of PAR utilize a critical perspective, emphasizing structures and processes through which power is exerted in smaller and larger social structures. Baer et al. (2003) refer to these structures and processes as the vertical linkages that tie a social group to the larger regional, national, and even global human society in often unequal social, political, and economic relationships. J. Schensul (2013) uses the term “eco-critical” to acknowledge the ways in which power differentials shape the interaction of individuals, groups, and organizations at various levels within a social ecosystem. Here we see an affinity between PAR and critical anthropology, including critical medical anthropology, which emphasizes structures of power and inequality in health or other systems that reinforce inequalities worldwide (Baer et al. 2003). Critical medical anthropology addresses the social origins of health and illness, including the ways in which poverty, discrimination, stigmatization, trauma, and other forms of structural violence contribute to poor community health and well-being. PAR can be an effective way to examine and address these structured inequalities. A SHIFT TO TRANSFORMATIONAL RESEARCH

Over time, we have moved toward the term “transformational research” (TR) in referring to forms of applied anthropology that address community or group inequity that result from unequal distribution of political, economic, social, and informational resources. By transformation, we mean acting to bring about changes in social structures to equalize power relationships and improve access to resources and power. The

14_107-Bernard.indb 192

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

193

concept also brings with it the notion of personal and group transformation that shifts consciousness to recognize more fully the roots of inequality and the responsibility of the researchers to collaborate with communities and groups in producing results and generating policy implications and interventions. The principles of transformational research emerging from anthropological experience include: 1. the use of anthropological research to reduce imbalances in political power and economic resources, limited access to needed resources, and environmental constraints; 2. developing and building personal and enduring relationships with members of local communities experiencing inequities; 3. engaging local groups and community members in research as an approach to strengthening community organizing and community development; 4. sharing knowledge and research technology resources; 5. involving researchers in community activism, development, and transformational change as a means of solidifying relationships with communities, shaping and transforming their sociopolitical perspectives, and improving research; 6. enhancing the ability of marginalized populations to access, utilize, and conduct research that addresses their needs and aspirations; 7. co-constructing research with community members and groups that seeks to understand and have maximum effect on remedying social inequities and structural violence; 8. generating empirical data, using a wide range of both qualitative and quantitative methods as a basis for the positions and actions of both community members and anthropologists; and 9. contributing through these processes to anthropological theory, method, and data related to human and community development and transformational change and its limitations.

Personal and political values come into play in taking this stance insofar as it requires a commitment to specific partners, processes, and forms of social change, especially as viewed through the lens of community collaborators. Rigorous empirical research is required to raise critical questions and answer them through the production of results that fare well under scrutiny. This is critical to ensure the credibility of the results, and those presenting them, and to maximize their impact. Finally, anthropologists who take a transformational research approach contribute directly to the goals, aspiration, and actions of the communities with which they seek to collaborate. At the same time, they maintain an active role in anthropology through publications, presentations at anthropology meetings, and elected and appointed positions in anthropological associations. And whenever possible, they introduce their community collaborators to these opportunities to contribute to scholarly work and intellectual debate.

Opportunities for Conducting Transformational Research There are several bases that can provide anthropologists with the freedom and flexibility to operationalize transformational research principles. UNIVERSITY DEPARTMENTS OR CENTERS

University departments or centers offer some flexibility for anthropological involvement in underserved communities or groups. Faculty members have time, student

14_107-Bernard.indb 193

6/9/14 10:02 AM

194

Stephen L. Schensul et al.

resources, and academic freedom to pursue a variety of topics. They must, however, overcome the natural inertia of remaining on campus rather than venturing into communities and succumbing to the pressure for rapid publication and grant getting, the greater value sometimes accorded a distant field site to one that is local, and the natural delays and need for negotiation that must accompany this kind of community and community involvement. Anthropologists who seek to be involved both in the university and in the community must commit significant time and effort to both initiatives. Effective programs that seek to implement transformational research cannot just send out students to communities but must also have faculty engagement so that students can learn under the guidance (apprenticeship) of an experienced anthropologist. Courses that integrate in-class examination of anthropological theory and practice with field-based hands-on guided experience in action research in the community can provide a powerful learning approach for students. Community service learning, which has gained considerable attention within the academy in recent years, provides one context for this approach to merging engaged pedagogy and transformational research (Butin 2010). INTERNATIONAL ORGANIZATIONS

Intervention programs are frequently based in international organizations (WHO, UNFPA, UNICEF, World Bank), bilateral organizations (USAID, CIDA, GTZ), and private voluntary organizations (PVOs such as PATH, Pathfinder, John Snow, Bill and Melinda Gates Foundation, Ford Foundation, Rockefeller Foundation, and many others). Anthropologists have been playing prominent roles in these organizations since the 1980s. Their assignments can include a specific set of responsibilities for formative research, adapting the intervention to the cultural context and evaluating outcomes. Under some circumstances, anthropologists can define part of their role to include community involvement and community facilitation (see S. Schensul 1974). However, the flexibility to engage and develop approaches that meet action research guidelines are achieved only when anthropologists have a clear view of their skills and identities, a set of objectives, and the opportunities to negotiate flexibility into their roles. COMMUNITY-BASED RESEARCH INSTITUTES

Institutes such as the Hispanic Health Council and the Institute for Community Research in Hartford provide both a community base for and a strong commitment to empirically based transformational research (see J. Schensul 2010; S. Schensul and J. Schensul 1978; Singer 2003; Singer et al. 1990). The original mission of the Hispanic Health Council was to conduct research training and advocacy to improve the health of the Puerto Rican community of Hartford. Over its 30-year history, the relative role and importance of research has shifted, depending on funding, personnel, and the need for community service. Its goal, however, remains to integrate research and services to promote health improvements among Latinos in central Connecticut. The current mission of the Institute for Community Research is to conduct research in collaboration with community partners to promote justice and equity in a diverse, multiethnic, multicultural world. It engages and supports community-based research

14_107-Bernard.indb 194

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

195

partnerships to reverse inequities, promote positive changes in public health and education, and foster cultural conservation and development. For the past nearly 25 years, the Institute for Community Research (ICR), together with many partners, has experimented with a variety of different approaches to carrying out research, integrated with education, training, policy promotion, participatory action research, and community cultural development. Both organizations represent community-based research organizations (CBRO), a type of institution that places research goals, methodologies, and results in the hands of community constituencies. There are a growing number of activist-oriented community-based research and evaluation organizations that, if organized, would offer a compelling force for social justice and transformational change (J. Schensul 2010). The CBRO model allows for flexibility in hiring practices, administrative modalities, collaborative structures, and the use of research for social justice purposes.

Methods for Transformational Research Transformational research must use an empirically based methodology that triangulates qualitative and quantitative data. All of the specific data collection methods in this handbook could be used within a framework of transformational research. However, there are several unique features of the ways it is organized and implemented that constitute an identifiable methodology for this approach to the applied study of the human condition. ENTRY INTO THE COMMUNITY

The process of entering a community for conducting research has been well described in anthropology (J. Schensul and LeCompte 2013; S. Schensul et al. 1999, 2013). For the most part, anthropologists provide a general description of the goals and methods of their research and convey these explanations to community gatekeepers, leaders, and key informants. These explanations are quickly forgotten as these potential research allies and facilitators focus more on the personality and motivation of the field researcher. Over time, the development of rapport and relationships and the researcher’s trustworthiness can create an environment that allows anthropologists to proceed with their own fieldwork. There are instances in the course of researcherdirected fieldwork when a group or community might ask an anthropologist to provide data and input that might address a need for intervention or advocacy (called “sharing and support” in engaged anthropology or “testimony” in advocacy anthropology). In transformational research, anthropologists are not interested simply in entering the field to study an issue or question of their own choice. The transformational researcher enters a community seeking collaborative relationships with groups experiencing injustices or disparities, to identify with them a range of issues that are problematic for them. These issues may be shaped by the limitations of the anthropologist’s organizational base, interests, and community priorities. Information on these issues and the individuals and groups addressing them can be collected through key informant interviews, observation at governmental, organizational, and community meetings, and available archival and secondary data (e.g., school performance, court and police records, prior surveys). This initial phase can identify competing sectors and organizations, factions, and differential approaches to change as well as partners for change.

14_107-Bernard.indb 195

6/9/14 10:02 AM

196

Stephen L. Schensul et al.

For example, S. Schensul’s entry into the Mexican American community on the west side of Chicago (S. Schensul 1973) was facilitated by a key informant who arranged meetings with key community activists and provided entry into community meetings, bars, and recreational settings These contacts and settings made it clear that official Chicago had failed to recognize the growing Hispanic population from the point of view of their health, educational, and recreational needs. Ten years later, when S. Schensul entered the Puerto Rican community in Hartford, key informants presented the same picture in the realms of health, housing, and education (S. Schensul and Borrero 1984). Margaret Weeks entered the AIDS community by seeking out relationships with all AIDS-related organizations in the Hartford area. Similarly, Jean Schensul’s efforts to situate the Institute for Community Research by developing a support base for research in a skeptical urban environment required meetings and negotiations with numerous community organizations representing different ethnic/racial and special interest constituencies over a two-year period, after which it became relatively easy to enter into research collaborations. DEVELOPING RELATIONSHIPS WITH CITIZEN ACTION GROUPS

Transformational researchers seek to use the results of the research for social change or social justice ends. For this reason, they look for partners who are interested in using information to advocate for and introduce change strategies at multiple levels. These partners and collaborators may be individuals, but it is more likely that like-minded individuals from the community have banded together in citizen groups or community-based organizations with an agenda of service and change. Once these groups have been identified, anthropologists need to develop relationships and rapport with them and identify opportunities to demonstrate the contributions of research and information gathering that mesh with the agenda of the group. Relationships start with attending meetings, conducting key informant interviews with leaders, identifying group goals, understanding the group’s history and functioning, and building rapport, while finding opportunities to demonstrate the utility of research. Every community has issues and residents who have formed or will form collectives to address those issues. Regardless of whether the scope of the problem is big or small, an action researcher needs to engage with these individuals and collectives. The engagement may not be about what the researcher is interested in or focused on, but rather about any issue in which individuals and groups are interested. The most important element is that the anthropologist shows interest in community needs and remembers that even if the research is in a topic area with which the anthropologist is unfamiliar, partnerships with other researchers can be sought to fill the gap. The engagement is most likely to start with small-scale contributions that researchers can deliver relatively easily. In doing so, transformational researchers develop rapport and relationships with members of the group, become familiar with group dynamics, learn about how advocacy can occur, and begin a process of joint planning for both the research and the action. It has been our experience that one actionable issue evolves into others, bringing about long-term relationships that can generate successful outcomes. As one example, S. Schensul began his work in the Department of Community Medicine at the University of Connecticut School of Medicine, seeking to link the

14_107-Bernard.indb 196

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

197

school with citizen groups involved in health action. One of them was a public housing tenants’ group addressing community health problems. At an initial meeting, tenants determined that we could be helpful with the “rat problem.” Tenants were being accused by the Hartford Housing Authority and the City Health Department of throwing garbage on the streets rather than in the available dumpsters. Tenants, in turn, complained that the Housing Authority failed to respond to requests to pick up the garbage or repair what residents believed were leaky dumpsters, leading to the rat problem. Schensul and anthropology students were asked by tenant leaders to identify and document the leaky dumpsters and the rat holes near the dumpsters. At 5:00 a.m., they photographed the dumpsters as they were emptied by the private garbage service, documenting garbage spewing from massive holes on the sides and bottom. These photographs and the mapping of rat holes were the data they presented at a meeting of the City Health Department. The outcome of the meeting resulted in the provision of new dumpsters to the community and a change in the dumpster contractor. The success of this event showed tenants that they could advocate for their own needs and use research as a weapon in this process. This action expanded the tenants’ group, involved both faculty and students of the University of Connecticut, and, after three years of struggle, led to the establishment of a Community Health Center in the housing project. MUTUAL LEARNING

Anthropology has a long tradition in which the field researcher becomes a learner while becoming integrated into the study community. A relationship that involves the researcher in regular contact in a citizen’s action group provides an opportunity for mutual learning. Anthropologists give information on formulating research questions, systematic data collection methodology, opportunities for systematic data collection to contribute to action and training of group, and community-based organization (CBO) personnel to participate in all aspects of the research process. At the same time, involvement in the organization offers the anthropologist an opportunity to learn about the power structure in a local area, what issues can be addressed, what approaches to advocacy might be effective, how to disseminate research results and to whom, how to transform research results into action, and what the successes and failures of underrepresented groups working with the researchers have been in seeking transformational change. TIMING AND ACTION

Anthropologists involved in basic research in the academy are generally under time pressure to generate results and publications primarily for the promotion and tenure process. Transformational researchers, however, must respond in a timely and sometimes immediate way to another form of pressure: to maximize the impact of their data and results so that community groups achieve their desired outcomes. Events that have profound implications for the community—such as the imminent dislocation of residents, a bill being heard in a legislative body that affects health care or educational choice, a change in school policy, or an incident involving discrimination at a local hospital—

14_107-Bernard.indb 197

6/9/14 10:02 AM

198

Stephen L. Schensul et al.

generally require rapid organizing, public action, and a series of time-limited outcomes. The transformational researcher, committed to the objectives that have been collaboratively established with community and other groups, must deliver research results at the right time and in the right format. S. Schensul was well into his work in the Mexican American community on the west side of Chicago in the 1970s when he was asked by a community group to develop a research plan that could document problems in the implementation of the Teaching English as a Second Language (TESL) program. In response, he devised a methodology requiring three months of observations in the classroom, in-depth interviews with school administrators, teachers, and parents, and a month of data analysis and report writing. The community group appreciated the effort but pointed out that they wanted to document the implementation of the program and present the results to the U.S. Civil Rights Commission meeting in Chicago in two weeks. In what was later described as commando research (see S. Schensul 1978), the group decided to organize teams of three persons, with each including a member of the research staff. Watches were synchronized and each team presented itself at exactly 11:00 a.m. in each school in the district and asked to see the TESL classroom. In one hour, the teams in the 18 schools had talked with the principals, observed the classrooms, and conducted brief interviews with both the teachers and the students. The results showed that TESL was being poorly implemented, teachers were monolingual English speakers, classrooms were in basement boiler rooms and other dilapidated areas of the school, and both teachers and children complained about learning and teaching limitations. A plan for data collection that was originally planned for three months was more effectively implemented through an hour of data gathering and a week of data analysis and brief report writing. The process provided a compelling case of TESL noncompliance, leading the U.S. Civil Rights Commission to mandate that the Chicago Public School System upgrade implementation. This example illustrates the use of rapid research methodologies that can facilitate some action research. Rapid ethnographic assessment methods, which developed as streamlined versions of more traditional ethnographic research, have been used internationally in public health and in response to other community issues for many years. Rapid assessment techniques include qualitative data collection strategies such as direct observation of behavior, in-depth interviews, and informal interviewing in the field setting. Often these approaches are complemented by the collection of survey data and the examination of secondary data sources (e.g., government records, public health information, media reports). Commonly, rapid assessments involve communities in the formulation of research issues, data collection, data analysis, and the translation of findings into action initiatives (Beebe 2001; Scrimshaw and Gleason 1992; Trotter and Singer 2007). However, ongoing commitment to a group or community consists of both rapid and long-term research that builds a body of data and trusted collaborations available both for current and future challenges. At the same time, community collaboration may slow the process of moving from data collection to results. S. Schensul, principal investigator of a grant to the Hispanic Health Council, sought to speed up the process of analysis of survey data in the Puerto Rican community by taking it back to the university to work on with colleagues. The

14_107-Bernard.indb 198

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

199

council staff responded that they wanted to be full participants and that the data and analysis should take place at the council. S. Schensul reluctantly agreed and, consequently, several months were added to the time required for analysis and dissemination of the data. On the other hand, joint analysis resulted in better interpretations and a stronger sense of ownership of the data. FORMULATING DATA COLLECTION ACTIVITIES

The transformational researcher involved in relationships with communities and groups needs to develop a research agenda to do two things: (1) gain a broad-based understanding of the context of both the community or local setting and its relationship to the wider society; and (2) focus on targeted topical issues consistent with the action agenda of the community or group or new issues that arise unexpectedly, either through new or unexpected events or from ongoing research. Data can be collected through key informants, in-depth interviews with residents and members of the group, primary and secondary survey data that characterize the community and its various subpopulations, and documentation of the resources available to the community from the political, economic, and service sectors. J. Schensul (2012) refers to this as community-level assessments. For example, the Hispanic Health Council received a grant from the Office of Minority Health of NIMH to document the health and mental health needs of the Puerto Rican community in Hartford. One question on the study’s survey focused on contraceptive use. The first set of 30 respondents indicated very low use of reversible contraceptives but a very high level of sterilization, particularly among women younger than 21 years of age. The survey was refined with more extensive questions about sterilization such as timing and consent. The survey found that sterilization was highly prevalent, that many women were sterilized after their second child at an age younger than 21, and that many women were unclear that the procedure was, in most cases, irreversible. The collection of these survey data and the follow-up qualitative interviews were presented at a press conference and a legislative subcommittee, leading to significant changes in the laws and procedures associated with sterilization (S. Schensul et al. 1982). A three-year study of HIV exposure among residents of senior housing in Hartford and Chicago, conducted by the Institute for Community Research in Hartford, indicated higher-than-anticipated levels of depression symptoms. The research team investigated further by talking with residents and management about depression and isolation. The results showed that depression was indeed a problem that needed to be addressed. An interdisciplinary research team developed another three-year study examining barriers to mental health care in older adult residents of senior housing (J. Schensul, Robison et al. 2006). The study consortium of clinics and older adult advocates expanded the research model and added many questions investigating factors contributing to depression as well as to inability to access care. Questions about social networks were added to try to address the social isolation that advocates mentioned in discussion (Disch et al. 2007). Similarly, the Hispanic Health Council initiated a survey on community perceptions of food insecurity among Latinos, including the availability of food in house-

14_107-Bernard.indb 199

6/9/14 10:02 AM

200

Stephen L. Schensul et al.

holds in the last week of every month. This approach was used because of community complaints about low incomes and how food stamps were insufficient for a full month of family food consumption in an era of rising food prices. Working with community food activists, findings from this research were used to address limited food resources in Hartford (Himmelgreen et al. 2000; Romero-Daza et al. 1999). The specific action agenda of a citizen group will initially structure the data collection activities. These activities will yield new perspectives and issues that become integrated into the action agenda. In addition, broad-based data collection is needed to undergird future activities and, as we have seen, can, by itself, generate action issues. Over time, both the broad-based and the focused data collection activities accumulate a base of data that simultaneously can serve both action needs and contribute to overall anthropological knowledge. SEEKING EXTERNAL FUNDING

The ability to maintain an extensive research agenda that can keep up with advocacy and action on community issues requires staff and financial resources. The recognition in 1974 (the Latina Mother-Infant Research Project awarded to Mujeres Latinas en Acción, Chicago) and 1977 (the Latino Mental Health Research Project, Hispanic Health Council, Hartford) that community groups could be the recipients of NIH funding opened a new era in the way transformational research could be conducted. The recent change allowing joint principal investigators creates the opportunity for multiple institutions (community groups, research institutes, universities) to share more equitably in research grants. The acquisition of research grants requires the transformational researcher to compete against primarily university-based peers without always having the extensive infrastructure required to be successful. Writing competitive research grants to government and foundation sources calls for a comprehensive understanding of the literature and skill in research design (e.g., randomized controlled trials, quasi-experimental design), intervention development and implementation, and qualitative and quantitative data analysis techniques. At the same time, grants need to link the proposed grant application to the action agenda. The generation of good proposals is a creative process that can involve community partners in conceptualization and methods when done well. Collaborative grant writing always produces more authentic, grounded, and realistically do-able and community-relevant proposals. Community-based grant writing initiatives provide a foundation for unification of the efforts of multiple community organizations. In Hartford, for example, multiple agencies working on issues of HIV came together as consortia to conceptualize and address significant issues in the community including substance use among youth, HIV prevention, substance abuse among pregnant women, depression in older adults, and a variety of other concerns. Anthropologists at the Institute for Community Research and the Hispanic Health Council worked closely together and with researchers from other institutions to articulate community concerns in grants submitted to federal funding agencies (J. Schensul and LeCompte 1999; Singer 2000b; Weeks et al. 1995). In all cases, receipt of the grants facilitated the discussion of other issues of concern in community organizations working for social change, especially with HIV.

14_107-Bernard.indb 200

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

201

TRAINING LAY PEOPLE TO PARTICIPATE IN RESEARCH AND INTERVENTION STUDIES

A key element of transformational research process is including community action groups and their members in all steps of the research process. Members of these groups frequently have lived experience in the issues facing the community, entry into multiple sectors, language capacity, commitment, and recognition that there is more to know and that research can be an important tool in action and development. They lack formal training and degrees in research conceptualization and methods, yet their knowledge of the community and commitment to action make them valuable partners in the research process. Further, Aunger (2004) argues that including community members as researchers can help to identify and correct measurement error by exposing the biases of internal versus external observers. Collaborating with community thus enhances what Aunger refers to as “reflexive ethnographic science.” Mosavel et al. (2011) used journaling as a tool for community researchers in Cape Town, South Africa, to reflect on the research process and to identify unique ethical implications of training lay people to conduct research. It is the responsibility of the transformational researcher to work collaboratively with action leaders who will be collaborators in conceptualizing research and community people who will assist in carrying out the research. The ideal goal is that action leaders come to understand research, frame research questions, and assess appropriate methodology, while researchers understand the dynamics of action and the role that research data can play in enhancing successful outcomes. There are a number of ways in which lay leaders become community researchers. One primary approach is to include integration of community residents as members of a study’s research team. Residents have many skills that can be brought to bear on intervention or basic research projects. These include knowledge of community history and politics, extensive social capital, ability to reach so-called hidden populations (such as drug users, young gay men, or women who have experienced reproductive violence), and a sense of what research tools and approaches might or might not work in their communities of reference. With proper training, their legitimacy with the study population enables them to reach and interview people that “outsider” researchers might be unable to reach regardless of their years in the community. Many organizations, university research centers, and researchers hire community residents as members of research teams. Hierarchies of knowledge and authority, coupled with established pay scales, make it difficult, however, to ensure that community researchers, who may sometimes have less than a high school degree, are paid equitably for their skills or are treated with respect for their knowledge. An action research approach, combined with the effort to situate research within community organizations, can do much to alleviate these structural inequities. Training community residents to do research has deep roots in anthropology, stretching back to Boas’s collaboration with George Hunt. In our own work, it began with training women in Chicago’s west side Mexican American community in the 1970s. Working together, we obtained funding to conduct research on the benefits of continuing the cuarentena, the traditional 40-day period following the birth of a child when mother and child are sequestered and the mother supported by friends and family members as the child begins its new life (Gaviria et al. 1982).

14_107-Bernard.indb 201

6/9/14 10:02 AM

202

Stephen L. Schensul et al.

Both the Institute for Community Research and the Hispanic Health Council have hired and reciprocally trained community researchers as core members of research teams for nearly three decades. Anthropologists J. Schensul, Singer, and Weeks have directed multiple community research initiatives to address pressing health needs including substance abuse, HIV/AIDS, sexually transmitted infections, hepatitis, community violence, depression, female-initiated protection methods, community violence, and other topics. All of these projects involved the training of outreach and intervention staff from the community in multi-method research (e.g., J. Schensul 2006; Singer et al. 1991; Weeks et al. 2006). A second approach is to train community residents to do their own research. We have referred to this approach earlier as PAR for political action, or critical PAR, and the development of critical consciousness and collective self-efficacy. We have described the process as one way of democratizing science (J. Schensul 2002)—that is, making the methods, theories, and results of science available for use by lay or community researchers. Our first effort to implement this approach took place in the context of a National Institute for Education–funded project based at La Casa de Puerto Rico, in Hartford, to train community educational activists to learn research and evaluation methods so as to monitor the newly granted bilingual consent decree. In this three-year project, J. Schensul and a team of community advocates trained more than 35 community educational research activists to do ethnographic research to monitor important educational processes in the community; many of these trainees are still active in the Hartford area and beyond, and several obtained educational leadership doctorates. ICR has used a similar approach with adult activists in a wide variety of its projects. The first of these, funded by a local foundation, involved more than 40 community organizations and resident groups as partners in an alternative census and neighborhood survey process. Resident groups and CBOs in each of 11 city neighborhoods and six municipalities structured their own questionnaires, with help from ICR research staff, to focus on topics of concern to them. They also created an alternative form of the census short form. They engaged in data collection and conversations around data analysis and received the data, plus a report on each neighborhood/ municipality and instructions about how to make use of their neighborhood data set (LeCompte et al. 1992). A second four-year project brought together five women’s organizations and women from different parts of the city of Hartford to define, collect, and analyze data on women for use by women’s advocacy groups, including the project steering community. The women spoke to city and state representatives about the issues on which their research focused; they received college credit for their work, and many went on to become leaders in city political circles (J. Schensul et al. 2008). ICR has also implemented this approach with children and youth for more than 20 years, with federal, state, and foundation funding; more recently, suburban youth have used ethnographic methods to study many issues affecting them, including stress, teen pregnancy, teen dating violence, sexual harassment, abuse, dropping out of school, drug use, hustling, racism, and diverted substances (prescription medications). They have used the results to make public presentations to legislators at the state and local levels, to advocate for

14_107-Bernard.indb 202

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

203

jobs for youth, to speak out against racism, to develop prevention strategies, and to advocate against binge drinking and illicit substance use. ICR studies have also involved community residents in the creation of culturally appropriate intervention materials in a process that matches their local knowledge with a critical analysis of scientific knowledge and the production and implementation of materials that reflect their synthesis of the two (J. Schensul et al. 2008). Recently, a study led by researchers J. Schensul and S. Reisine of the University of Connecticut School of Dental Medicine with older adults in senior housing integrated local and expert knowledge about oral health into a set of interactive health promotion materials and delivered them in social marketing campaigns to large numbers of residents in senior housing in the Hartford area. In a similar example, Margaret Weeks provided leadership for the Risk Avoidance Partnership (RAP) study, funded by the National Institute on Drug Abuse, that was designed to train illicit drug users to become Peer Health Advocates (PHAs) to deliver intervention to their drug-using peers in high risk sites and when it would be most appropriate and effective. The goal was to support the prevention of HIV, hepatitis, sexually transmitted diseases, and other common infectious and chronic diseases and risks among injection drug users and crack smokers in Hartford. The study was designed using intensive ethnography of the process and effects of the PHA training program and the PHA-delivered intervention on their networks, tracking the network characteristics of the PHAs and their contacts for changes in risk and intervention action, and using multiple time point survey and narrative interviews with a cohort of Hartford drug users (PHAs and their contacts). A key feature of the study was PHA involvement in the development of prevention materials (referred to as the RAP Flipbook) to be used by other PHAs in intercept interventions with members of their drug-using networks. A mixed-method and integrated qualitative/quantitative approach with a longitudinal design allowed for a deep examination of both process and outcomes of the intervention at the community and drug-user macro network levels needed to understand the intervention being implemented and its effects (DicksonGomez et al. 2006; Weeks et al. 2006; Weeks et al. 2009). A third initiative involves training community-based organizations to conduct research. Usually, CBOs are interested in understanding their constituencies better or expanding their constituencies, obtaining information on a new problem so they can seek funding to address it, evaluating their own programs, or working more effectively with researchers. Anthropologists have been involved for some time in a variety of efforts to train CBOs to do research for their own use. Internationally, this takes the form of rapid appraisal or community assessment (Theis and Grady 1991). For the most part, the CBPR movement, which emphasizes the engagement of organizations in healthrelated research, does not have a track record of training organizational partners to do their own research, although there are exceptions (e.g., Sanstad et al. 1999). More recently, research capacity building for CBOs has been fostered by the Clinical and Translational Science Awards (CTSAs) movement, under the rubric of community engagement. The more progressive CTSAs, such as those in New York and San Francisco, are providing some ethnographic training to their community-based partners. ICR has offered research training to CBOs in Hartford and New Haven under

14_107-Bernard.indb 203

6/9/14 10:02 AM

204

Stephen L. Schensul et al.

the umbrella of Yale’s Center for Interdisciplinary Research on AIDS, with which it has partnered since its inception. Training has consisted of orientation to ethnography, basic data collection methods and analysis, and participatory and other forms of evaluation. Most recently, ICR has embarked on training with CBOs that provides a framework for critique of and more effective knowledge-based participation in clinical and other forms of controlled trials. Such training prepares CBOs and their clients to develop an informed voice, as science policy is formulated and implemented with respect to advances in clinical medicine (e.g., genetically tailored medicine), new medications, and new forms of medical technology. S. Schensul, Verma et al. (2009) have trained many members of the study community in Mumbai, with special emphasis on observation, key informant and in-depth interviews, free lists and pile sorts, and administration of survey instruments. Malta and colleagues (2008) trained activists from the commercial sex worker population and service sector in Brazil in grant writing to assist them in addressing identified needs related to HIV prevention. DISSEMINATION AND USES OF RESULTS

The publication of research results in the form of articles in peer-reviewed journals, books, and book chapters are an essential part of conducting scientific research, important for a productive image in competing for research grants, and a requirement for individual academic advancement. We strongly urge transformational researchers to publish in these forms as they are crucial to action goals and to individual careers. It is also critical that such work be coauthored with community collaborators who make vital contributions to conceptualization, methods, data collection, and interpretation of findings. For transformational researchers, however, these modes of dissemination are necessary but insufficient. First and foremost, they have a responsibility to broadly disseminate findings in the communities in which research is conducted. This means working with community advisers in identifying the best methods to share findings and the ideal format and language for community dissemination. Further, transformational researchers must not only produce results but be involved in the translation of their policy and intervention iterations. These realms may include: community education (Singer 1993), policymaker awareness (Singer 1993), community and culturally appropriate interventions (S. Schensul, Saggurti et al. 2009), press releases web postings (J. Schensul 2010), pictorial representations (J. Schensul 2010), and gallery exhibits and public programming (J. Schensul 2010). In all these areas, it is most important for the translational researcher to modify the format to enhance comprehension, gauge audience response to allow for further modification in researcher interpretation, and identify strategies for change and intervention that emerge through discussion of results. The results of research for purposes of policy change and intervention programming is always a work in progress if the end goal is community development and positive social change. LONG-TERM RELATIONSHIPS

We have had the opportunity and commitment to work with groups and communities over extended periods of time. S. Schensul worked in the Mexican American

14_107-Bernard.indb 204

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

205

community in Chicago for seven years on-site and continued his contacts with it for many years. In a conference on the “Science of Community Intervention” in Chicago in 2009, Schensul, in collaboration with community colleagues P. Ayala, M. Capoccio A. Giachello, and G. Stern, presented a case study that encompassed over 40 years of community development work in Chicago’s west side Mexican community. S. Schensul was actively involved in inner-city Hartford for 14 years, in Sri Lanka for 15 years, and in India for more than a decade. J. Schensul has been involved in Hartford for more than three decades and India for over a decade, Margaret Weeks has worked in Hartford for two decades and China for two decades, and Merrill Singer for almost three decades in Hartford. With this continuity, we, and our collaborators, have been able to compile the results of short- and longer-term action research projects into a substantial database that can provide ongoing information and measurements of change. Building a base of action-related knowledge and making progress in terms of action simply cannot be fit into one- to two-year time increments. Instead, success in contributing to social changes that make a difference in the lives of residents requires 5–10 years to ensure sustainability and institutionalization. Both timely response and long-term commitment are essential components of transformational research. DISCUSSION

Transformational research as an approach may not be appropriate for all anthropologists. However, for anthropologists who choose this route, working with communities experiencing resource challenges and inequities can be both consuming and highly satisfying. Satisfaction comes from long-established, rewarding, and often life-long personal relationships; the pleasure of doing research that is immediately useful; the positive effects of research on the quality of life of community residents; and the building of community institutions and organizations. At the same time, regardless of the base, transformational research involves simultaneous involvement in multiple sectors, each of which demands considerable time and effort and commitment. Simultaneous involvement in ongoing community-based action research efforts, while continuing to engage in a meaningful relationship with the discipline of anthropology, requires meeting the different needs, expectations, obligations, and outputs of each sector, which frequently do not overlap. The requirements of a transformational research approach mean that anthropologists rarely act alone. Thus, ideas, theories, initiatives, methods, analysis, results, and interpretations require continuous negotiation. These negotiations can lead to delays and modifications in research design, methods of data collection and analysis, and interpretation. In some cases, initiatives are rejected completely. Dissemination and use of the data by communities for purposes of policy change and organizational transformations is never through peer-reviewed articles, book chapters, and academic presentations. Transformational research requires anthropologists to develop modes of dissemination that are amenable to lay audiences of policymakers, community residents, electronic and print media, and specialized groups (e.g., adolescents, teachers, intravenous drug users). Most anthropologists leave graduate school with few skills in this area and must learn on the job.

14_107-Bernard.indb 205

6/9/14 10:02 AM

206

Stephen L. Schensul et al.

Early in his work with the Westside Community Mental Health Program in Chicago, S. Schensul developed a detailed report on the communities in the catchment area for the Board of the Illinois State Psychiatric Institute. He later learned that 5 of the 20 pages had been inadvertently left out of the copies; even more devastating was that not one of the members of the board had noticed, leaving a clear unobtrusive measure of the degree to which it was read. For the most part, anthropologists are unprepared for the rough-and-tumble world of action where just having the data that firmly support a particular position may be important but insufficient to affect actions. S. Schensul, in his early days in the Mexican community on the west side of Chicago, became involved with community activists in the effort to fight displacement of a series of buildings that were to be torn down for a gas station. He collected data on the demography of the residents, their rental costs, the hardships of finding a new residential location, and the lack of need for another gas station in the community given the close proximity of many others. The data were presented to a positive response by the Chicago Zoning Board. In Schensul’s naiveté, he failed to understand the power of the oil companies. A week after the hearing, he passed by the area to find that the buildings were torn down, the gas station put up, and customers were pumping gas. Communities are complex entities with many different sectors. The anthropologist may be committed to work with “the community,” but which sector and what if there are significant differences in point of view and action? For example, the Mexican American community in Chicago in the early 1970s consisted of a majority of first-generation migrants directly from Mexico, with significant numbers of Mexicans from the southwest United States (“bracers,” derived from braceros) as well as increasing numbers of second- and third-generation Chicanos. Most of the activists in the community came from this second- and third-generation group and were committed to the developing “brown power” movement. At the same time, many of the more middle-class bracers held conservative views that clashed with those of the younger activists. The recent migrants in the community kept a low profile, since many of them were undocumented. S. Schensul had entered the community through the activist sector, which matched him in terms of age and interests. It was only after systematic surveys and qualitative data collection that he and the activists recognized the divisions in the community and developed strategies to address the needs of the different sectors. Schensul needed a postgraduate course in power politics and community mobilization, which he acquired over the course of seven years in working with community activists. The issues with which anthropologists who are committed to action research are involved are not easily addressed, and although there may be short-term “victories,” they are, for the most part, long-term “sloughs.” After a 40-year history of anthropological involvement in the Westside Mexican American community, anthropologists and activists concluded that the issues of la migre, bilingual education, gangs, health disparities, culturally sensitive services, and many others showed some improvement but were still very much present in 2009, just as they were in 1969. The health and human problems that stem from structural inequities cannot be addressed entirely at the local level and are likely to take generations or decades to

14_107-Bernard.indb 206

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

207

resolve. Furthermore, the immediate consequences of inequities may not be resolved through taking direct action, but rather through economic development and political mobilization. Anthropologists must understand the long-term nature of change and appreciate the process of action and the development of new institutions and organizations that increase a community’s ability to advocate for its needs and develop resources to meet those needs. Transformational research can make anthropology a more engaged, public, actionoriented discipline, while still making a major contribution to the science of anthropology (S. Schensul 1985). The typical anthropological study involves the development of theory, the testing of that theory through data collection in the field, and the modification of that theory based on the results of data analysis. Transformational research pushes the anthropologist beyond this point to test the adequacy of the modified theory by generating policy change that emerges from the modified theory to be tested in subsequent data collection procedures. These ongoing iterations include the natural validation of theory through action and intervention as experimentation. These iterations must also account for the impact of policy changes and events in the community and in the larger society over time. The long-term commitment of anthropologists and their partners to transformational research allows this sequence of theory, data collection, and intervention to evolve multiple times, leading both to an improved anthropological science and to a well-defined and important contribution to enhance the social and political infrastructure, health, and well-being of underresourced and marginalized or invisible groups and communities. REFERENCES

Agency for Healthcare Research and Quality (AHRQ). 2003. Community-based participatory research: Assessing the evidence. Rockville, MD: Agency for Healthcare Research and Quality. Agrawal, B. C. 1981. An experiment in action anthropology: Yanadi action research project in India. Human Organization 40: 87. Aunger, R. 2004. Reflexive ethnographic science. Walnut Creek, CA: AltaMira. Austin, D. E. 2004. Partnerships, not projects! Improving the environment through collaborative research and Action. Human Organization 63: 419–30. Baer, H. A., M. Singer, and I. Susser. 2003. Medical anthropology and the world system. Westport, CT: Praeger. Beck, S. S. 2009. Public anthropology. Anthropology in Action 16: 1–13. Beck, S. S., and C. A. Maida. 2013. Public anthropology in a borderless world. New York: Berghahn Publishers. Beebe, J. 2001. Rapid assessment process. Lanham, MD: AltaMira. Bennett, J. W. 1996. Applied and action anthropology: Ideological and conceptual aspects. Current Anthropology 37: 23. Berg, M., E. Coman, and J. Schensul. 2009. Youth action research for prevention: A multi-level intervention designed to increase efficacy and empowerment among urban youth. American Journal of Community Psychology 43: 345–59. Berg, M., and J. Schensul, guest eds. 2004. Approaches to conducting action research with youth (Special issue). Practicing Anthropology 26(2). Borofsky, R. 2000 Commentary: Public anthropology. Where to? What next? Anthropology News 41: 9–10. Borofsky, R. 2011. To laugh or cry? Anthropology News 41: 9–10.

14_107-Bernard.indb 207

6/9/14 10:02 AM

208

Stephen L. Schensul et al.

Brydon-Miller, M., et al. 2008. Popular education and action research: Voices from the field. In Handbook on educational action research, ed. S. Noffke and B. Somekh, 495–508. Thousand Oaks, CA: Sage. Butin, D. W. 2010. Service-learning in theory and practice: The future of community engagement in higher education. New York: Palgrave Macmillan. Cammarota, J. 2008. The cultural organizing of youth ethnographers: Formalizing a praxis‐ based pedagogy. Anthropology & Education Quarterly 39: 45–58. Checker, M. 2009. Anthropology in the public sphere, 2008: Emerging trends and significant impacts. American Anthropologist 111: 162–69. Cook, W. 2008. Integrating research and action: A systematic review of community-based participatory research to address health disparities in environmental and occupational health in the USA. Journal of Epidemiology and Community Health 62: 668. doi:10.1136/ jech.2007.067645. Cornwall, A., and R. Jewkes. 1995. What is participatory research? Social Science & Medicine 41: 1667–76. Davis, S. M., and R. Reid. 1999. Practicing participatory research in American Indian communities. The American Journal of Clinical Nutrition 69: 755S–59S. Dickson-Gomez, J., M. Weeks, M. Martinez, and M. Convey. 2006. Times and places: Process evaluation of a peer-led HIV prevention intervention. Substance Use & Misuse 41: 669–90. Disch, W. B., J. Schensul, K. E. Radda, and J. T. Robison. 2007. Perceived environmental stress, depression, and quality of life in older, low income, minority urban adults. In Quality of life in old age, ed. H. Wollenkopf and A. Walker, 151–65. International and Multi-Disciplinary Perspectives Series: Social Indicators Research Series, Vol. 31. Netherlands: Springer. Fals Borda, O. 1979. Investigating reality in order to transform it: The Colombian experience. Dialectical Anthropology 4: 33–55. Fals Borda, O. 1987. The application of participatory action-research in Latin America. International Sociology 2: 329–47. Fluehr-Lobban, C. 2010. Collaborative anthropology as twenty-first-century ethical anthropology. Collaborative Anthropologies 1: 175. Foley, D. E. 1999. The Fox Project. Current Anthropology 40: 171–92. Freire, P. 1970. Pedagogy of the oppressed. New York: Criterion. Gatenby, B., and M. Humphries. 2000. Feminist participatory action research: Methodological and ethical issues. Women’s International Forum 23: 89–105. Gaviria, M., G. Stern, and S. Schensul. 1982. Sociocultural factors and perinatal health in a Mexican-American community. Journal of the National Medical Association 74: 983. Gearing, F., R. Netting, and L. R. Peattie. 1960. Documentary history of the Fox Project, 1948– 1959: A program in action anthropology. Chicago: University of Chicago Press. Geilhufe, N. 1978. Discrimination against women in jail: An analysis of advocacy anthropology. Human Organization 37: 202. González, N. N. 2010. Advocacy anthropology and education working through the binaries. Current Anthropology 51: S249–S258. Goodman, L. A., A. Litwin, A. Bohlig et al. 2007. Applying feminist theory to community practice: A multilevel empowerment intervention for low-income women with depression. In Advancing social justice through clinical practice, ed. E. Aldarondo, 265–90. New York: Routledge. Greenwood, D. J., W. F. Whyte, and I. Harkavy. 1993. Participatory action research as a process and as a goal. Human Relations 46: 175–92. Gross, D. D. 2002. Commentary: Anthropology as social work: Collaborative models of anthropological research. Anthropology News 43: 4–4.

14_107-Bernard.indb 208

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

209

Hagey, R. S. 1997. The use and abuse of participatory action research. Chronic Diseases in Canada 18: 1–4. Himmelgreen, D. A., R. Pérez-Escamilla, S. Segura-Millán et al. 2000. Food insecurity among low-income Hispanics in Hartford, Connecticut: Implications for public health policy. Human Organization 59: 334–42. Hyland, S. 2005. Community building in the twenty-first century. Santa Fe: School for Advanced Research. Israel, B. A., R. Lichtenstein, P. L. Lantz et al. 2001a. The Detroit Community-Academic Urban Research Center: Development, implementation, and evaluation. Journal of Public Health Management & Practice 7: 1–19. Israel, B. A., A. J. Schulz, E. A. Parker, and A. B. Becker. 1998. Review of community-based research: Assessing partnership approaches to improve public health. Annual Review of Public Health 19: 173–202. Israel, B. A., A. J. Schulz, E. A. Parker, and A. B. Becker. 2001b. Community-based participatory research: Policy recommendations for promoting a partnership approach in health research. Education for Health: Change in Learning & Practice 14: 182–97. LeCompte, M. D., W. Millroy, and J. Preissle, eds. 1992. The handbook of qualitative research in education. San Diego, CA: Academic Press. Lewin, K. 1946. Action research and minority problems. Journal of Social Issues 2: 34–46. Lewin, K., and D. Cartwright. 1952. Field theory in social science: Selected theoretical papers: London: Tavistock. Li, J., S. Shaw, M. Singer, and S. Clair. 2003. The tensions of unity: Challenges of communitycentered research. Practicing Anthropology 25: 8–11. Low, S., and S. Merry. 2010. Engaged anthropology: Diversity and dilemmas: An introduction to Supplement 2. Current Anthropology 51: S203–S226. Maguire, P. 2006. Uneven ground: Feminisms and action research. In Handbook of action research: Participative inquiry and practice, ed. P. Reason and H. Bradbury, 60–70. London: Sage. Malta, M., S. Monteiro, R. M. J. Lima et al. 2008. HIV/AIDS risk among female sex workers who use crack in southern Brazil. Revista de Saúde Pública 42: 830–37. Marcelin, L., B. Page, M. Singer et al. 2008. The impact of international aids in the life of Cité Soleil inhabitants. Unpublished manuscript, Interuniversity Institute for Research and Development (INURED), Port-au-Prince, Haiti. Minkler, M. 2005. Community-based research partnerships: Challenges and opportunities. Journal of Urban Health 82:2: ii2–ii12. Minkler, M., and N. Wallerstein. 2011. Community-based participatory research for health: From process to outcomes. San Francisco: Jossey-Bass. Minkler, M., N. Wallerstein, and N. Wilson. 1997. Improving health through community organization and community building. Health Behavior and Health Education: Theory, Research, and Practice 3: 279–311. Mosavel, M., R. Ahmed, D. Daniels, and C. Simon. 2011. Community researchers conducting health disparities research: Ethical and other insights from fieldwork journaling. Social Science & Medicine 73: 145–52. doi:10.1016/j.socscimed.2011.04.029. Parker, E. A., B. A. Israel, M. Williams et al. 2003. Community action against asthma. JGIM: Journal of General Internal Medicine 18: 558–67. Radda, K. E., and J. Schensul. 2011. Building living alliances: Community engagement and community-based partnerships to address the health of community elders. Annals of Anthropological Practice 35: 154–73. Reid, C., A. Tom, and W. Frisby. 2006. Finding the “action” in feminist participatory action research. Action Research 4: 315–32.

14_107-Bernard.indb 209

6/9/14 10:02 AM

210

Stephen L. Schensul et al.

Romero-Daza, N., D. Himmelgreen, R. Pérez-Escamilla et al. 1999. Food habits of drug-using Puerto Ricans in inner-city Hartford. Medical Anthropology 18: 281–98. Rylko-Bauer, B., M. Singer, and J. Van Willigen. 2006. Reclaiming applied anthropology: Its past, present, and future. American Anthropologist 108: 178–90. Sanstad, K. H., R. Stall, E. Goldstein et al. 1999. Collaborative community research consortium: A model for HIV prevention. Health Education & Behavior 26: 171–84. Schensul, J. 2002. Democratizing science through social science research partnerships. Bulletin of Science, Technology & Society 22: 190–202. Schensul, J. 2005. Strengthening communities through research partnerships for social change: Perspectives from the Institute for Community Research. In Community building in the twenty-first century, ed. S. Hyland, 191–218. School of American Research advanced seminar series. Santa Fe: School of American Research. Schensul, J. 2006. Life at the crossroads. NAPA Bulletin 26: 163–90. Schensul, J. 2010. Malinowski Award presentation: Engaged universities, community based research organizations and third sector science in a global system. Human Organization 69: 307–20. Schensul, J. 2013. Community-based research organizations: Co-constructing public knowledge and bridging knowledge/action communities through PAR. In Public anthropology in a borderless world, ed. S. Beck and C.A. Maida. New York: Berghahn Books. Schensul, J., and M. Berg. 2004. Youth participatory action research: A transformative approach to service learning. Special issue of Michigan Journal of Community Service Learning on Anthropology and Service Learning, ed. J. Howard and A. Keene, Vol. 10: 76–88. Schensul, J., M. Berg, and S. Nair. 2012. Using ethnography in participatory community assessment. In Methods for community-based participatory research for health, 2nd ed, ed. B. Israel, E. Eng, A. Schultz, and E. Parker, 161–88. New York: Jossey Bass. Schensul, J., M. Berg, D. Schensul, and S. Sydlo. 2004. Core elements of participatory action research for educational empowerment and risk prevention with urban youth. Practicing Anthropology 26: 5–8. Schensul, J., M. Berg, and K. M. Williamson. 2008. Challenging hegemonies: Advancing collaboration in community-based participatory action research. Collaborative Anthropologies 1: 102–37. Schensul, J., and M. D. LeCompte. 1999. The ethnographer’s toolkit. Lanham, MD: AltaMira. Schensul, J., and M. D. LeCompte. 2013. Essential ethnographic methods: A mixed methods approach, Vol. 3. Lanham, MD: AltaMira. Schensul, J., K. Radda, E. Coman, and E. Vazquez. 2009. Multi-level intervention to prevent influenza infections in older low income and minority adults. American Journal of Community Psychology 43: 313–29. Schensul, J., J. Robison, C. Reyes et al. 2006. Building interdisciplinary/intersectoral research partnerships for community-based mental health research with older minority adults. American Journal of Community Psychology 38: 79–93. Schensul, J., and S. Schensul. 1992. Collaborative research: Methods of inquiry for social change. In The handbook of qualitative research in education, ed. M. D. LeCompte, W. L. Millroy, and J. Preissle, 161–200. New York: Academic Press. Schensul, J., and G. Stern. 1985. Introduction: Collaborative research and social action. American Behavioral Scientist 29: 133–38. Schensul, J., M. Weeks, and M. Singer. 1999. Building research partnerships. In Research roles and research partnerships, ed. J. Schensul and M. D. LeCompte, 85–164. Lanham, MD: Rowman & Littlefield.

14_107-Bernard.indb 210

6/9/14 10:02 AM

6: Participatory Methods and Community-Based Collaborations

211

Schensul, S. 1973. Action research: The applied anthropologist in a community mental health program. In Anthropology beyond the university, ed. A. Redfield, 106–19. Athens: University of Georgia Press. Schensul, S. 1974. Skills needs in action anthropology: Lessons from El Centro de la Causa. Human Organization 33: 203–9. Schensul, S. 1978. Commando research: Innovative approaches to anthropological research. Practicing Anthropology 1: 13–14. Schensul, S. 1980. Anthropological fieldwork and sociopolitical change. Social Problems 27: 309–19. Schensul, S. 1985. Science, theory, and application in anthropology. The American Behavioral Scientist 29: 164–85. Schensul, S., and M. Bakszysz-Bymel. 1975. The role of applied research in the development of health services in a Chicano community in Chicago. In Topias and utopias in health, ed. S. Ingman, 71–89. The Hague: Mouton. Schensul, S., and M. G. Borrero. 1984. Introduction: The Hispanic Health Council. Urban Anthropology 11: 1–8. Schensul, S., M. G. Borrero, V. Barrera et al. 1982. A model of fertility control in a Puerto Rican community. Urban Anthropology 11: 81–99. Schensul, S., N. Saggurti, R. Singh et al. 2009. Multilevel perspectives on community intervention: An example from an Indo-US HIV prevention project in Mumbai, India. American Journal of Community Psychology 43: 277–91. Schensul, S., and J. Schensul. 1978. Advocacy and applied anthropology. In Social scientists as advocates: Views from the applied disciplines, ed. G. H. Weber and G. J. McCall, 121–65. Newbury Park, CA: Sage. Schensul, S., and J. Schensul. 1979. Advocacy and applied anthropology. In Social scientists as advocates: Views from the applied disciplines, ed. G. Weber and G. McCall, 121–66. Beverly Hills, CA: Sage. Schensul, S., and J. Schensul. 1982. Self-help groups and social advocacy: A contrast in beliefs and strategies. In Self-help groups, ed. G. Weber and L. Cohen, 298–336. New York: Sage Press. Schensul, S., J. Schensul, and M. D. LeCompte, eds. 1999. Essential ethnographic methods: Observations, interviews, and questionnaires, Vol. 2. Lanham, MD: AltaMira. Schensul, S., J. Schensul, and M. D. LeCompte. 2013. Initiating ethnographic research: A mixed methods approach, Vol. 2. Lanham, MD: AltaMira. Schensul, S., R. K. Verma, B. K. Nastasi et al. 2009. Sexual risk reduction among married women and men in urban India: An anthropological intervention. In Anthropology and public health: Bridging differences in culture and society, 2nd ed., ed. M. Inhorn and R. Hahn, 362–93. New York: Oxford University Press. Scrimshaw, N., and G. Gleason, eds. 1992. Rapid assessment procedures: Qualitative methodologies for planning and evaluation of health related programs. Boston: INFDC. Singer, M. 1990. Another perspective on advocacy. Current Anthropology 31: 548–50. Singer, M. 1993. Knowledge for use: Anthropology and community-centered substance abuse research. Social Science and Medicine 37: 15–26. Singer, M. 2000a. Commentary: Why I am not a public anthropologist. Anthropology News 41: 6–7. Singer, M. 2000b. Updates on Projects Recovery and CONNECT. In Careers in anthropology: Profiles of practitioner anthropologists, ed. P. Sabloff, 64–66. NAPA Bulletin #20. Washington, DC: National Association for the Practice of Anthropology (American Anthropological Association).

14_107-Bernard.indb 211

6/9/14 10:02 AM

212

Stephen L. Schensul et al.

Singer, M. 2003. The Hispanic Health Council: An experiment in applied anthropology. Practicing Anthropology 25: 2–7. Singer, M. 2009. Introduction to syndemics: A critical systems approach to public and community health. San Francisco: Jossey-Bass. Singer M., C. Flores, L. Davison et al. 1990. Owning AIDS: Latino organizations and the AIDS epidemic. Hispanic Journal of Behavioral Sciences 12: 196–211. Singer, M., C. Flores, L. Davison et al. 1991. Puerto Rican community mobilizing in response to the AIDS crisis. Human Organization 50: 73–81. Singer, M., and M. R. Weeks. 2005. The Hartford model of AIDS practice/research in community intervention and AIDS. In Community interventions and AIDS: Targeting the community context, ed. E. Trickett and W. Pequenot, 153–75. Oxford: Oxford University Press. Sydlo, S., J. Schensul, D. C. Owens et al. 2000. Participatory action research curriculum for empowering youth. Hartford, CT: Institute for Community Research Tax, S. 1958. The Fox Project. Human Organization 17: 17–19. Tax, S. 1975. Action anthropology. Current Anthropology 16: 514. Theis, J., and H. M. Grady. 1991. Participatory rapid appraisal for community development: A training manual based on experiences in the Middle East and North Africa. London: International Institute for Environment and Development. Trotter, R., and M. Singer. 2007. Rapid assessment: A method in community-based research. In Communities assessing their AIDS epidemics: Results of the rapid assessment of HIV/ AIDS in eleven U.S. cities, ed. B. Bowser, E. Quimby, and M. Singer, 9–28. Lanham, MD: Lexington Books. Van Willigen, J. 1993. Applied anthropology: An introduction. Westport, CT: Bergen and Garvey. Wallerstein, N. B., and B. Duran. 2006. Using community-based participatory research to address health disparities. Health Promotion Practice 7: 312. Weeks, M. R., J. Dickson-Gomez, K. E. Mosack et al. 2006. The risk avoidance partnership: Training active drug users as peer health advocates. Journal of Drug Issues 36: 541–70. Weeks, M. R., J. Li, J. Dickson-Gomez et al. 2009. Outcomes of a peer HIV prevention program with injection drug and crack users: The risk avoidance partnership. Substance Use & Misuse 44: 253–81. Weeks, M. R., S. Liao, F. Li et al. 2010. Challenges, strategies and lessons learned from a participatory community intervention study to promote female condoms among rural sex workers in southern China. AIDS Education and Prevention: Official Publication of the International Society for AIDS Education 22: 252. Weeks, M. R., J. Schensul, S. Williams et al. 1995. AIDS prevention for African American and Latina women: Building culturally and gender appropriate intervention. AIDS Education and Prevention 7: 251–63. Whyte, W. F. E. 1991. Participatory action research. Newbury Park, CA: Sage.

14_107-Bernard.indb 212

6/9/14 10:02 AM

Part II

ACQUIRING INFORMATION

14_107-Bernard.indb 213

6/9/14 10:02 AM

14_107-Bernard.indb 214

6/9/14 10:02 AM

CHAPTER 7

Sampling and Selecting Participants in Field Research Greg Guest

WHAT IS SAMPLING?

Sampling is the process of selecting a subset of items from a larger population for inclusion into a study. Items can be individuals, groups, chunks of time, events, places, or anything that you wish to collect information on or about. Sampling is one of the most important aspects of research design. One reason is that the degree of generalizability (or not) of findings and their representativeness relative to the larger population (or not) largely stem from one’s sampling strategy. Likewise, validity of one’s findings is related to sampling. If the individuals you talk to know little about the research topic, your resulting data will not be very informative. Similarly, if your topic of inquiry is complex and involves multiple stakeholders, and you collect data from only one group of those stakeholders, your findings and conclusions will be significantly limited in scope and relevance. This chapter covers the basics of sampling in field research. The first section deals with the process of selecting whom to include in your study and provides suggestions for framing sampling populations, subpopulations, and constituent strata. The second main section of the chapter addresses how to sample items from your study population(s) and covers the two primary approaches to sampling (as well as censuses)—probabilistic and non-probabilistic. Within each approach are several commonly used sampling methods. I describe the characteristics of each and outline basic instructions on how to execute each method. CHOOSING WHOM TO INCLUDE IN YOUR STUDY

Defining your research population(s) is probably one of the most difficult tasks in research design. Consider these questions: • What is the primary population or populations in your study? • Are there subpopulations within the population that are of particular interest? • Would it benefit your study to include groups of individuals other than the primary population? That is, who are the key stakeholders? • Are there specific individuals whom you should include? • How will you conceptually and methodologically relate the various samples within your study?

The obvious answer to all of these questions is that your research objectives should be the primary guiding factor behind any sampling decisions and your research 215

14_107-Bernard.indb 215

6/9/14 10:02 AM

216

Greg Guest BOX 7.1. DEFINING KEY TERMS

The following sampling terms are mentioned throughout the chapter. Being familiar with them up front will enhance the reading process. Note that these terms have commonly used synonyms. They also may be defined differently by other researchers, especially across disciplines.

Study Population This is the entire group of elements in which your study is interested. This is often a group of individuals (e.g., firefighters in New York City, migrant workers in North Carolina, female sex workers in Bangkok), but it can also be composed of larger units such as families, institutions, communities, cities, and so forth. The study population is entirely defined by the researcher and is based on the study objectives.

Sampling Frame The sampling frame is a list of all the elements in a study population. It is always defined by your study population. If, for example, your study population is vendors in market X, then your frame would be a list of all vendors in that market. Note that the sampling frame is the largest possible sample of a study population.

Subpopulation This is a subset of the study population. Subpopulations are also defined by the researcher. Subpopulations are often determined through stratification of your population by one or more variables of interest.

Sampling Unit This is the thing (person, place, event, etc.) that is selected for inclusion into the study. For example, if you are sampling high school students, one student would be your sampling unit. Sampling units are not always people. They can be objects, places, points in time, or whatever types of things that inform your research question.

Study Sample The study sample is the group of sampling units chosen from the larger population for inclusion into your study.

Elementary Unit This is the person or item that we collect data from or about. It is often the same as the sampling unit, but not always. For example, we may randomly choose households for inclusion into a study (sampling units), but may ask questions from and about one individual within each house selected (elementary unit).

14_107-Bernard.indb 216

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

217

Probability Sampling This is any method of sampling that employs some form of random selection of sampling units. Probability sampling allows the researcher to determine the odds of each sample unit being selected (e.g., one can estimate confidence intervals).

Non-probability Sampling This is any method of sampling that does not utilize random selection procedures. Non-probability sampling does not allow the researcher to determine the odds of each sample member being selected.

Stakeholder In social science research, a stakeholder is an organization or individual with a stake of some kind in the execution and/or outcome of a research project. Stakeholders can be those with the power and authority to implement (or block) research, action, or policy in a study community. Stakeholders also include individuals or groups who will be affected by a research initiative in some way.

design in general. This answer, although absolutely correct, is nonetheless not very helpful from a pedagogical standpoint. I find it more helpful instead to compartmentalize the process and look at each piece of the sampling process in a stepwise order (see also Table 7.5).

Defining Your Primary Study Population One of the first tasks in the sampling design process will be to define your primary study population. In the good old days before the changes from accelerated globalization, the lone ethnographer studied relatively small and geographically bounded indigenous groups. Your study population might have been the !Kung Bushmen of the Kalahari, the Yanomami, or the Trobriand Islanders. With increasing globalization, physical and cultural boundaries of indigenous peoples have become blurred. And the ethnographic approach has become adopted by researchers in other fields working in domestic settings. A landmark domestic ethnographic study, from sociology, for example, was Street Corner Society. It portrayed the lives of Italian Americans living in the slums of Boston (Whyte 1993 [1943]). Another research trend has been the growth of applied field research and the corresponding extension of sociocultural fieldwork into disciplines outside of anthropology, such as public heath, marketing, education, geography, sociology, and social psychology. The result is that for most contemporary field research, study populations are not often as clear cut as they were in traditional ethnographic research. As Johnson (1990, 49) comments: “The heterogeneity of complex societies has taxed some of the traditional methods of ethnography that have been used in the study of relatively small homogeneous communities.” Considering this complexity, how do you begin to define your primary study population? The biggest clue in this puzzle is often found—and should be found—in the title of your proposed research and the primary report/article/book you intend to write as a

14_107-Bernard.indb 217

6/9/14 10:02 AM

218

Greg Guest

result of your research. Are you studying topic X among African American adolescents in Durham, North Carolina? Miskito lobster divers in Honduras? Recent Latina doctoral graduates across the United States or the Jίvaro of Ecuador? At the surface, this seems pretty straightforward. But each of these populations can easily be made more or less inclusive. You may decide, for example, to expand your population geographically to include African American adolescents across the entire state of North Carolina. Or perhaps, after considering your research objectives some more, you feel it necessary to expand the demographic scope of your study and include all minority doctoral graduates within the United States. Conversely, time or budget constraints might force you to reduce your research scope, so you might narrow your study to, say, Miskito lobster divers who have suffered injuries, or the Jίvaro in the Kapawi River region. As Fetterman (1998) notes, determining exclusion criteria is the critical part of the process: “The decision is not who we shall admit but rather who we must reject—given all the people who qualify” (p. 32). The two fundamental questions it really comes down to are: (1) to whom do you wish to generalize, either theoretically or statistically, your study findings; and (2) what scope is feasible within your timeline and the resources you have available? Figuring this out invariably entails an arduous journey composed of an iterative interplay among thinking, gathering more information, and revising your concept.

Determining Subpopulations After you’ve defined your primary study population, you will next need to think about if and how you want to stratify your study population. Stratification—or dividing your study population into subpopulations—is done for two primary reasons. One is to enhance the diversity within and hence the representativeness of your study sample. For many research topics, you would expect variation in knowledge, behavior, and life experiences across a sampling frame so you want to make sure that you capture this variability. Stratification ensures that you have at least some individuals from different subpopulations. Another reason to stratify is if you want to explicitly assess and report on variation within your study population. Basic sociodemographics are a good place to start thinking about the stratification process. Gender, age, marital status, education, ethnicity, and socioeconomic status are commonly used characteristics because they often affect how people experience and perceive the world. For highly targeted research, which is typical in applied research contexts, stratification is often more specific. In one of my studies for example, we were interested in ascertaining the propensity for unprotected sex among women who were involved in an HIV-prevention program. Our primary study population was defined as women at higher risk for HIV in two African study sites, and we used sexual history eligibility criteria to define this population. But my colleague and I also hypothesized that different subsets of these women would report different propensities for unprotected sex than would others. Women who engage in transactional sex—that is, explicitly exchange sex for money or gifts—might, for example, be more motivated to have unprotected sex than other women in the population because clients will pay more for sex without a condom. We had to make sure that we accounted for this when sampling for our study, so we added having had transactional sex within the past year as a stratum.

14_107-Bernard.indb 218

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

219

In short, how you define your strata is a function of how you hypothesize or expect individuals/items in your study population to vary with respect to your research topic. One of the most basic and commonly analyzed forms of variation in social– behavioral research is between groups or subpopulations. If your primary purpose for stratification is direct comparison among groups, you need to be mindful of the implications it has for your sample size. For each dimension (i.e., subpopulation or stratum) you wish to add to your comparative analysis, your overall sample size will necessarily double, assuming you retain the same sample size justifications. So be thoughtful and frugal when making these decisions. Scope creep is one of the researcher’s worst enemies during the research design process. Not all sampling involves people. When conducting observational techniques, for example, you may be more interested in observing activities, events, or places. The two factors that most commonly influence variability in observational methods are time and place. Agricultural activities and events, for example, often vary by season. If you want to observe the entire range of these activities, you’ll need to sample through at least one cycle of seasons. The temporal scale on which other activities vary might be more granular. They might differ, for example, between weekdays and weekends or as a function of the time of day. Events may also vary by location, such as between poor or wealthy neighborhoods or between houses with direct access to water and houses with restricted access. In sum, if you want a representative sample of whatever it is you’re observing, you will need to account for temporal and spatial variability. Some field research, particularly that with a phenomenological purpose, will only be concerned with one primary population. But many research questions are much larger in scope and encompass social structures, shared cultural meaning, or processes and events involving multiple social roles or stakeholders. If your research falls into this latter category—and most anthropological research does—you need to figure out who these key stakeholders are and which ones to include in your study. This is the primary principle behind the concept of triangulation (admittedly an overused word in the social science lexicon). Incorporating multiple perspectives enhances basic validity. The well-known parable of three blind-folded wise men, each touching a different part of an elephant, is instructive here. The first wise man, touching the elephant’s trunk, concluded a snake was in the room. The second, holding a leg, concluded a tree was in the room. The third, grasping the tail, concluded that a rope was in the room. Getting multiple viewpoints is always a good idea; you don’t want to be the blind-folded “wise” man. Inclusion of multiple stakeholders can also be essential from a political and practical standpoint. You wouldn’t want to exclude policymakers from your study if the goal of your research is to inform an intervention that will ultimately require their support. Likewise, you wouldn’t want to exclude the recipients of the potential interventions that will result from your study. One tool to help organize your thoughts and make sampling decisions is a simple matrix. Table 7.1 contains a social role matrix from Johnson’s (1990, 43) ethnographic research in an Alaskan fishing camp. The various social roles (in this case, jobs) within the camp comprise the rows, and the columns contain information re-

14_107-Bernard.indb 219

6/9/14 10:02 AM

14_107-Bernard.indb 220

6/9/14 10:02 AM

Low Low High Moderate Moderate Moderate Moderate High Low High High Low High Moderate

Boat captain

Crew

Carpenter

Head port engineer

Cook

Office manager

Head carpenter

“Beach gang”

Superintendent

Port engineer

Gopher

Tender worker

Night watchman

Bookkeeper

Source: Johnson 1990, 43.

Ability to Interact with Other Roles

Social Role Matrix

Definable Social Role

Table 7.1.

Moderate

Diverse

Limited

Diverse

Localized, detailed, and limited Diverse

Diverse

Localized, detailed and limited Diverse

Diverse

Diverse

Diverse

Localized, detailed, and limited Localized, limited

Access to Information

Moderate

Low

Very low

Moderately high

High

Low

High

Moderate

Low

Moderately high

High

High

Low

Low

Accessibility to Organizational Sectors

Moderate

Low

Very low

Moderately high

High

Very high

Moderate

High

High

Moderate

Moderately high

Moderately high

Moderately high

Moderately high

Power within Organization

Control of resources

Limited

Limited

Skill, technical know-how, and control of resources Control of resources

Decision-making status

Skill and control of files and resources Skill, technical know-how, and control of resources Ability to perform favors

Skill, technical know-how, and control of resources Skill, technical know-how, and control of resources Skill and control of resources

Skill and hard work

Skill and capital investment

Basis of Power

7: Sampling and Selecting Participants in Field Research

221

Table 7.2. Sample Stakeholder Template Stakeholder Group

Level (Who do they communicate with?)

Values/Goals

Concerns

Host community members Study participants Local leaders Religious leaders Ethics committee(s) Community advisory board National policymakers National/international advocates Funders/Sponsors Local/National media Wider scientific community Source: Adapted from Robinson et al. 2011, 19.

garding interaction, power, and access to information. Table 7.2 contains an abridged example of a stakeholder matrix used for communicating about clinical trials. Depending on your research topic, the rows and columns in your matrix will vary. And you may not need to include as many details as in these tables. But, at the very least, you should list all of the possible types of individuals and groups that either have a stake in, or can inform, your research objectives. From there, you can begin to expand, or more than likely, whittle down the range of roles or stakeholders you are able to include in your study. Matrices and other tools can be helpful, but you may not have enough information to fill in the cells, or even know what the relevant rows (social roles or stakeholders) should be. If you’re unsure of how you should stratify your population (if at all) or don’t know who the primary stakeholders are relative to your research objectives, it means you need to gather more information. Where and how you get this information will depend on which type of sampling approach you’re planning to employ—a priori or inductive.

A Priori versus Inductive Sampling The degree of flexibility in your study procedures, your choice of data collection methods, and the amount of time you have to carry out your study will determine whether you employ an a priori or an inductive sampling approach. A priori sampling is the norm for structured, probability-based methods, in which sampling procedures are determined before data collection begins and are, for the most part, immutable once set in motion. This includes most survey research and direct, systematic observation methods.

14_107-Bernard.indb 221

6/9/14 10:02 AM

222

Greg Guest

Qualitative inquiry may also have to be inflexible with respect to sampling procedures. Reviewers of research proposals and protocols (e.g., funders, institutional review boards [IRBs]) typically require researchers to state the sampling parameters for a study before it begins (Cheek 2000). Many IRBs require the submission of a protocol amendment if you change your sampling procedures after the protocol has been approved and that you submit a protocol violation report, post facto, if you’ve implemented a significant change in sampling procedures (e.g., new types of participants or exceeding approved sample sizes) after data collection begins. Such requirements are antithetical to the very principles of inductively based research designs such as ethnography and qualitative research in general (Charmaz 2006, 30). Nevertheless, such requirements are not uncommon, and researchers must do the best they can to establish appropriate sampling strategies prior to collecting data. Your two best resources to guide you in these efforts are the existing literature and people who have expertise with your study population and topic. For the former, look to existing literature to see how other researchers have framed the problem and stratified their population. What stakeholder groups have they included/excluded? Use these sources as a foundation and adapt (and improve on) what has already been established. Other people are also excellent sources of information. Consult colleagues and other researchers in the field and get their advice. Talk to individuals from the study communities before finalizing your research plan; if possible, take a reconnaissance trip to your field site(s). Depending on the IRB that reviews your research protocol, you may be able to build some flexibility into your research strategy and sampling plan. Inductive sampling—or “theoretical sampling” in the grounded theory literature (Charmaz 2006; Glaser and Strauss 1967)—is founded on such flexibility. An inductive sampling approach allows for the inclusion of groups and types of cases not originally specified or conceived of in the original study design. This approach is common in longer term, community-based research such as ethnography. The flexibility to follow new leads during fieldwork and to take advantage of new information as it is collected and reviewed is a major strength of inductive sampling and of ethnographic research in general. Note that even if you have intimate knowledge of the research topic and study population, and a well-thought-out and appropriate sampling plan, it’s always a good idea to leave some wiggle room to follow up on emergent data and hypotheses (resources and protocol permitting).

Locating Appropriate Participants So far I’ve talked about study participants in the abstract, as functional groups or types of stakeholders. But there is more to effective sampling than just selecting the appropriate strata and stakeholders within a population or community. Within any group, specific individuals will prove to be better study participants than are others. As Spradley (1979) puts it, “Although anyone can become an informant, not everyone makes a good informant” (p. 45). Some individuals are simply more knowledgeable about certain subjects than others. In many research contexts, finding knowledgeable individuals is fairly straightforward. If you are interested in boat-building techniques, your experts will be individuals who actually build boats in your study community.

14_107-Bernard.indb 222

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

223

Similarly, traditional healers will likely exhibit a high degree of knowledge surrounding traditional healing practices. For more diffuse research topics, the process is not quite as simple. Handwerker (2001, 91) reminds us that “everyone is an expert in what he or she knows”; so how does one identify and sample, say, “cultural experts”? In traditional, small-scale societies, almost all adults are “thoroughly enculturated” (Spradley (1979, 47), but what about more complex societies or organizations? Proxy measures can be helpful in addressing this question. Generally, the longer and more intensely an individual is a part of something, the more enculturated they are likely to be with respect to that activity or culture. So, if you need to find experts in an organizational culture, you might consider locating individuals who have worked in that organization for a long time and who are actively involved in its organization. Being currently involved is another key dimension because cultures evolve and memories degenerate. Also keep in mind that some groups of individuals might be more willing to talk to you than others. When conducting fieldwork in the Yucatán, I had a difficult time interviewing women because they were often in their homes and reluctant to talk to non-local men in general. Gender-based barriers like this are even more pronounced in more intensively gender-segregated cultures. Other populations present access issues due to the nature of their activities, which may be highly stigmatizing (homosexuality in some countries) or criminal in nature (illegal drug use, black market economics, etc.). Although establishing credibility and trust within any type of community is essential to good fieldwork, it is indispensable (yet much more difficult to achieve) in these “hard-to-reach” or “hidden” populations. Getting introductions from individuals outside of, but who have gained trust among, a hidden population is an effective way to gain initial access. Choosing referral-based sampling strategies, which I discuss in detail later, is another. Validity of information is another important consideration. Johnson (1990) conveys an illustrative example from his fieldwork in an Alaskan fishing camp. Johnson, who possessed carpentry skills when he entered the field, became a member of the fishing community by working as a boat carpenter. Although the fishermen knew Johnson was a researcher, he was able to integrate into the community and establish strong rapport with study participants because of his skills. During his fieldwork, two novice anthropologists happened to arrive in the camp and began asking questions. Johnson recalls overhearing a group of fishermen intentionally (and mischievously) providing the new arrivals with what he knew to be misinformation. Also consider who is too willing to talk to you. A well-known joke among anthropologists is that the first individuals to approach a newly arrived investigator are often those who are marginalized from the community in some way (no one else in the community wants to talk to them). Again, this is why it’s important to include a variety of individuals and groups within your sampling plan. Misinformation is not a new concept in social behavioral research. A large body of literature exists, across multiple disciplines, that addresses the issue of self-report accuracy. The potential for misinformation is always present in any research that involves humans providing information. The best a researcher can do is establish good rapport with study participants and their community, collect data from multiple sources and

14_107-Bernard.indb 223

6/9/14 10:02 AM

224

Greg Guest

with multiple methods, and exercise good judgment. That said, for survey research, certain technologies, such as audio-computer-assisted-self-interviews (ACASI), have been developed with the intent to minimize social desirability bias and thereby enhance the accuracy of self-reported data. The rationale behind ACASI and similar methods is that by removing the interviewer from the interview process, the respondent will feel more comfortable answering sensitive questions truthfully. To date, methodological studies on ACASI indicate that its effectiveness in reducing social desirability bias (as measured by the increase in reporting of sensitive and socially undesirable behaviors) varies by context. So, if you’re thinking of using a technique like ACASI, check first to see if others have used it among your study community and/or topical area of inquiry and what results they have had. Note that a key disadvantage of ACASI-like methods is that they are only useful for structured, noninductive forms of questioning.

BOX 7.2. WHAT DO I CALL PEOPLE WHO ARE PART OF MY STUDY? Traditionally, individuals who provided information in ethnographic studies were called “informants,” and the term is still used today by some social scientists and ethnographers. In the past two decades, however, the term “participant” has increasingly been used to describe individuals who take part in research, particularly qualitative inquiry. This evolution is due to both the negative connotations associated with usage of the word informant in criminal investigations and the trend toward increased democratization of research. Participant connotes more of a two-way process. The use of participant has not yet taken hold in other, more structured, forms of inquiry. “Respondent” is still widely used, for example, to describe individuals who answer structured questions in survey research. In biomedical research, the vast majority of researchers still use the terms “subject” or “patient” to describe individuals within a study, although the use of participant in these contexts is growing.

UNITS OF OBSERVATION AND ANALYSIS

As you begin to achieve clarity around who or what your sampling frame is, and how you will stratify it, you should start thinking about how you will define your units of observation and units of analysis for your study. A unit of observation is the item from, or about which, you collect data (conceptually the same as an “elementary unit” in sampling terms). The most common unit of observation in socio-behavioral research is the individual. When we conduct in-depth interviews or implement a survey, we do so with individuals. For observational methods, the unit of observation is also often at the individual level (i.e., we observe and record individual behaviors), but researchers just as often observe at an event or activity level. You may, for example, be describing the details surrounding a rite of passage like traditional male circumcision in Africa or a bachelor party in the United States. Note also that observational units need not be

14_107-Bernard.indb 224

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

225

people or include behaviors. You may observe vehicles, boats, livestock, or whatever items inform your research objectives. The unit of analysis is distinct from the unit of observation. It refers to how you will organize your data for analytic purposes. Your units of analysis are those elements in your study that you plan to: (1) assess variation among; and/or (2) compare against each other. In most socio-behavioral research, your units of analysis are individuals. Social scientists are often concerned with examining variation within a group of individuals or comparing two or more groups of people on a particular dimension or outcome. Sometimes, though, the unit of analysis is defined at a higher level of abstraction and can include items such as households, events, time periods, geographic regions, and the like. In many cases, the unit of observation and unit of analysis are the same, but not always. You may, for example, collect data about individuals in a household, aggregate those data at the household level, and analyze variation among households. Units of observation can, therefore, be smaller than units of analysis in a study, but never vice versa. Bernard provides researchers with a simple but important rule when it comes to selecting units of observation and analysis: Always collect data at the most granular level possible. The reason is simple: “You can always aggregate data collected on individuals, but you can never disaggregate data collected on groups” (Bernard 2013, 44; emphasis in the original). GENERAL APPROACHES TO SAMPLING

Once you’ve figured out who and what you want to include in your research study and have determined your units of observation and analysis, the next big decision to make is which sampling strategy(ies) you will employ in selecting those units from the larger population. And this needs to be considered for each population you wish to include in your study. Two major categories of sampling approaches exist—probabilistic and non-probabilistic. A third category—which is not technically sampling—are censuses. I describe these approaches below and provide basic instructions on how to execute the common variations of each type.

Censuses A census collects information from or about every item in a study population (technically, it’s not a sampling method because no selective process is involved). Censuses are often associated with large quantitative data collection efforts—such as the U.S. Census—but they can also be employed in research of a smaller scale. For any research, a census is always preferable to sampling because it eliminates the need to infer from a sample to a larger population. Censuses are not very common in field research due to logistical constraints such as time and money. Most study populations are large, and including everyone from a population in a research project is usually prohibitively time consuming and costly. Nevertheless, there are certainly situations in which the population of interest is small enough to permit a census (e.g., all of the traditional healers in a community or all of the elders in a small village). As a general rule of thumb, if you can feasibly obtain a census, do so. A census eliminates the potential for criticisms pertaining to generalizability of your findings (at least relative to your study population). If

14_107-Bernard.indb 225

6/9/14 10:02 AM

226

Greg Guest

you can’t employ a census, however—which will often be the case—you will need to choose some form of sampling. Two general approaches exist from which to choose—probabilistic and non-probabilistic.

Probabilistic Sampling A probabilistic, or probability, sample is derived using some form of random selection. Random selection means that the probability of each sampling unit being chosen is known. For most types of random sampling, the chance of selection is equal, but in some cases it is not, and samples must be “weighted.” This weighting is taken into account when analyzing and interpreting the results. The defining feature for any probability sample is that one can calculate the odds or probability (i.e., provide confidence intervals) that the sample represents the larger population well. The basis for creating a random sample is the generation of a set of random numbers, and the assignment of those numbers to individual sampling units. In contemporary research, random number generation is invariably carried out using computer software (Excel or any statistical program) or via the Internet (e.g., www.randomizer .org), although random number tables are still available and used occasionally. When should a researcher use probability sampling? Traditionally, variations of the following two related reasons are given for employing probability sampling: 1. When statistical methods one plans to use require a probabilistic sample (which is the case for most statistics). 2. When there is a need to statistically generalize results from a sample to the larger study population. Statistical generalization includes probability estimates such as sampling error, standard error, and confidence intervals.

Given these reasons, it makes sense that most population-based surveys employ some form of probability sampling. Indeed, this goes for many studies that can acquire or generate a decent sampling frame and that involve quantitatively oriented data collection and analyses such as structured surveys, structured observation techniques, and confirmatory content analysis of text. If these criteria apply to one of more of the data collection methods and analyses you have planned for your study, a probability sample is going to be your most likely option. Before considering the various types of probability sampling methods, though, the first question you as a researcher need to answer is: Can I obtain or realistically generate a decent (not perfect) sampling frame? How you answer this question will determine, in large part, the type of probabilistic sampling method you choose. ESTABLISHING A SAMPLING FRAME (OR NOT)

At the beginning of this chapter, I defined a sampling frame as a list of all of the items in a study population. The basic question is this: Can you acquire or create a reasonably exhaustive list of all the items in your study population? For some populations, this is a relatively simple process. If your population comprises all of the students in a school or employees in a company, lists will most certainly exist somewhere within these institutions that contain everyone’s name (getting access to these lists is another

14_107-Bernard.indb 226

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

227

matter). If you’re interested in households you can probably access census data, at least in most developed countries. Think creatively about where such lists might exist for your study population(s). For an HIV prevention study in Uganda, Guest et al. (2010) needed a sampling frame of all of the long-distance truck drivers in the country. With the help of local researchers, the research team tracked down several truck driver association membership lists and used these as a basis for an initial sampling frame. Female vendors in Kampala’s Owino market (of which there are more than 8,000) comprised another target population in the same study. On his first visit to the field site, Guest visited the market administrator with his local collaborator to get permission to conduct the study. During this brief conversation, the administrator mentioned that he had a list of all the market vendors that the field team could copy. Within minutes, a sampling frame was (unexpectedly) obtained. In many cases, particularly in less-developed areas of the world, censuses and membership lists are hard to come by or are of dubious quality. If you’re intent on having a decent sampling frame, in many research contexts you’ll have to create your own by enumerating all of the items in your population. For most large-scale research endeavors, this is a costly and time-consuming process. For more moderately sized initiatives, creative workarounds can be developed. Using satellite images is a prime example. With Google Earth and similar remote sensing applications readily available, the task of establishing sampling frames at the household level is no longer as difficult as it once was. Simply identify and number the houses on the remote image and then spend a little time ground-truthing the image, confirming which buildings are residences and occupied, and so on. You now have a solid sampling frame. And if you have a good sampling frame, you can employ one of the most common and robust types of probability samples—a simple random sample. SIMPLE RANDOM SAMPLE (SRS)

Probably the most common probability sample, an SRS is also one of the most versatile and easiest to employ, both logistically and analytically. In an SRS, each unit has the same selection probability and the sampling unit is identical to the elementary unit. The procedure is simple and based on three fundamental steps. 1. Assign a number from 1 to N to each element in the population, with N being the total number of known elements in the population. 2. Obtain a list of n different numbers (n being the number of elements you want in your sample) within the 1 to N range, each one obtained using a random process. 3. Choose the elements in the population corresponding to these numbers.

There are several advantages of an SRS. It is conceptually the simplest probabilistic sampling method and thus easy to analyze. Many statistical analysis packages and standard statistical methods assume this type of sampling. An SRS is also self-weighting, since every element has an equal and independent chance of selection. One of the challenges associated with an SRS, however, is that it requires a sampling frame. In many research contexts this is simply not possible. Another potential drawback to an SRS is that the resulting sample may be spread out over a large geographic

14_107-Bernard.indb 227

6/9/14 10:02 AM

228

Greg Guest

area if your population is geographically dispersed. Imagine if your sampling units were rural houses within an area of 2,000 square miles. Data collection would require a substantial amount of travel. Another limitation is that an SRS might not adequately capture subpopulations if they are hidden or small in number, relative to the larger population. Suppose, for example, for analytic purposes, you wanted roughly an equal number of men and women in your study but only 10% of the population was female. If that 10% rendered an n of, say, 15 females an SRS would not provide enough women in your sample to include in a meaningful statistical comparison. To compensate for these limitations, other probability sampling approaches, described below, can be employed. STRATIFIED SAMPLE (RANDOM)

The defining characteristic of stratified sampling is the conceptual division of a population into two or more subpopulations, or strata (see Figure 7.1 for a visual example). Stratified sampling is primarily used when an analysis calls for inference to subpopulations that may not be represented in adequate numbers within the larger population. It generates a more balanced sample, relative to one’s research objectives. For example, Guest and a colleague needed to devise a sampling strategy for injecting drug users (IDUs) in East Africa. Existing data indicated that 95% of IDUs are men, but our research objectives included a gender-based comparison. We needed to make sure, therefore, that we had enough female IDUs in our sample to enable a statistical comparison between male and female IDUs so we opted for a stratified procedure. For any stratified sample, however, you will need to weight your results if making inferences to the larger population from which the strata were drawn (in our case, IDUs in the field sites). A random stratified sample requires a sampling frame as well as a good

Figure 7.1. Stratified sample of 5 men and 5 women (n = 10, from a population of 50).

14_107-Bernard.indb 228

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

229

estimate of the relative proportions of the chosen strata within the study population. The procedure for creating a stratified sample is relatively straightforward: 1. Classify each element into one stratum. Strata must be mutually exclusive and sampling units must belong to only one stratum. 2. Select a simple random sample from each stratum. 3. If relative proportions of subgroups in your sample are different than those in the actual population, you will need to weight your findings if making inferences to the larger population. SYSTEMATIC SAMPLE

If you can’t procure or generate a sampling frame, an excellent alternative is a systematic sample. A systematic sample involves the selection of every nth unit within a study population. For example, we might select every 20th name in a phone book, visit every 8th house in a village, or interview every 5th person entering a market or attending a health clinic. Although theoretically systematic sampling is not true random sampling (once the first unit is selected, the odds of selection for every other unit have been predetermined), it is as robust as an SRS for the vast majority of research contexts. If you can get a decent systematic sample in field research your study is in good shape, at least from a sampling perspective. The process involves four general steps: 1. Determine the desired sample size. 2. Estimate the size of your study population. 3. Determine your sampling interval. This is calculated by dividing the estimated population size by the desired sample size. 4. Randomly select a starting point, and then select every nth item from that point.

Because a systematic sample does not require a sampling frame, it is a versatile option for many field research contexts. There are some limitations to the method, however. One is that you need to be able to estimate the population size fairly well. If you overestimate, you will reach your sample size prematurely, before traversing the entire population. In contrast, underestimation will result in exhausting the sample population before obtaining your desired sample size. For a systematic sample it’s helpful if sampling units are configured in an orderly or defined manner. If you want to sample, say, people attending a sporting event, it’s best if: (1) there are defined entrances from which to count and sample; or (2) seating is fixed and orderly. If the sampling context is more fluid or less orderly, then systematic sampling becomes more difficult. Imagine, for example, obtaining a systematic sample at a large political rally in an open space, where people are milling about. Workarounds can be created for such contexts (e.g., creating invisible transects across the crowd and selecting every nth person along those lines), but, in general, diffuse and mobile environments are not ideal for systematic sampling. Another potential problem, though rare, is order bias, If the assemblage of units to be counted has some sort of regular patterning (e.g., people are seated in an alternating male/female configuration), this can introduce obvious bias into the sample.

14_107-Bernard.indb 229

6/9/14 10:02 AM

230

Greg Guest

TIME–SPACE SAMPLING

One of the sampling techniques not often covered in research design books is time–space, or time–location, sampling (TSS). It is sometimes used when probability estimates are desired but when sampling frames at the individual level can’t be created. TSS is particularly useful for reaching populations that congregate in known and geographically bounded venues and has been successfully used in research studies among gay men in the United States (MacKellar et al. 2007), sex workers in Africa (Kayembe et al. 2008), and Brazilian truck drivers (Ferreira et al. 2008). The TSS process is composed of three basic phases: 1. Formative Research. This phase is ethnographic in nature and typically involves one or more of: secondary data review and analysis, in-depth interviews with key informants, and participant observation. The primary purpose of the formative research is to establish an initial sampling frame of appropriate locations/venues and times, called “venue-day-time-units” (VDTs). 2. Enumeration. Through more ethnographic investigation, venues are characterized in more detail, particularly the range and proportion of the study population at each venue. Enumeration at this level will permit the researcher to estimate the “effective yield” at each venue. 3. Venue Selection and Sampling. Once a final sampling frame of VDTs is constructed, an SRS of VDTs is obtained—preferably weighted based on the effective yield of VDTs—and data collection is carried out in the specified places and times.

In one of Guest’s studies in two African cities (Mbeya, Tanzania and Tema, Ghana), for example, VDTs were constructed of bars/night clubs and their busiest time period (8 p.m. to midnight, Thursday–Saturday) to sample men at higher risk for HIV (Guest et al. 2006). A portion of the sampling frame containing VDTs for this study is presented in Figure 7.2 and illustrates the (randomly) assigned venues to specific dates. Note that an optional probabilistic element can be added to a TSS procedure. The sampling unit in a TSS is the VDT. Within each VDT, however, individuals can be

Figure 7.2. Time-space sampling of bars/clubs in Tema, Ghana.

14_107-Bernard.indb 230

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

231

further sampled using virtually any other type of sampling method. In the study described above, for example, once a field team was at a VDT, systematic sampling was carried out to choose individual men within the venue. RESPONDENT-DRIVEN SAMPLING (RDS)

Since the 1990s, RDS has become the go-to sampling technique for hard-to-reach populations when probabilistic samples are desired. A form of chain-referral sampling, RDS is based on the principle that individuals utilize their social networks to identify and recruit their peers into a study (Heckathorn 1997). RDS is distinguished from non-probabilistic, snowball sampling in two important ways. First, in RDS a limited number of recruiting “coupons” (usually three) with identifiers are given to study participants. This allows researchers to map out the social connections among study participants. And, by limiting the number of coupons per individual, selection bias is reduced and referral chains are lengthened. The procedure involves the following steps: 1. Locate “seeds” to begin the recruiting process. Try to obtain as diverse a group of seeds as possible so as to saturate the population faster. 2. Provide seeds with (usually) three coupons to distribute to others in their network (which is presumably your population of interest). Coupons have unique ID numbers on them that connect them to both the seed and the individuals he or she recruits. 3. Seeds recruit peers in the first “wave,” and those peers return the coupons to the research team when they arrive for whatever data collection activity is planned. 4. These peers then recruit three more individuals in a second wave, and so on until “equilibrium,” or redundancy in recruitment, is achieved. In most cases, four–six sampling waves are enough to reach equilibrium (see Figure 7.3).

RDS is an exceptional sampling method for hidden populations who exhibit a certain degree of social connection (Magnani et al. 2005). RDS is believed by some scholars to be able to provide probability-based population estimates and standard errors (Semaan 2010), but there is considerable debate within the sampling literature as to whether RDS allows for unbiased estimation reasonably well (Gile and Handcock 2010). Despite the theoretical debate surrounding this sampling method, if a probability sample is needed for a hidden population and the only choice is between an RDS, a non-probability snowball sample, or no sample at all, RDS is the clear choice. CLUSTER SAMPLING

In cluster sampling, it is a cluster of elementary units—as opposed to the units themselves—that are randomly selected. Then, every item in the chosen clusters is included in the study. Clusters are clearly demarcated, mutually exclusive groups of items (see Figure 7.4). Clusters can be anything from districts within a large city, or rural villages within a region, to sections of a large outdoor market. In these examples, elementary units might be households, market stalls, or individuals. Generating a (one-stage) cluster sample involves five basic steps: 1. Define the clusters in your population. 2. Assign a number from 1 to N to each cluster in the population (N = total number of clusters).

14_107-Bernard.indb 231

6/9/14 10:02 AM

232

Greg Guest

Figure 7.3. Visual depiction of RDS.

3. Select a sample of clusters using a random process. 4. Take the clusters in the population corresponding to these numbers. 5. ALL of the elements in these clusters constitute your sample (for a one-stage sample).

Although cluster samples have the advantage of not requiring a traditional sampling frame and can simplify field logistics (i.e., they require less travel), there are several significant drawbacks to the method. One is that the researcher must be able to define boundaries between clusters (i.e., there must be identifiable clusters), which is not always possible. Precision is also decreased, due the possible correlation among elements within clusters. The most egregious shortcoming of cluster sampling, however, is the geographic bias it can introduce into your study. Imagine, for example, that you want to select 10 out of the 50 districts in a large city for inclusion into a study (note also that there are no real guidelines as to how many clusters should be selected). What if the random selection process identifies 10 of the city’s poorest districts? Your sample would be completely missing middle-class and affluent areas. The same could happen for any important variable (e.g., religion, ethnicity, access to water, etc.). In my view, this bias is so problematic that I recommend using cluster sampling as a method of last resort and is why it’s not included in the sampling decision tree at the end of this section.

14_107-Bernard.indb 232

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

233

Figure 7.4. Cluster sample, with two clusters (n = 10).

Non-Probabilistic Sampling The difference between non-probability and probability sampling is simple: Nonprobability sampling does not involve random selection. This does not necessarily mean that non-probability samples aren’t representative of the larger study population. But it does mean that non-probabilistic samples cannot depend on probability theory, which is to say that non-probability sampling does not allow the researcher to determine the odds of each sample unit being selected. Non-probabilistic sampling is the norm when using qualitative data collection methods. The primary reason for this is qualitative inquiry is usually not intended or designed for statistical generalizability. Rather than being concerned with populationbased estimates and confidence intervals, qualitative research seeks to generate rich, explanatory data. Probability samples provide little benefit to this end, and the effort needed to acquire a proper sampling frame, as is necessary for most probability samples, is not justified for most forms of qualitative inquiry. Related to this are the topics on which qualitative research often focuses. In many cases, such studies seek to understand common processes, shared experiences, and understandings, or identify basic cultural knowledge and norms. It doesn’t take a probabilistic sample to gain insight into these types of topics. In fact, as discussed in more detail below, sample sizes of as small as six individuals are often adequate in this regard. How many villagers would you need to ask, for example, to find out what time the local market opens? How many fishermen in a small community would you need to interview to get a general picture of what types of fish are typically sought after? Much experience and knowledge is so widely shared or processes so standardized that only a few individuals are needed to address a research question. Even in cases where one is interested in variability—as opposed to commonalities and patterns—qualitative inquiry is typically used to establish only the range of the phenomena of interest.

14_107-Bernard.indb 233

6/9/14 10:02 AM

234

Greg Guest

PURPOSIVE SAMPLING

The most commonly employed non-probabilistic sampling approach is purposive, or purposeful (also sometimes called “judgment” sampling). Perhaps the most intuitive way to think of purposive sampling is that you choose study participants based on the purpose of their involvement in the study. In Bernard’s words (2013, 164): “You decide the purpose you want your informants (or communities) to serve, and you go out and find some.” The logic and power of purposive sampling, argues Patton, “lie in selecting information-rich cases for study in depth . . . those [cases] from which one can learn a great deal about issues of central importance to the purpose of the inquiry” (Patton 2002, 230). Operationally, a purposive sample requires establishing one or more eligibility criteria for inclusion into your study. Your criteria could be as simple and broad as “adult males in village x.” Or perhaps your research is aimed at understanding a very specific experience, such as traveling through security checkpoints in a conflict zone. One of your criteria might be having traveled through at least 10 checkpoints in the past month. Often, multiple sampling criteria are used. You may wish to limit the above sample to adult males who have traveled through at least 10 UN-staffed checkpoints in districts X, Y, and Z. The number of criteria you choose is a function of the degree of specificity or breadth you wish your study sample to exhibit. You also need to consider the analytic purpose of your sample. Your sampling criteria would be vastly different, for example, if looking for common themes across a homogeneous group of individuals versus identifying the maximum range of themes across a heterogeneous group. Table 7.3 provides a framework from which to start thinking about the various dimensions of purposive sampling, and includes 14 variations. Note that this list is not exhaustive, nor are the sampling variations necessarily mutually exclusive. INTEGRATING SAMPLES

In many types of research, including ethnography, sampling approaches of various types are routinely combined. There are many possible permutations. For example, in Guest et al. (2010), three different sampling approaches were used for one data collection method—time/space, systematic, and cluster. Multiphase sampling is common practice in larger, more complex field studies. A simple yet illustrative model of sampling integration comes from the field of mixed methods. Onwuegbuzie and Collins (2007) reduce the process of integrating samples to two underlying dimensions— timing of procedures and the type of relationship between sampling units within two or more samples (see Figure 7.5). The key factor to consider when selecting if and how to integrate sampling methods is the degree to which you want to connect data from your samples, theoretically and analytically. There are no explicit rules when it comes to integrating sampling methods within a study. Be creative and do whatever makes sense for your research objectives and study context.

Timing • Sequential: Samples are selected sequentially, at different times. Previous sample informs subsequent sample. Subsequent sample is dependent in some way on prior sample. • Concurrent: Samples are drawn at the same (or close in) time. Samples are independent of one another.

14_107-Bernard.indb 234

6/9/14 10:02 AM

Table 7.3. Purposive Sampling Variations Sampling Type

Characteristics

Example

Homogeneous

Sample is homogeneous on one or more dimensions. Useful if the population to which the results will be inferred is also homogeneous. One of the most common sampling approaches. Data are often collected from multiple homogeneous groups. Learning from unusual manifestations of the phenomenon of interest. Often used to find key factors associated with extreme behavior. Such cases may or may not be identified with quantitative data (e.g., the nth percentile). Information-rich cases that manifest the phenomenon of interest intensely, but not extremely. Differ from extreme/deviant cases in that they are in a normal range and not on the extreme end of the curve. Illustrate or highlight what is typical, normal, average. If selected well, a few individuals can provide accurate insight into general patterns and processes across a larger population. Typical cases are illustrative, not definitive. Particularly good for understanding phenomena that are widely shared across a culture or community. Permits logical generalization and maximum application of information to other cases because if it’s true of this one case, it’s likely to be true of all other cases.

You wish to understand how, in general, mothers in study community treat their children’s diarrhea. You may also choose to sample a group of traditional healers as well.

Extreme or Deviant Case

Intensity

Typical Case

Critical Case

Confirming and Disconfirming Cases

Elaborating and deepening initial analysis; seeking exceptions; testing variation. Often used to validate (confirming) or expand/ negate (disconfirming) data-driven models. Often sought out near the end of a research project to establish boundaries around one’s interpretation of a dataset.

Politically Important Cases

Purposefully eliminating or including politically sensitive cases. Sample individuals to attract attention to the study or to avoid attracting undesired attention. Picking all cases that meet some predetermined criterion, Often used to assess adherence to, or deviance from, set standards or guidelines. Purposefully picking a wide range of cases for a sample to get variation on

Criterion

Maximum Variation

Identifying outstanding successes or notable failures, such as the only two individuals in the community to have a college degree. Or, the only men in a hunting society who choose not to hunt. Sampling very successful and not-sosuccessful fisherman. Or individuals who are above average/below average on a given dimension.

Interviewing adolescents who recently underwent a traditional circumcision. The assumption is that the process and experience is fairly typical across individuals so relatively few participants need to be sampled.

If the most highly educated members of the community are having difficulty understanding a newly introduced law or political process, one can be fairly certain the rest of the community will also have problems. Your data reveal a widely shared theme. Taking a second look through the same dataset in search of data that diverge or contradict that theme. Sample individuals or events that will inform those observations further. Imagine all of the 30 people you interview have positive views of the newly constructed dam nearby. You purposively seek out individuals in the community who have negative views of the dam, and try to understand why they have different perspectives. Local leaders or political activists, for example, are often included in a study to increase political cooperation.

Sampling children who have been abused in a local orphanage, or all middle school children who failed the past school year. You’re interested in documenting the diversity of opinions regarding a new (continued)

14_107-Bernard.indb 235

6/9/14 10:02 AM

Table 7.3. (Continued) Sampling Type

Stratified Purposeful

Characteristics dimensions of interests. Document unique or diverse variations that have emerged in adapting to different conditions. Identify important common patterns that cut across variation (cut through the noise of variation). Also useful for generating a wide range of responses. Used when time/resource constraints prohibit sampling multiple groups. Study population is broken down into strata that have some theoretical importance to the study objectives. Illustrative characteristics of particular subgroups of interest; facilitate comparisons.

Quota

A variation of a stratified sample. Relative size of sub-strata are proportional to their relative size in the larger population

Theory-based

Finding real-world manifestations of a theoretical construct of interest so as to elaborate and examine the construct and its variations. Cases are chosen based on their ability to inform parts of the theoretical model.

Snowball or Chain Referral

Utilizes participants’ social networks to identify other participants. Ask existing participants to refer others based on certain criteria (often attributes similar to the referring participant). Very useful for sampling hard-toreach populations, but vulnerable to recruiting biases. Note that this is really a recruiting technique, since chain referrals are only used to achieve types of samples. However, since chain referral techniques are typically referred to as sampling methods in the literature (correct or not), they are included in this table. Technically not a form of purposive sampling. Collect data from whatever cases present themselves. Convenience sampling has the lowest credibility of all sampling approaches, so this technique should be either be viewed as a last resort or as a way of gathering exploratory data to allow for more targeted sampling strategies.

Convenience or Haphazard

Example agricultural reform law that is soon scheduled to take effect. Select your individuals for sample to enhance the range of viewpoints. Include individuals with specific roles and backgrounds that you would expect to lead to diverse perspectives.

Examples of commonly used strata include: gender, ethnicity, age, education, income level. Can be much more specific such as the amount of experience using a particular technology or degree of exposure to a program or activity. A general quota sample of men and women with an n of 20 would break down to 10 men and 10 women, since the sex ratio in general populations is 50/50. If, however, your study population was injection drug users, of which 90% are men, your quota sample of 20 would be composed of 18 men and 2 women. Models can come from various sources such as existing literature or primary data. For example, your survey data indicate that there is a statistical association between formal education and gender norms that young men express. You sample individuals—perhaps school teachers and parents—to better understand why this association exists. Your study objective is to understand what factors influenced individuals’ decision to engage in piracy or prostitution. Once you find one individual willing to talk, you ask for referrals to others. You may also ask individuals engaged in these activities to suggest other types of individuals who might have some insight into the research problem (i.e., to increase the number of viewpoints to bear on the subject).

Conducting research in conflict zones, for example, may only permit convenience samples due to security issues. In ethnographic research, initial interviews are often convenience based, until the researcher becomes more familiar with the study community.

Source: Adapted from Patton 2002, 243–44.

14_107-Bernard.indb 236

6/9/14 10:02 AM

7: Sampling and Selecting Participants in Field Research

237

Figure 7.5. Sample integration model (adapted from Onwuegbuzie and Collins 2007).

Relationship between Samples • Identical: Same sampling units and sample sizes. Useful if you want to directly compare responses to different data collection activities at the observation unit level. • Parallel: Samples are drawn from the same population, but are composed of different units. Use if there is no need to directly compare data at unit level, or if “identical” sampling is too burdensome for participants. • Nested: Smaller sample is a subset of a larger sample. Use if need to either directly compare between samples, or infer from a smaller second sample to the larger sample. • Multilevel: Samples from two or more different populations or two or more levels of analysis (definition depends on author). Particularly useful if triangulating or comparing independent groups or levels of analysis. HOW DO I CHOOSE A SAMPLE?

With so many options to choose from, how does one go about selecting an appropriate sampling method or methods for a research initiative? The strengths, weaknesses, and requirements for each of the sampling methods have been discussed above and should provide readers with some degree of guidance. Figure 7.6 simplifies and synthesizes this information into a single graphic. The decision tree provides a very rough set of guidelines to help you choose a sampling approach. Walking through the tree from the top, ask yourself the following questions: 1. How large is the population of interest? If it’s extremely small (e.g., <20), for example, you can probably obtain a census. 2. Do you need to statistically generalize, or do your methods require a probability sample? If yes to either of these questions, start heading in the direction of probability sampling. 3. If opting for a probability sample, is it possible to obtain/generate a sampling frame? If not, you can rule out an SRS or a stratified sample and start thinking about if and how a systematic or time-space sample can be achieved.

14_107-Bernard.indb 237

6/9/14 10:03 AM

238

Greg Guest

4. If choosing the non-probability side of the tree, consider how much control you will have over your recruitment and sampling procedures. If you can control to some degree what units you can select for inclusion, you should be able to avoid convenience sampling. 5. Another factor to consider is whether you need to include subgroups and whether these subgroups are underrepresented in the population. If so, your sample will likely need to be stratified in some way. 6. Also ask yourself how the population is configured in time and space. If considering a systematic sampling approach, you need to establish whether or not the population is defined and orderly enough in time and space to systematically count each unit. If your study population can be found in large numbers in specific geographic areas, then time/space sampling is a good option. 7. Is your population hidden and socially connected somehow? If so, consider employing a form of chain referral sampling. 8. If a non-probability sampling approach is chosen, how inductive and flexible are your study procedures? If relatively fixed, any type of theoretical/inductive sampling is not logistically feasible.

Figure 7.6. Sampling method decision tree.

14_107-Bernard.indb 238

6/9/14 10:03 AM

7: Sampling and Selecting Participants in Field Research

239

The tree in Figure 7.6 is just a tool to help researchers think about choosing a sampling approach. The pathways are by no means perfect. Sampling is a complex process that is contingent on a multitude of factors and cannot be entirely captured in a simple heuristic device. Use the figure cautiously as a starting point to begin thinking about a sampling plan. CHOOSING SAMPLE SIZES

Probability-Based Probabilistic sample sizes are usually estimated mathematically based on preselected parameters and objectives (i.e., x statistical power with y confidence intervals). If your study will be based primarily on quantitative data, you will want to determine a priori your odds of observing a predefined degree of difference (i.e., statistical power). Statistical power is the probability you will detect a meaningful difference, or effect, if one were to occur (studies often use power levels of 0.80 or higher [i.e., an 80% chance or greater of finding an effect if one was really there]). Each sampling case is unique with respect to the analytic context, so estimating the “correct” sample size (i.e., conducting a power calculation) may require different pieces of information. Generally, the more parameters you can plug into your calculation, the more precise your estimate will be. Several of the following parameters are common inputs for power calculations: • • • • • • • •

the size of your population; the desired error level (e.g., 5%); the desired level of confidence (e.g., 95%); the number of variables you will include in your analysis; the type of statistical analysis you plan to employ; the desired power coefficient; whether a one- or two-tailed test will be used; and the expected proportion of the primary outcome variable (if your study has a primary outcome).

Numerous sample size calculators—both online and in software packages—are available to help estimate statistical power. Some are relatively simple in that they don’t require much information but produce crude estimates (e.g., www.raosoft.com/ samplesize.html), while others are more sophisticated, requiring more information up front but yielding more precise estimates (e.g., G*Power). For readers interested in the theory and math behind probability sampling sizes, a good reference is Jacob Cohen’s classic Statistical Power Analysis for the Behavioral Sciences (Cohen 1988). In some cases, you might not have control over how many items you can include in your study, resulting in an “expedience”-based sample that is smaller than you would like. Although not optimal, this does not necessarily mean that your study is destined for failure. Do the best you can to get an appropriate sample size and, when reporting on the study, provide an explanation for any sampling shortfall. Most experienced researchers and reviewers understand the logistical constraints of fieldwork. Note also, as illustrated in Box 7.3, that large sample sizes are not always necessary; once a sample size moves past a certain number range, the return on precision becomes disproportionally smaller.

14_107-Bernard.indb 239

6/9/14 10:03 AM

240

Greg Guest BOX 7.3. A STATISTICAL REASON WHY “MORE” ISN’T NECESSARILY “BETTER”

Researchers often overestimate the benefits of increasing sample sizes above a minimum threshold, believing that “the more the better.” Mathematically, it can be shown that gain in precision associated with increasing the sample size beyond 2–3% margin of error (MOE) diminishes significantly as the sample size grows. The following graph shows how the maximum MOE associated with the estimate of a proportion (based on a population of 5 million and a 95% confidence level) changes based on the sample size. Note that as the sample size exceeds 1,000, the error line flattens, remaining virtually unchanged as the sample size increases.

Graph 7.1. Margin of error associated with the estimate of a proportion changes based on the sample size.

And, as shown in Table 7.4, it takes an increase of only 118 respondents to move from a 6% MOE to a 5% MOE, but it takes a sample increase of 466 to move from +/−4% to +/−3% MOE, and an additional 1,333 respondents to reduce the MOE from +/−3% to+/−2%, and so on. Table 7.4. Increase in Sample Size Required to Reach a Given Margin of Error Sample Size (at 95% C.L.)

Margin of Error (+/–)

Step increase in Sample Size (N)

Percent Improvement in Margin of Error

267 385 601 1,067 2,400 9,586

+/− 6.0% +/− 5.0% +/− 4.0% +/− 3.0% +/− 2.0% +/− 1.0%

+118 +216 +466 +1333 +7186

+/− 1% +/− 1% +/− 1% +/− 1% +/− 1%

Source: Adapted from Hintze, n.d.

14_107-Bernard.indb 240

6/9/14 10:03 AM

7: Sampling and Selecting Participants in Field Research

241

Note that the sample size graphic and calculations in Box 7.3 do not take into account subgroup analyses or more sophisticated statistical procedures such as multivariate analyses. As a general rule, though, the more subgroups (i.e., strata) you wish to compare and the more variables you wish to include in your analysis, the larger the sample you will need.

Non-Probability Based: Qualitative Data Non-probability sample size recommendations for qualitative data are not common in the literature, and the few recommendations that do exist exhibit a certain degree of variability (Guest et al. 2006). Bernard (2000, 178), for example, observes that most ethnographic studies are based on 30–60 interviews. Bertaux (1981) argues that an n of 15 is the minimum acceptable sample size in qualitative research. Morse (1994, 225) recommends at least 6 participants for phenomenological studies; 30–50 participants for ethnographies, grounded theory studies, and ethnoscience studies; and 100–200 sampling units in qualitative ethology. Creswell (2006), on the other hand, recommends between five and 25 interviews for a phenomenological study and 20–30 for a grounded theory study. Relating his recommendations to sample heterogeneity and one’s research objectives, Kuzel (1992, 41) suggests six to eight interviews for a homogeneous sample and 12 to 20 data sources when looking for disconfirming evidence or trying to achieve maximum variation. Although these rules of thumb may be, at best, based on anecdotal experience, none of them are presented with supporting evidence. The majority of the qualitative research literature does not make any specific recommendations for non-probabilistic sample sizes but posits that sample sizes in qualitative research should be determined inductively and continue until “theoretical saturation” is reached (e.g., Bluff 1997; Byrne 2001; Fossey et al. 2002; Morse 1995; Sandelowski 1995). Theoretical saturation is “the point at which gathering new data about a theoretical category reveals no new properties nor yields any further theoretical insights about the emerging grounded theory” (Charmaz 2006, 189). The pervasiveness of the concept in the discussion of sample sizes is further evidenced by the trend for some journals, in a variety of research fields, to require that theoretical saturation be the main criterion by which to justify and judge sample size adequacy in qualitative inquiry.1 Using saturation as a method to determine sample sizes has two major shortcomings. First, procedures or tangible guidelines for operationalizing the concept of saturation are elusive. To date, I have not seen any literature that indicates how saturation should or can be determined. How do you know when no new properties are revealed or new theoretical insights yielded? Interpreting Charmaz’s definition literally, particularly, the word “no,” one would be hard pressed to ever reach saturation. Certainly any newly administered data collection event would reveal some new property about the topic, even if minute in detail. More research needs to be conducted in this area of sampling. Second, employing saturation as a sampling yardstick is problematic from a practical standpoint as well. Saturation can only be determined after data collection has begun. Yet, as mentioned earlier, inductive sampling is often not possible due

14_107-Bernard.indb 241

6/9/14 10:03 AM

242

Greg Guest

Figure 7.7. Code creation of the course of data analysis (adapted from Guest et al. 2006).

to funding and IRB requirements or logistical and resource constraints. Waiting to reach saturation in the field may not be an option. How does one go about estimating when saturation will occur before data are collected? Only a few empirical studies exist to inform this question. Guest et al. (2006), for example, successively analyzed data from a study involving 60 in-depth interviews with female sex workers in two West African cities. The authors systematically documented the degree of data thematic expression and variability over the course of their analysis. After analyzing 12 of the 60 interviews, Guest et al. noted that 100 codes had been created and applied to the data. This represents 88% of the total number of codes (114) that were created and applied to the entire dataset of 60 interviews. Seventy percent of the 114 codes were identified within the first six interviews (Figure 7.7). The magic number of 6 to 12 interviews for identifying high-level themes/issues is consistent with one other empirically based study (Morgan et al. 2002), the mathematically derived cultural consensus theory (Romney et al. 1986), and an evidence-based recommendation from the field of human-computer interaction (Nielsen and Landauer 1993). The findings and recommendations from these studies come with two important caveats, however. One caveat is that all studies are different. The estimates above were derived from studies that shared three important factors. Each of the studies involved a relatively homogeneous sample, used a semi-structured data collection approach, and was interested in finding patterns across the sample. Not all forms of inquiry have each of these characteristics. As evidenced by the diversity of chapters in this volume, a plethora of approaches to data collection exist, each varying with respect to analytic purpose and degree of structure. And the range of possible research topics in anthropology is infinite. Nevertheless, some general observations can be made regarding saturation. Guest et al. (2013), for example, posit that the rapidity at which saturation is reached is related to five factors:

14_107-Bernard.indb 242

6/9/14 10:03 AM

7: Sampling and Selecting Participants in Field Research

243

1. The degree of instrument structure. The more structure embodied in the instrument, the sooner saturation will be reached. For studies using an unstructured instrument, or no instrument at all, saturation may never be reached. 2. The degree of sample homogeneity. The more homogeneous the sample, the quicker saturation is achieved. Groups that are alike on various dimensions are more likely to think in similar ways and have similar experiences. 3. The complexity and focus of the study topic. For more complex and intricate topics, it will take longer to reach saturation than simpler and more targeted topics. 4. Study purpose. Finding high-level common themes across a sample will generally require fewer sampling units than identifying the maximum range of variation within a sample. If you’re interested in finding the big issues, a small sample is often sufficient. Conversely, if your study objectives require the comprehensive documentation of all the minutia in your sample, you’ll need to sample substantially more than 6 or even 12 units. 5. Analyst categorization style. Some folks are “splitters.” They tend to see detail in everything and create codebooks accordingly. On the other end of the continuum are “lumpers”—individuals who like to group things into a few large conceptual categories. Codebooks created by splitters will invariably include a lot more codes than codebooks created by analysts with a lumper bent. The smaller the codebook being used to code the data, the quicker saturation will be achieved.

At one extreme, if you have a heterogeneous sample, a loosely structured instrument, a highly complex topic, are interested in assessing the range of variation, and your analysts are all splitters, you may never reach saturation! The magic numbers of 6 and 12 per group would be meaningless. They are estimates only. Also keep in mind that in some cases you may not even need, or be able to locate, six interviewees. If you are conducting a case study analysis, for example, there may be only one or two cases that fit your criteria. And an n of two is perfectly acceptable for many case study analyses. Use the findings and recommendations from the few existing empirical studies as a reference point. Use your judgment and knowledge of your research sites, populations, data collection methods, and analytic purpose to estimate your sample sizes. The second caveat is that, regardless of what makes methodological sense, your audience may have different standards. Despite the evidence regarding the adequacy of small sample sizes for many types of qualitative inquiry, many journals (especially those outside of anthropology) are less likely to except manuscripts based on qualitative data with smaller sample sizes. Likewise, Ph.D. committees, clients, and funders will more often than not insist on larger sample sizes than are really necessary. So, although increasing one’s sample size might not be necessary from a scientific perspective, you need to consider your audience’s predilections. In most cases, it doesn’t hurt to increase your sample size, and doing so enhances due diligence (real or perceived). In other words, plan for larger samples if you can.

Non-Probability Based: Quantitative Data Much survey research and other quantitative data collection methods employ non-probability samples. Intercept sampling at public facilities and nonsystematic observational activities fall into this category. Power calculations (see above) can be used in such contexts, with the caveat that the statistical generalizability of findings

14_107-Bernard.indb 243

6/9/14 10:03 AM

244

Greg Guest

will be limited. If for some reason statistical power cannot be calculated, the work of Graves (2002, 72–75) provides a data-driven recommendation for non-probability sampling, at least for comparison of frequencies. In his book, he presents an example of a 2 x 2 contingency table of height and weight of the San Francisco 49ers. Using a non-probabilistic sample of 30, Graves calculated a chi-square value of 3.75 for the association between height and weight (not statistically significant at the .05 level). However, when the sample size was doubled, but the relative proportions kept constant, the chi-square value doubled to 7.5, which is statistically significant. Based on these calculations, Graves (2002, 73) recommends collecting samples of between 60 and 120 for such correlative analyses. It is important to note here that statistical significance, as reflected in p-values, should not be the only consideration when reporting or reviewing statistical findings, because statistical significance is directly influenced by sample size. From a mathematical perspective, the larger the sample, the more likely one is to find a statistically significant relationship. Moreover, as Nakagawa and Cuthill (2007) observe, significance testing does not provide researchers with a measure of the magnitude of an effect of interest. It would not be overly informative, for example, if a weight loss program simply stated that members of the program lost a statistically significant amount of weight within six months. A discerning consumer would want to know how much weight (on average) members lost in that time period. The same principle applies to statistical reporting in general. It is therefore considered good practice to report the effect size once statistical significance has been observed (Nakagawa and Cuthill 2007; Wilkinson et al. 1999). Effect size can refer to a standardized measure (e.g., r, Cohen’s d, and odds ratio), or to an unstandardized measure (e.g., the raw difference between group means). To minimize confusion when talking about statistical findings, Onwuegbuzie and Daniel (2005), suggest using the phrase “statistical significance” and its variants when interpreting p-values, and the phrase “practical significance” when interpreting effect sizes. RECRUITING

Recruitment of study participants refers to the process by which participants are first informed about and, if eligible, asked to join a study. Recruitment is the means by which a sample is obtained. Table 7.5 outlines some of the more commonly used recruitment techniques and their characteristics (adapted from Guest et al. 2013). And keep in mind that involving local stakeholders and/or investigators is always a good idea in field research. Not only is their cooperation necessary for the successful implementation of a study, but they can tell you which recruitment techniques are most appropriate—and more importantly, which ones are inappropriate—for your study population. Even experienced ethnographers, who may be very familiar with a particular community, rely on local community members to provide guidance in the research process.

14_107-Bernard.indb 244

6/9/14 10:03 AM

7: Sampling and Selecting Participants in Field Research

245

Table 7.5. Common Recruitment Techniques and Characteristics Recruitment Type

Technique

Characteristics

Media-based

Posters/Flyers

Inexpensive, targeted coverage

—indirect and passive

Newspaper/Magazine

Expensive, relatively wide coverage

—does not accommodate probability sampling

Radio/TV

Expensive, wide but not targeted coverage

Internet

Inexpensive, little control over coverage

Investigator Initiated

Door-to-Door

Labor intensive, higher response rates

—conducive to probability sampling

Facility-based

Good response rates, generalizability limited to facility patrons

Intercept

More difficult (and sometimes not possible) to obtain a probabilistic sample

Email

Limits sample to Internet users, potentially low response rate (dependent on population and topic)

Phone

Limits sample to those with phones, obtaining sampling frame may be difficult

Mail

Inexpensive, lower response rates

Chain referral (participants refer other participants)

Good for hard-to-reach populations, participants must have substantial social ties

External referral (non-participants refer participants)

Can be effective, depending on who is referring (e.g., community leaders), possible ethical issues relating to participant volition

Socially based —does not accommodate probability sampling

Source: Adapted from Guest et al. 2012.

SUMMARY

Sampling is an extremely important component of the research design and implementation process. Developing an appropriate sampling plan can be a daunting task, involving the conceptual (and sometimes mathematical) integration of numerous competing factors. Table 7.6 is designed to help readers navigate this intricate process by summarizing and characterizing the key components in sampling design. The components are listed in the order they are presented in the chapter and represent a logical temporality in which each step might be initiated relative to the other steps. In real life, though, research design is a fuzzy and ever-evolving process, and the order in which the steps appear below is by no means immutable or necessarily linear. And, if your study involves multiple data collection methods, these steps will need to be applied to each different data collection method you employ. For readers who wish to read more about sampling, Johnnie Daniel’s Sampling Essentials (Daniel 2012) is an excellent resource.

14_107-Bernard.indb 245

6/9/14 10:03 AM

Table 7.6. A Summary of the Key Components in Planning Sampling and Recruiting Step

Considerations

Determine study’s primary population(s)

Key questions: • To whom do you wish to generalize, either theoretically or statistically, your study findings? • What scope is feasible within your timeline and the resources you have available? Tip: The title of your proposal/study/report should be a big clue

Determine sub-populations and stratification strategy

Key questions: • How much diversity do you need in your sample to reasonably say it’s “representative” of the larger population? • Do you plan to explicitly assess and report on the variability within your sample? • Do you plan to directly compare two or more subpopulations? Tips: • If yes, to bullet 1, make sure to include as diverse a range of individuals (or items) as possible • If yes to bullet 2, try to keep the sample sizes of subpopulations similar • If yes to bullet 3, try to keep the sample sizes of subpopulations similar AND, make sure that you have large enough sample sizes for each comparative group. • For observations data collection, pay special attention to temporal variability • Refer to Tables 7.1 and 7.2

Identify secondary study population(s)

Key questions: • What individuals or groups outside of your primary population are involved in the activity, experience, process that you are studying? • Which of these are likely to have different perspectives? Tip: Refer to Tables 7.1 and 7.2

Choose a priori or inductive approach

Key questions: • From a political, logistical and regulatory perspective, how much flexibility do you have with respect to study procedures, and sampling specifically? Tips: • Non-probability-based sampling for an individual method of data collection does not accommodate inductive approaches (findings from a probability-based method can, however, lead you to include another, independent probability based sample and/or method). • Opt for as much flexibility as you can. Like having extra funds, it’s nice to have in the bank if you need it.

Determine units of observation

A unit of observation is the item from, or about which, you collect data. It can be an individual, place, event, period in time, etc. Tip: Always collect data at the most granular level possible

Determine units of analysis

Your units of analysis are those elements in your study which you plan to assess variation amongst, and/or compare against each other. Tip: Units of analysis can never be smaller than your units of observation

14_107-Bernard.indb 246

6/9/14 10:03 AM

Step

Considerations

Select sampling method(s)

Key questions: • Is it feasible to collect data from/about every item in the study population? • Are your procedures flexible (i.e., allow for inductive sampling)? • How much control do you have over who you sample? • Do you wish to statistically infer your findings to a larger population? • How, if at all, do you want to stratify your population • Can you establish a reasonable sampling frame? • How is your population configured in time and space? • Is your population “hidden” and socially connected? Tip: Refer to Figure 7.6

Establish sample integration procedures (applicable only if selecting two or more samples)

Key questions: • How, if at all, are your samples dependent on one another? • How will your samples be employed chronologically with respect to each other? • At what level of abstraction do you wish to link data from my samples (if at all)? Tip: Refer to Figure 7.5

Determine sample size

Probabilistic Key questions: • How large is your study population? • How much precision do you need? • What are your analytic methods and objectives? Tips: • Online sample size calculators or software can give crude sample size estimates • Generally, the more variables and subgroups analyses you wish to include in your study the larger the sample you will need • If possible, consult a statistician for a large study with a quantitative analysis as your primary outcome Non-Probabilistic Key questions: • How much sampling flexibility do you have in your study (i.e., can you truly collect/analyze data until you reach “saturation”) • How much structure do your instruments/procedures have? • What is the degree of homogeneity in your sample relative to your topic of interest? • How complex is your topic? • What is the analytic purpose behind your sample selection? • Who is the primary audience for your research findings and what is their standard of evidence? Tip: It never hurts to over-sample, as long as: a) you have the time and funds, and b) it doesn’t detract from the quality of the data

Develop recruitment procedures

Key questions: • Do you want a probabilistic sample? • What resources do you have on the ground for recruitment? • How much can you include local stakeholders in your recruitment process and research overall? Tip: Refer to Table 7.5

14_107-Bernard.indb 247

6/9/14 10:03 AM

248

Greg Guest

NOTE

1. The term “theoretical saturation” derives specifically from the grounded theory literature and its procedures. Unfortunately, the term is often incorrectly used to describe general data saturation outside of a grounded theory context, in which theory building may not be part of the research purpose. REFERENCES

Bernard, H. 2000. Social research methods: Qualitative and quantitative approaches. Thousand Oaks, CA: Sage. Bernard, H. 2013. Social research methods: Qualitative and quantitative approaches, 2nd ed. Thousand Oaks, CA: Sage. Bertaux, D. 1981. From the life-history approach to the transformation of sociological practice. In Biography and society: The life history approach in the social sciences, ed. D. Bertaux, 29–45. London: Sage. Bluff, R. 1997. Evaluating qualitative research. British Journal of Midwifery 5: 232–35. Byrne, M. 2001. Evaluating the findings of qualitative research. Association of Operating Room Nurses Journal 73: 703–6. Charmaz, K. 2006. Constructing grounded theory: A practical guide through qualitative analysis. Thousand Oaks, CA: Sage. Cheek, J. 2000. An untold story: Doing funded qualitative research. In Handbook for Qualitative Research, 2nd ed., ed. N. K. Denzin and Y. S. Lincoln, 401–20. Thousand Oaks, CA: Sage. Cohen, J. 1988. Statistical power analysis for the behavioral sciences, 2nd ed. New York: Psychology Press. Creswell, J. 2006. Qualitative inquiry and research design: Choosing among five approaches, 2nd ed. Thousand Oaks, CA: Sage. Daniel, J. 2012. Sampling essentials: Practical guidelines for making sampling choices. Thousand Oaks, CA: Sage. Ferreira L., E. de Oliveira, H. Raymond et al. 2008. Use of time-location sampling for systematic behavioral surveillance of truck drivers in Brazil. AIDS and Behavior 12: S32–S38. Fetterman, D. 1998. Ethnography: Step by step, 2nd ed. Thousand Oaks, CA: Sage. Fossey, E., C. Harvey, F. McDermott, and L. Davidson. 2002. Understanding and evaluating qualitative research. Australian and New Zealand Journal of Psychiatry 36: 717–32. Gile, K., and M. Handcock. 2010. Respondent-driven sampling: An assessment of current methodology. Sociological Methodology 40: 285–327. Glaser, B., and A. Strauss. 1967. The discovery of grounded theory: Strategies for qualitative research. New Brunswick, ME: Aldine Transaction. Graves, T. 2002. Behavioral anthropology: Toward an integrated science of human behavior. Walnut Creek, CA: Rowman & Littlefield. Guest, G., A. Bunce, and L. Johnson. 2006. How many interviews are enough? An experiment with data saturation and variability. Field Methods 18: 59–82. Guest, G., H. Burke, D. Shattuck et al. 2010. Concurrency and HIV risk behavior among heterosexual men in Ghana: A mixed methods analysis. Poster presented at the International AIDS Conference, Vienna, Austria, July 18–23. Guest, G., E. Namey, and M. Mitchell. 2013. Collecting qualitative data: A field manual for applied research. Thousand Oaks, CA: Sage. Handwerker, P. 2001. Quick ethnography: A guide to rapid multi-method research. Lanham, MD: AltaMira. Heckathorn, D. 1997. Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems 44: 174–99.

14_107-Bernard.indb 248

6/9/14 10:03 AM

7: Sampling and Selecting Participants in Field Research

249

Hintze, W. N.d. In survey research, is more better? The diminishing returns and costs of large sample sizes. Unpublished paper. Johnson, J. C. 1990. Selecting ethnographic informants. Newbury Park, CA: Sage. Kayembe, P., M. Mapatano, A. Busangu et al. 2008. Determinants of consistent condom use among female commercial sex workers in the Democratic Republic of Congo: Implications for interventions. Sexually Transmitted Infections 84: 202–6. Kuzel, A. 1992. Sampling in qualitative inquiry. In Doing qualitative research, ed. B. Crabtree and W. Miller, 31–44. Newbury Park, CA: Sage. MacKellar, D. K. Gallagher, T. Finlayson et al. 2007. Surveillance of HIV risk and prevention behaviors of men who have sex with men—A national application of venue-based, time-space sampling. Public Health Reports 122: S39–S47. Magnani, R., K. Sabin, T. Saidel, and D. Heckathorn. 2005. Review of sampling hard-to-reach and hidden populations for HIV surveillance. AIDS 19(suppl 2): S67–S72. Morgan, M., B. Fischoff, A. Bostrom, and C. Atman. 2002. Risk communication: A mental models approach. New York: Cambridge University Press. Morse, J. 1994. Designing funded qualitative research. In Handbook for qualitative research, ed. N. K. Denzin and Y. S. Lincoln, 220–35. Thousand Oaks, CA: Sage. Morse, J. 1995. The significance of saturation. Qualitative Health Research 5: 147–49. Nakagawa, S., and I. Cuthill. 2007. Effect size, confidence interval and statistical significance: A practical guide for biologists. Biological Reviews 82: 591–605. Nielsen, J., and T. K. Landauer. 1993. A mathematical model of the finding of usability problems. Proceedings of INTERCHI 93: 206–13. Onwuegbuzie, A. J., and K. Collins. 2007. A typology of mixed methods sampling designs in social science research. The Qualitative Report 12: 281–316. Onwuegbuzie, A. J., and L. G. Daniel. 2005. Evidence-based guidelines for publishing articles in research in the schools and beyond. Research in the Schools 12: 1–11. Patton, M. 2002. Qualitative research and evaluation methods, 3rd ed. Thousand Oaks, CA: Sage. Robinson, E., D. Baron, L. Heise et al. 2011. Communications handbook for clinical trials. Research Triangle Park, NC: Family Health International. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–38. Sandelowski, M. 1995. Sample size in qualitative research. Research in Nursing and Health 18: 179–83. Semaan, S. 2010. Time-space sampling and respondent-driven sampling with hard-to-reach populations. Methodological Innovations Online 5: 60–75. http://www.pbs.plym.ac.uk/mi/ pdf/05-08-10/7.%20Semaan%20English%20%28formatted%29.pdf (accessed February 14, 2012). Spradley, J. 1979. The ethnographic interview. Belmont, CA: Wadsworth. Whyte, W. 1993 [1943]. Street corner society: The social structure of an Italian slum. Chicago: University of Chicago Press. Wilkinson, L., and the APA Task Force on Statistical Inference. 1999. Statistical methods in psychology journals: Guidelines and explanations. American Psychologist 54: 594–604.

14_107-Bernard.indb 249

6/9/14 10:03 AM

14_107-Bernard.indb 250

6/9/14 10:03 AM

CHAPTER 8

Participant Observation Kathleen Musante ( DeWalt)

Participant observation is a method in which a researcher takes part in the daily activities, rituals, interactions, and events of a group of people as one of the means of learning both the explicit and tacit aspects of their life routines and culture. Participant observation is considered almost universally as the central and defining method of ethnographic research and foundational in cultural anthropology, but it also has early roots in sociology and has been incorporated into qualitative research in a number of disciplines since at least the second half of the twentieth century. Since Malinowski published the description of his approach to ethnographic fieldwork, the types of questions addressed and the contexts in which research take place have shifted, yet the core elements of participant observation have remained stable. Moreover, ethnographic research based on participant observation has not only moved beyond functionalist theory, under which it developed, but has had a material and continuing impact on theoretical development in anthropology and other social sciences. As a way of knowing about the world of others, it provides a particular and unique way of generating novel understandings of the participants in our research. As a method, it is also at the core of contemporary discourses concerning the nature of observation, the degree to which we can hold our ideas up to a fair test, and the extent to which we can conduct research in ethical ways. In this chapter, I distinguish between what I see as the particular method of participant observation and the broader category of fieldwork. For writers such as McCall and Simmons (1969), Spradley (1980), Van Maanen (1988), Grills (1998), Marcus and Fischer (1986), and Agar (1996), the term “participant observation” subsumes the bulk of what some call fieldwork. Spradley (1980) used the term to refer to the general approach of fieldwork in ethnographic research, and Agar (1996) used it as a cover term for all of the observation and formal and informal interviewing in which anthropologists engage. For Agar, the interview is more important than the participation, and observation serves as the source of questions about which to interview. However, he sees participant observation as providing the context for the rest of the enterprise. In a similar vein, Schensul and LeCompte (2013) write that “Participant observation represents the starting point in ethnographic research” (p. 83). They see participant observation as the foundational method for ethnographic research. For Bernard (2011, 257), participant observation is a “strategic method” (i.e., a method that comprises several methods at once). In Bernard’s sense, one or more of the elements of a strategic method can be chosen, depending on the question being asked. Participant 251

14_107-Bernard.indb 251

6/9/14 10:03 AM

252

Kathleen Musante

observation “puts you where the action is and lets you collect data . . . any kind of data that you want, narratives or numbers” (2011, 343). In essence, for Bernard, it’s an approach to field research. Even as the nature of the field has changed over the years, participant observation remains a fundamental strategic method. For this chapter, I focus my comments on participant observation as one among several qualitative research methods used in ethnographic fieldwork. Although participant observation, as described below, may be foundational to the development of anthropology, it is not the only qualitative method used by anthropologists, nor do all anthropologists rely on participant observation as a key method in ethnographic research. For example, Agar (1986) argues that time spent participating and writing field notes is less important in research than in-depth interviewing. Other chapters in this volume focus on other qualitative and quantitative methods and techniques. The method of participant observation explicitly incorporates the collection and recording of information gained from participating in a social setting and observing what is happening in the setting explicitly into the analysis. All humans are participants and observers in all of their everyday interactions, but researchers engage in the systematic use of this information in formal analysis. Participant observation is a way to collect data in natural settings by ethnographers who observe and take part in the common and uncommon activities of the people being studied. Central to the method is a particular approach to the recording of observations in field notes. Participant observation may be the only way to capture tacit aspects of culture as praxis (Zahle 2012) as well as explicit culture. “Explicit culture makes up part of what we know, a level of knowledge people can communicate about with relative ease” (Spradley 1980, 7). Tacit aspects of culture may not be directly observable and often remain outside our awareness or consciousness. Tacit knowledge may become embodied in the way we learn to stand, sit, move, modulate our voices, and perform—the day-today practice of living in a culture (Desjarlais 1992; Desjarlais and Throop 2011; Zahle 2012). It certainly includes emotional responses to places, situations, and individuals (Crapanzano 2010; Davies and Spencer 2010; Hage 2010). We may find it difficult to articulate what it is that makes us feel comfortable (or uncomfortable) because these aspects of cultural knowledge often remain outside of our general consciousness. It is participation in the context around us that allows us to gain insight into the tacit. With thought and reflexive reflection, these phenomena can be available for analyses (Davies 2010b; Jackson 2010; Zahle 2012). HISTORICAL DEVELOPMENT OF PARTICIPANT OBSERVATION

Social scientists, including anthropologists, had carried out ethnographic fieldwork before him, but Malinowski (1961 [1922], 1978 [1935]) is usually credited with developing “something novel” (Sanjek 1990b; Stocking 1983)—an approach to research founded on long-term fieldwork and participation in daily activities. Firth (1985) noted that Malinowski did not invent long-term research, living with the subjects of research, or working in the vernacular; what Malinowski contributed was to “supply principles of systematic, intensive collection and interpretation of field data to a degree of sophistication not known before” (p. 30). Or, as Tedlock (1991) has said: “Malinowski’s invention lay in elevating the fieldwork method into a theory” (p. 69).

14_107-Bernard.indb 252

6/9/14 10:03 AM

8: Participant Observation

253

Malinowski’s discussion of his approach still serves as the fundamental description of the method of participant observation: Soon after I had established myself in Omarkana, Trobriand Islands, I began to take part, in a way, in the village life, to look forward to the important or festive events, to take personal interest in the gossip and the developments of the village occurrences; to wake up every morning to a new day, presenting itself to me more or less as it does to the natives. I would get out from under my mosquito net, to find around me the village life beginning to stir, or the people well advanced in their working day according to the hour or also the season, for they get up and begin their labors early or late, as work presses. As I went on my morning walk through the village, I could see intimate details of family life, of toilet, cooking, taking of meals; I could see the arrangements for the day’s work, people starting on their errands, or groups of men and women busy at some manufacturing tasks. Quarrels, jokes, family scenes, events usually trivial, sometimes dramatic but always significant, form the atmosphere of my daily life, as well as of theirs. It must be remembered that the natives saw me constantly every day, they ceased to be interested or alarmed, or made selfconscious by my presence, and I ceased to be a disturbing element in the tribal life which I was to study, altering it by my very approach, as always happens with a newcomer to every savage community. In fact, as they knew that would thrust my nose into everything, even where a well-mannered native would not dream of intruding, they finished by regarding me as a part and parcel of their life, a necessary evil or nuisance, mitigated by donations of tobacco. (1961 [1922], 7–8)

Malinowski further cautioned against living in compounds apart from the people under investigation like other “white men” do and insists on the need to live in the community. His approach differed from earlier forms of fieldwork in that it included an emphasis on everyday interactions and observations recorded in chronologically organized notes rather than on using directed inquiries into specific behaviors. Writing more than 70 years later, anthropologist Philippe Bourgois, living for more than 4 years in the neighborhoods in which he worked, described his approach to research in a more contemporary context in similar terms: I spent hundreds of nights on the street and in crackhouses observing dealers and addicts. I regularly tape recorded their conversations and life histories. Perhaps more important, I also visited their families, attending parties and intimate reunions—from Thanksgiving dinners to New Year’s Eve celebrations. I interviewed, and in many cases befriended, the spouses, lovers, siblings, mothers, grandmothers, and—when possible—the fathers and stepfathers of the crack dealers featured in these pages. (1995, 13)

More recently, working in an online environment, anthropologist Tom Boellstorff (2010) created an avatar and participated fully in the massively multiplayer online role-playing game (MMORPG) Second Life. As a participant observer he (i.e., his avatar) set up a household, made friends, participated in group activities, and engaged in informal (as well as more formal) interviewing. He attended weddings and parties, went to dance clubs and bars, dropped in on friends, and hung out. He bought things, sold things, and even chatted with other participants on the methodological issues of conducting research on a virtual society. Over the course of several years, he spent

14_107-Bernard.indb 253

6/9/14 10:03 AM

254

Kathleen Musante

thousands of hours in Second Life. His virtual fieldwork in Second Life was interspersed and overlapped with his actual field research on sexuality and HIV/AIDS in Indonesia. To differing extents, each of these ethnographers practiced the method of participant observation by living in the community they were researching, taking part in usual and unusual activities, “hanging out,” and conversing (as compared with interviewing), while consciously observing and ultimately recording what they observed. The basic method is no longer confined to village ethnography; it has been adapted to the study of “global nomads” (D’Andrea 2007) and other dispersed communities. Malinowski may have been the first anthropologist to describe this approach as a research method, but he wasn’t the first person or the first anthropologist to practice it. Rosalie Wax (1971) begins her discussion of the history of participant observation with the mention of Herodotus and other ancient writers, and, for later times, points to amateur writers such as Condrington, Callaway, and Bogoras, who spent extended time with the people they wrote about, spoke the languages, and described everyday life in the nineteenth century. Atkinson and Hammersley (1994) see participant observation as primarily a twentieth-century phenomenon, but they trace the philosophical and methodological roots of participant observation in the historicism of the Renaissance, and, in the nineteenth century, to the development of hermeneutics as an approach to understanding humans in different settings and time periods. The first anthropologist to write about using something akin to participant observation appears to be Frank Hamilton Cushing. Cushing was assigned by the Smithsonian Institution’s Bureau of Ethnology to collect information about Zuni Pueblo in the southwestern United States in 1879 (Hinsley 1983; Sanjek 1990b). His supervisor, Spencer Baird, expected him to spend about three months in Zuni Pueblo, but Cushing spent four and a half years there. Cushing wrote Baird in 1879 saying: “My method must succeed. I live among the Indians, I eat their food, and sleep in their houses. . . . On account of this, thank God, my notes will contain much that those of all other explorers have failed to communicate” (Green 1978, 136–37). Cushing learned to speak Zuni and was inducted into a Zuni Pueblo, and then the Bow priesthood (Green 1978; Green et al. 1990; Sanjek 1990b). In 1881, after two years with the Zuni, Cushing wrote to Baird saying: “I would be willing to devote, say, a year or two more to it to study for a period almost as great, from the inside, the life of the Zuni, as I have from the outside.” Cushing’s insistence on an internal, holistic, and organic understanding of Zuni life and culture born of long-term participation, fluency in the language, and intuitive, even poetic, insight presage both Malinowski’s approach and more contemporary approaches to ethnography. Cushing’s successor in Zuni and others criticized him for having become too involved with Zuni culture to write analytically and objectively about it, and he was accused of having “gone native” (Hinsley 1983; Tedlock 1991). However, Cushing left Zuni, married, and, with only a brief return to Zuni, spent the rest of his career in New England. Thomas Eakins’s well-known, romanticized, and controversial 1895 portrait of Cushing shows him dressed in Zuni clothing and surrounded by Zuni artifacts in a New England setting. The 1881–82 photo taken by John Hilliers, although taken in Zuni, is similar in that the setting and the dress appear contrived and artificial. Both

14_107-Bernard.indb 254

6/9/14 10:03 AM

8: Participant Observation

255

of these works suggest a more self-conscious adoption of the trappings of Zuni culture than the kind of adoption of the culture that going native might imply. Muller (2006) suggests that Frank Cushing’s Zuni-man identity ultimately has more to do with the ways in which the material culture of a subordinate group can be used by an individual from a dominant culture to construct an identity that provides him with authoritative power and privilege. It may be that this case illustrates a persistent conflict in the method of participant observation (i.e., the interplay of power in and identity for researchers who insert themselves into the lives of the subjects of their research), rather than the danger of going native. Beatrice Potter Webb (1926), the daughter of a British industrialist, was interested in understudying the plight of the London’s working-class poor. She sought to gain acceptance in London’s working-class neighborhoods and, in 1883, with the aid of her mother’s nurse, she disguised her identity and visited working-class neighborhoods. Later, she took a job as a rent collector in public housing and finally worked as a seamstress in a London sweatshop. Webb’s approach contained many of the elements characteristic of participant observation, but she did not live in the neighborhoods in which she was working. She certainly observed, but the degree of participation was limited to that which a rent collector would have had. Also, it is not clear that she systematically recorded field notes, although the stories of individuals she encountered do appear in her writing (Webb 1926; Webb and Webb 1902, 1932). Margaret Mead may have independently arrived at a method similar to Malinowski’s at about the same time. Sanjek (1990a) suggests that Mead had not read Malinowski’s book Argonauts of the Western Pacific when she traveled to Samoa in 1925 to conduct her first research based on original fieldwork among the Manu’a. In this project, she focused on the lives of adolescent girls but also carried out a more general ethnographic study of Manu’an social organization. Mead’s description of her approach, in the introduction of her ethnography of Manu’a, is similar in many ways to Malinowski’s and she speaks of “speech in action” as the heart of the method. My material comes not from half a dozen informants but from scores of individuals. With the exception of two informants, all work was done in the native language. . . .Very little of it was therefore gathered in formal interviews but was rather deviously extracted from the directed conversations of social groups, or at formal receptions which the chiefs of a village afforded me on account of my rank in the native social organization. . . . The concentration upon a small community and detailed observations of daily life provided me with a kind of field material rarely accessible to the field ethnographer. (Mead 1930, 5)

While a good deal of information was gathered through informal interviewing and conversation, Mead also undertook to learn the skills required of Manu’an girls: A rather disproportionate amount of my knowledge about Samoan custom and style came through my exposing myself to teaching—both in matters of etiquette, dancing, recitation of fa’alupenga—the stylized courtesy phrases and the making of artifacts. . . . I felt it was necessary to actually labor through the specific tasks . . . which a Samoan girl had to perform.

14_107-Bernard.indb 255

6/9/14 10:03 AM

256

Kathleen Musante

I combined the learner or novice’s point of view with that of the ethnographer, a more explicit interpretation of the term “participant observation”—a term that had not then been invented—than is usual, or even necessary, in any kind of field work. (1969, xix)

It is not clear when the actual term “participant observation” came into use to describe the method. The earliest use I can find is in the work of educational and community development researcher and educator Eduard C. Lindeman (1924; see also Maginn 2007). He argued that observation is a form of asking questions (“What is the individual doing?”), and that asking questions is a form of observation. He was also an early advocate of what has become known as the emic/etic approach. He said that the full answer to the question “What is going on?” comes both from the point of view of the researcher and from the point of view of the participant. To this end, then, he advocated using “participant–observers” in social research. In using this term, however, Lindeman was referring to participants who have been trained to be observers—“cooperating observers” (Lindeman 1924, 191)—rather than investigators who have adopted a participant role among a group of people.1 By the mid-1920s, a number of papers and books reference Lindeman’s concept of participant–observer. However, references to the method as described by Malinowski and Mead do not appear in the sociology literature until the 1930s. By 1940, the term “participant observation,” with reference specifically to the method as described by Malinowski and Mead, was in wide use in both anthropology and sociology and was included in the titles of papers by Lohman (1937) and F. Kluckhohn (1940). C. Kluckhohn’s (1943) obituary of Malinowski specifically references participant observation as a method. Kluckhohn did not believe that Malinowski invented the method of participant observation, but acknowledges that he was one of its best practitioners. As the nature of the “communities” in which social research takes place—and the purposes to which research is put—participant observation has taken on new subjects and forms. Forms of participant observation that retain some of the historically relevant aspects of the method have been adapted to short-term research in aid of program planning, implementation, and evaluation (Bentley et al. 1988; Scrimshaw and Gleason 1992; Scrimshaw and Hurtado 1987; Slim and Mitchell 1992). Anthropologists have also used participant observation to examine rapidly developing political events, such as the Arab Spring and 2011 protests in Cairo’s Tahrir Square (Winegar 2012). Also, as Boellstorff (2010) illustrates, participant observation in virtual communities has become commonplace as researchers examine virtual communities of many types based on computer-mediated communication (see Constable 2003; Hine 2000, 2005; Snodgrass, this volume). Even as researchers engage in multi-sited research, participant observation continues to be a fundamental approach to knowing (Marcus 1998).

Method and Theory In addition to one of the first explicit descriptions of participant observation, another of Malinowski’s major contributions to anthropology was the development of the functionalist theoretical perspective that assumed “that the total field of data under the observation of the fieldworker must somehow fit together and make sense” (Leach 1957, 120). Sanjek (1990b) argues that Malinowski’s particular approach to fieldwork resulted in the development of the functionalist theoretical approach. Holy (1984) says

14_107-Bernard.indb 256

6/9/14 10:03 AM

8: Participant Observation

257

that his theoretical perspective predated his fieldwork and influenced his method of collecting information. Wax (1971) suggests that Malinowski needed to invent functionalism to justify both his method and his promotion of that method following his return to academia after the war. Whatever the actual succession of events and intellectual development was, the method of participant observation was closely tied to functionalist theory at its beginning. While historically linked with functionalist theory, the method of participant observation is not tied to it. I believe that participant observation has, in fact, been the catalyst for the development of a number of theoretical perspectives. Many of the experimental ethnographies of the late twentieth century were built on research methods that placed the ethnographer in the middle of social life—a participant even when the writing focused more closely on specific relationships (Alter 2000; Behar 1993, 1996; Dumont 1991; Shostak 1983). The ethnographers who so explicitly situated themselves within the ethnographic encounter seem to me to have come directly from applying an analytic stance to their participation in the contexts in which they were working. As anthropologists have “embedded ethnographic subjects of study within contexts of a world system” (Marcus 1995, 97), multi-sited ethnography has become a common feature in the discipline. Marcus notes several different approaches to multi-sited ethnography in anthropology. Each focuses on a different object or phenomenon that moves across place and time: people, things (commodities), metaphors, plots, lives, conflicts, and so on. In each case, Marcus argues, the ethnographic analysis is, at least in part, founded on a deep understanding of place, people, and landscape. Anna Tsing’s (2005) brilliant multi-sited ethnographic study of global (and local) environmentalism is grounded in her deep knowledge of the Indonesian communities in which she carried out ethnographic research based on participant observation. Comparing that to her more conventional ethnography In the Realm of the Diamond Queen (Tsing 1993), one sees the critical impact of her participation and observation on the development of her analysis of the role of “friction” in the intersection of local and global processes. Gewertz and Errington (2010) draw on participant observation to enrich their understanding of the local implications of the global market in meat and meat byproducts, both from the point of view of the affluent who consume the more desirable cuts and the people who consume the cuts affluent consumers consider less desirable. Their study of the economic, social, political, and cultural life of mutton flaps draws on years of ethnographic research in several different settings, placed in a broader context of global narratives of power, economics, and health. They follow not only the commodity (mutton flaps) but also the concepts of “healthy” and “unhealthy” foods, placing these within the context of postcolonial power relations, and national discourses of resistance to postcolonial capitalist exploitation. However, they build a more nuanced understanding of the local role of mutton flaps in the cultures and cuisines of Pacific Islanders as a result of years of participating in some of the contexts they describe. Susser’s (2009) analysis of the global politics of AIDS in South Africa takes a more typical World Systems theoretical approach to but, again, her nuanced analysis of the South African policy and individuals’ experience with AIDS in the context of global policy was, in part, generated by a reflexive analysis of her fieldwork in South Africa.

14_107-Bernard.indb 257

6/9/14 10:03 AM

258

Kathleen Musante

Other multi-sited ethnographies, such as Constable’s (2003) study of mail-order brides (see below) draw more on interview materials, but Constable’s participation in the online life of the brides and bride seekers certainly fits the classic model of participant observation, although in a virtual way. What seems common across theoretical positions are the expectations that: (1) we can learn from observation (keeping in mind that the observer becomes a part of what is being observed); (2) being actively engaged in the lives of people brings the ethnographer closer to understanding the participants’ point of view; and (3) achieving understanding of people and their behaviors is possible. I would argue that the method itself pushes the researcher to reevaluate previous data and theory. As Picchi points out: “Participant observation—disallows selective learning about a people. Adjusting to a new culture provides on a daily basis many different types of experiences that prevent anthropologists from concentrating too assiduously on any one aspect of people’s traditions” (1992, 144). WHY INCORPORATE PARTICIPANT OBSERVATION IN RESEARCH DESIGN?

Participant observation is the foundational method of anthropological research, but it is rarely the only method used in contemporary ethnographic research. Most research designs that incorporate participant observation also include more structured interviewing, structured observation, and a host of other methods. In other fields, participant observation is often secondary to designs built on other qualitative and quantitative research methods. However, irrespective of the topic or principal methods used in doing social scientific studies, participant observation provides several advantages to field research. First, it enhances the quality of the data obtained during fieldwork. Second, it enhances the quality of the interpretation of data, whether those data are collected through participant observation or by other methods. That is, participant observation is thus both a data collection and an analytic tool. Third, it encourages the formulation of new research questions and hypotheses grounded in in-the-scene observation.

Quality of Data Collection and Interpretation In some cases, participant observation may be the only viable approach to research. Researchers who have worked in “deviant subcultures” with such groups as drug dealers (Adler 1985; Bourgois 1995, 1996), bank robbers and gangsters (Katz 1988), gangs (Brymer 1998), and poachers (Brymer 1991) have often argued that long-term participation in the setting was the only possible way to gain enough of the trust of participants to carry out research. Further, the use of more formal methods might have put off research participants. Sociologist Patricia Adler (Adler 1985; Adler and Adler 1987) argues that she would have been unable to gain any information concerning the subculture of drug dealing if she had not gained the trust of her informants by hanging out with them. In fact, Adler argues that she and her husband Peter could not have carried out the research without actively participating in the consumption, if not the marketing, of the drugs involved.2 Philippe Bourgois (1995, 1) writes: “I was forced into crack against my will.” He means that, like the Adlers, the choice of a place to live inserted him in a setting in

14_107-Bernard.indb 258

6/9/14 10:03 AM

8: Participant Observation

259

which he had, in some measure, to take part in the culture—if not the behaviors—of crack dealing and use. The view of the barrio and the structure of crack dealing within it that Bourgois presents could only have been made by someone who dedicated a long-term commitment to the research and the community. Unlike the Adlers, who conducted much of their study covertly (see below), Bourgois made it very clear he was carrying out research. He often openly audiotaped events and conversations and taped semi-structured interview. Even with the research intent explicit, however, he was able to gain the trust and confidence of participants in an illegal activity. Brymer (1998) was able to use long-term field research and long-term personal relationships (which, in his description, fit the definition of participant observation) to gain insight into two distinct subcultures known for their wariness of outsiders. These were Mexican American gangs in the Southwest and hunters and poachers in North America. Brymer is convinced that he has a much more nuanced view and insight into gangs and gang members than other people. This was possible because of his knowledge of a particular dialect of Spanish called pachuco, which he acquired growing up with Mexican cowboys working without documentation in the southwestern United States; seven years of working in several cities studying Mexican American gangs; hanging out with gang members; hauling them around in his 15-year-old station wagon; and his use of informal interviewing techniques. Like Adler and Bourgois, Brymer’s study of poaching was the result of stumbling onto this activity after buying a piece of land frequently poached on by local groups. He notes that it took two years of “testing” him before the locals incorporated him into the group (which, remember, was poaching on his land). The serendipitous and opportunistic flavor of the initial period of these interesting field research experiences suggests that perhaps once pulled into the experience of participant observation, the researcher is never again fully a participant in any setting but always part observer on the lookout for a new project. Hine (2000) and Constable (2003) both discuss how participation in online communities, not only as lurkers but also as participants in Listservs, newsgroups, and chat rooms, allowed them to appreciate the participant’s point of view even as they pioneered computer-mediated research. Researchers have been entering virtual worlds as cultural settings since the 1990s. Venues such as Second Life have become the sites of studies of everything from virtual culture to virtual economics (e.g., Boellstorff 2010; and Snodgrass, this volume). Engaging in apprenticeships is a common approach to participating in a context under study (Coy 1989; Singleton 1989; Tedlock 1991). In studying Yolmo healing, for example, medical anthropologist Robert Desjarlais (1992) trained to become an apprentice Yolmo shaman. To do so, he found it necessary to learn how to move and to experience his body as a Yolmo. He argues that much of what ethnographers can learn regarding peoples’ lives is at the level of the body. He notes that as he learned how to hold his body, sit on his heels, how to sip tea, caught the meaning of jokes, and participated in the practice of everyday life, these interactions shaped his “understanding of local values, patterns of actions, ways of being, moving, feeling” (Desjarlais 1992, 26). Through time, experiencing the body in this manner (including the residual, intermingling effect it had on how I stepped through a village, climbed a hill, or approached others)

14_107-Bernard.indb 259

6/9/14 10:03 AM

260

Kathleen Musante

influenced my understanding of Yolmo experiences; it hinted at new styles of behavior, ways of being and moving through space that I did not previously have access to. By using the body in different ways, I stumbled on (but never fully assimilated) practices distinct from my own. Touching head to heart merged thinking and feeling (two acts unsegregated in Yolmo society); a sense of the body as a vessel dynamically compact let me see Yolmo forms as vital plenums of organ and icon; and my loose assemblage of bent knees and jointed bones contributed to the springboard technology that gradually brought some force and ease to my shamanic “shaking.” (Desjarlais 1992, 27)

The process by which this might take place, although difficult to convey in words, comes as the result of sharing the lives of people over a significant amount of time. As Mead wrote: Pictures taken in the field show the extent to which I adapted to the style of the people with whom I was working. In photographs taken in Bali I look disassociated, sitting among a people each of whom was separated from the others. In Samoa the pictures show me dressed up, sitting and standing to display my Samoan costumes and rank; in Manus I am alert and tense, half strangled by a child clinging around my neck; in Arapesh I have become as soft and responsive as the people themselves. (1970b, 320)

The embodiment of cultural forms also informs the interpretation of meaning. In the most obvious ways, it allows us to understand nonverbal communication and to anticipate and understand responses. It shapes the way in which we interact with others and, in a more fundamental way, it shapes the way we interpret what we observe. Zahle (2012) argues that the technique of participant observation is the only method that can access the tacit aspects of culture and make them available for analysis as the result of engaging directly in the praxis of culture through participation. The fieldworker who does not attempt to experience the world of the observed through participant observation will find it much harder to critically examine research assumptions and beliefs and themselves (see Clifford 1997, 91). A part of this process is coming not only to understand, intellectually, the perspective of participants in the context in which the researcher is working, but to “feel” the point of view of the other (Grills 1998; Katz 1988; Throop 2012). The problems inherent in dealing with information that is tacit and embodied rather than explicit and intellectualized are obvious. How do we incorporate embodied knowledge in our analysis and writing? How do we make the insights it affords explicit enough to approach it analytically? How do we convey to others the type of insight we have (without sounding like we have gone native)? Several researchers have recently focused on the incorporation of emotional responses and the embodiment of tacit knowledge into explicit analysis (Crapanzano 2010; Davies and Spencer 2010; Hage 2010; Hastrup 2010; Luhrmann 2010).

Formulating New Research Questions Grounded theory approaches, which argue that in the early stages of research hypotheses and even broader theory should be generated inductively through primary data collection, have long stressed the role of qualitative research, including participant observation, in generating hypotheses (Glaser and Strauss 1967; Strauss and Corbin

14_107-Bernard.indb 260

6/9/14 10:03 AM

8: Participant Observation

261

1990, 1998). Even if a researcher does not take a grounded theory approach, qualitative research in general, and participant observation in particular, encourages the continual reassessment of initial research questions and hypotheses and facilitates the development of new hypotheses and questions as new insights occur as a result of increasing familiarity with the context (see Johnson and Hruschka, this volume). The process of analysis is inherently iterative. The active, insightful investigator should be continually reviewing field notes and transcripts and continually tossing out old ideas and posing new questions for study during the fieldwork and post fieldwork phases of research. “Being there” in the fullest sense means that our ideas and notions are continually challenged and “resisted” by the actions and words of those within the setting (paraphrasing Becker 1970; Grills 1998, 4). In my experience, participant observation provides moments in which the “scales fall from our eyes” and a new understanding or hypothesis presents itself. To come full circle, the tacit understandings gained during participant observation facilitate the intuitive moments when a selection of notes about events, people, and conversations comes together to provide us with a deeper insight and understanding of behavior. Living and participating in the research context forces us to place our particular focus of study within the wider context. THE BALANCE BETWEEN PARTICIPATING AND OBSERVING: THE RESEARCHER’S ROLE

The approach that ethnographers use in their field research and in participant observation is highly individualistic. Just as learning about a new social or cultural context is experiential, so is learning to use participant observation effectively. The method draws heavily on behavioral skills and already established social skills wedded to a flexible approach to new social situations. Every person brings to field research a complex mix of personal characteristics and a particular theoretical approach. They are influenced by the field context within which they work. Learning to participate in a new context means acquiring a set of understandings and reactions, that, in fact, we may not fully appreciate until we begin analysis. (A good reason to begin analysis early.) Those skills include both learning to be an observer and learning to be a participant. Among them are: fitting in, “active seeing,” short-term memory, informal interviewing, recording detailed field notes, and, perhaps, most importantly, patience.

Basic Skills of Participant Observation Mead suggests that before the new fieldworker goes to the field she should make an inventory of her own talents and skills, strengths and weaknesses. Important skills might include: memory for faces, ability to reproduce nonsense material from memory, ability to reproduce sensible material from memory, relative memory for things seen and things heard, ability to write and observe simultaneously, width of vision, ability to predict what will happen behind one by the expression on the faces of those in front, tolerance for continuous observation of the same type—e.g., kneeling with smoke in one’s eyes for four hours recording trance behavior, . . . attention span inside which attention is of the same quality, ability to attend to an unpleasant situation, susceptibility to disqualifying disgust

14_107-Bernard.indb 261

6/9/14 10:03 AM

262

Kathleen Musante

reactions, ability to resist the impulse to interrupt an unpleasant or disturbing sequence of behavior, tendency to identify in a partisan fashion with preferred individuals, etc. (Mead 1970a, 249)

All of these skills are fundamental to the method of participant observation. However, Mead does not suggest that a particular strength or deficit in these skills and abilities qualifies or disqualifies the researcher as an effective participant observer but that an honest accounting can be a guide in the selection of the most effective questions, venues, and recording techniques for any particular fieldworker. While some ethnographers come to participant observation with a better grounding in some of these skills compared with others, almost everyone can learn them or improve on existing skills.

Forms of Participation: The Participant–Observer Continuum A review of the work of researchers using participant observation from Webb and Cushing, to Malinowski and Mead, and through contemporary researchers such as Adler, Desjarlais, Bourgois, and Hecht and Johnson shows that successful researchers have employed a wide variety of strategies that range between pure observation and almost full participation in a culture. There is no absolute right way to balance participation and observation. Both experienced and novice fieldworkers find a balance right for them. The key is to make the form of the balance explicit in analysis. Spradley (1980) identified what he described as a continuum in the “degree of participation.” For Spradley, the purer observational form, “nonparticipation,” occurs when cultural knowledge is acquired by watching television or reading newspapers, diaries, or novels. Much information can be acquired in this way even though no active interaction with people is required. He defines “moderate participation” as when the ethnographer is present at the scene of the action but does not actively participate or interact, or only occasionally interacts, with people in it. Many anthropologists, for example, will live in their own house or perhaps even in a nearby community. They essentially “commute” to the field to interview or to participate in only some of the everyday activities of the community.3 “Active participation” refers to an ethnographer who engages in almost everything that other people are doing to try to learn the cultural rules for behavior. Finally, in “complete participation,” the ethnographer is or becomes a member of the group that is being studied. Examples of this include ethnographers who are and study jazz musicians or anthropologists who become hobos, cab drivers, or boat builders for a time (Johnson et al. 2006; Riemer 1977). It can also include native ethnographers, who turn an observer’s eye on their own communities and carry out analysis (e.g., Appadurai 1981; Kraidy 1999). Adler and Adler (1987) developed a somewhat different categorization scheme based on how much researchers participate in the groups they are studying. The three categories they use are peripheral membership, active membership, and full membership. Peripheral membership applies to researchers who become part of the scene, or of one group within it, but keep themselves from being drawn completely into it. They interact frequently and intensively enough to be recognized by members as insiders and to acquire first-hand information and insight. In active membership, the researcher takes on some or all of the roles of core members. Hecht (1998), for example, began to work for one of the agencies that served the needs of the street children he was studying

14_107-Bernard.indb 262

6/9/14 10:03 AM

8: Participant Observation

263

in northeast Brazil. Cushing was initiated into the Bow priesthood while working with the Zuni. In full membership, the researcher becomes immersed in the group and takes on an identity of the group. In both active and full membership, the researcher takes on some or all of the core membership responsibilities and duties. These roles most closely resemble the role of the classic participant–observer. Johnson and colleagues (2006) discuss the advantages of what they term “active participation” in research, which is close to Spradley’s notion of complete participation and Adler and Adler’s category of active membership. In a study of migratory commercial fishermen on the west coast of the United States, Johnson (1982, 1998) worked as a boat carpenter, tender worker, bookkeeper, and commercial crewman over several seasons to better understand commercial fishing. Weatherford (1986) took a job as a night clerk in a pornography store during a study of the red light district of Washington, DC. As part of larger study on how Chinese and Taiwanese migrants maintain social networks in California, Avenarius (2004) worked as a part-time employee in a Chinese-owned optometry store. She also became a member of several ethnic community organizations, including a choir, several dance groups, and church organizations. For Johnson, Avanarius, and Weatherford, being treated as insiders opened up new levels of understanding. Indeed, the people of the communities in which they worked joked with them about the ineptness of other researchers who did not take active participation roles and were seen always as outsiders. The formulations by Adler and Adler and Spradley suggest a much more rigid structure for the role of the researcher than is often the case. All roles are fluid and shifting and continually being constructed. The identities of the researcher and researched and the role of the researcher as participant are continually reconstituted through the process of fieldwork (Erickson 2011; Rashid 2007; Wanat 2008; Winchatz 2006). Both native and nonnative ethnographers can be simultaneously “insiders” and “outsiders” (Wanat 2008). Furthermore, roles may be continually renegotiated both explicitly and implicitly. In research with Chicano youth in Chicago in the early 1970s, sociologist Ruth Horowitz (1996) spent a good deal of time on the streets with members of fighting gangs. She discusses how her role as participant/observer was negotiated with the youth gangs over time. She became a “lady” to some (not available sexually)—someone who could go with them to buy guns but not participate in fights or looking for fights with other gangs. Some researchers (Adler 1985; Adler and Adler 1987; Bourgois 1995) are forced to take a greater degree of participation than they anticipated. In a gentler way, sociologist of religion David Martin (personal communication, 2000) and anthropologist John van Willigen (1989) were both surprised when asked at different times in different places to preach sermons in the churches in which they were working. Martin took the opportunity to tell his story in a sermon. Van Willigen was even more surprised when they presented him with the collection (somewhat over $4.00) at the end of the service. Finnström (2008) draws on the work of Swedish anthropologist Kaj Arhem (1994) to speak of the tension between participation and analysis (observation): As anthropologists, we do our best to participate in the works, questions, joys, and sorrows of our informants’ everyday life. Then we take a few steps back, to be able to reflect

14_107-Bernard.indb 263

6/9/14 10:03 AM

264

Kathleen Musante

on what we have learnt and experienced, again to step forward and participate. This we do daily in the field work encounter. (Finnström 2008, 27)

The balance between observation and participation, the specific kind of membership role a particular researcher will adopt (or fall into), and the emotional involvement of the researcher can be influenced by a number of factors. Personal characteristics of the researcher may influence the level of participation that an individual may choose or be forced to adopt (see below for a discussion of gender). I believe that most researchers who wish to use the method of participant observation can train themselves to do so effectively. However, some people feel so uncomfortable in the role of participant that they will be more effective researchers by primarily relying on other methods. I have often heard both new and seasoned researchers express fear that no one will accept them, no one will talk with them, they will be ignored, and the like. In my experience, for most researchers, this is not true. After an initial period of adjustment for both the researcher and the group or community, most researchers using participant observation find that they are integrated at least to some extent into the scene. However, the concern as to whether one will be accepted often masks the concern that the researcher does not have the personality to accept the role of participant in a particular setting. And, frankly, some people do not. Briggs (1986) is quite honest about how hard she found it to accept the participant role allowed her by the Inuit group with whom she worked. She was never able to adopt the role of Kapluna daughter, and her personality made it difficult to interact intimately with the other members of the community. Many researchers like Briggs, however, with an unsatisfying experience with participant observation in one setting find that in a different cultural setting they can get along quite well. Other personal characteristics may make it difficult or even impossible for particular researchers to participate as fully as they would like. Gender, age, class, and ethnicity may pose barriers to participation in some arenas important to research. Studies of adolescents conducted by adult researchers may be limited to observation and interview. Hecht was able to participate in the lives of street children in Brazil primarily as a volunteer in an organization that served street children and as the “producer” for the taped “radio interviews” carried out by other street children. In Hecht’s case, because participation as a street child was not feasible, he recruited participant observers in the sense meant by Lindeman (1924). That is, he was able to work through street children who took tape recorders and interviewed other children. I see a positive trend in anthropological research as writers increasingly making explicit the role, degree of participation, and point of view they have adopted in research and writing (see Alter 2000; Scheper-Hughes 1992, 1996; Susser 2009). Sanjek (1990c) argues that this is an important part of establishing validity in ethnographic reporting. Some researchers have combined this with an explicit discussion of their own emotional involvement in participant observation as part of a self-reflexive ethnography (Hage 2010; Wanat 2008). Janelle Taylor (2008), for example, explores caring and dementia based on participation in and observation of her mother’s Alzheimer’s disease. Perusek (2012) examines how culture frames the experience of cancer for patients and their families by recounting his brother’s death. Throop (2012) writes about empathy

14_107-Bernard.indb 264

6/9/14 10:03 AM

8: Participant Observation

265

and loss, based on participant observation of funerals on the island of Yap, as refracted through his own loss of a childhood friend to testicular cancer. These examples bring the participant observer into the story in ways that add new insight. Researchers should be aware of the compromises in access, objectivity, and community expectations that are being made at any particular place along the continuum. Further, in writing an ethnography, the particular place of the researcher on this continuum should be made clear. Methodological notes, field notes, and diary entries should report the level of involvement of the researcher in the community or group being studied and the degree to which the researcher comes to identify with the community. For Tedlock (1991), exploring the dynamic tension between participation and observation is critically important to analysis. In the past, many ethnographers wrote personal accounts of their field research under pseudonyms so as to maintain their reputations as professional ethnographers. From Tedlock’s perspective, however, these accounts should be part of the data of anthropology. She argues that we should engage in the “observation of participation,” an approach that she terms “narrative ethnography.” Narrative ethnography combines the approaches of writing a standard monograph about the people being studied (the Other) with an ethnographic memoir centering on the anthropologist (the Self). What is most valuable in the kinds of accounts that she is advocating is that they go a long way toward demystifying the process of doing ethnography. That is, by examining how other anthropologists have dealt with the degree of participation and with their emotional involvement, beginning ethnographers can better appreciate the circumstances, emotions, and reactions they are likely to experience when they begin their own field research.

Gaining Entry Gaining entry to a field site and beginning the process of building a research relationships (what we often call “rapport”; see below) can be a daunting experience for new researchers and experienced researchers in new settings. Entry into a field setting is often taken in stages. For nonnative research outside the home country, this may begin with taking care of issues such as research permits and visas. Even for native researchers, the communities in which we wish to carry out research often have gatekeepers (Kawulich 2011; Maginn 2007; Rashid 2007; Wanat 2008) (i.e., local leaders and organizations who represent, or claim to represent, the community in which the research will take place or who have access to the setting in which the research will take place). Some field situations do not have gatekeepers, but this is uncommon (Schensul and LeCompte 2013). For example, in studies of older adults in communities in Kentucky, we first made contact with the county judge executive of each county, and then with the county extension agents, representatives of community development organizations, and leaders of community organizations such as homemakers’ clubs, retired teachers’ organizations, and the directors of the senior citizens’ centers before we began contacting other agencies and individuals (DeWalt and DeWalt 2010). Kawulich (2011) outlines the process of identifying and contacting gatekeepers in Muscogee (Cree) communities. She identifies the establishment of trusting relationships,

14_107-Bernard.indb 265

6/9/14 10:03 AM

266

Kathleen Musante

honesty, and showing respect for local concerns, as critical in working through gatekeepers. Leith Mullings (Mullings 2000; Mullings and Wali 2001; Mullings et al. 2001) and her colleagues worked toward a participatory research endeavor and formed alliances with organizations such as the New York Urban League and set up an effective working community advisory board to introduce the research to the community and to guide the specific questions and approaches to the research (see S. Schensul et al., this volume). Nicole Constable (2003), when using computer-mediated communication as a critical component of her study of global romance and “mail-order” marriages, was encouraged by the moderator of an online group to become a member of the group as part of her research. He suggested that she lurk for a few days and then make her presence as a researcher known to the wider membership. When this was done, she received a number of hostile messages—she was “flamed.” She finally decided to discontinue her participation in that group. After she withdrew from the group, she received a larger number of responses supporting her participation and encouraging her to stay. In Constable’s case, the “gatekeeper” was supportive, but a number of participants were not, resulting in her withdrawal. However, the supportive moderator initiated an offshoot group of about 40 members who were comfortable with the research and welcomed her as a participant. Guimaraes (2005), in participant observation in a Brazilian Internet multimedia online sociability platform called Palace, early on met (through his avatar) a key member of the community who was organizing an online motorbike tour of Palace. His sponsor then introduced him to other members as “an anthropologist researching Palace” and assured the other members that Guimaraes was “a nice guy.” According to Guimaraes, “while talking about the organization and the pleasures of motorbike riding and travelling (either online or offline), it was possible to introduce myself as an ethnographer and establish rapport with some key members of the group as many of the organizers were Gods or Wizards” (p. 149).4 The members of Palace came to accept Guimaraes as a participant and researcher. Securing the support of gatekeepers and community leaders is necessary but not sufficient to entering a research setting. Most of the work we do is carried out with individuals who agree to participate with us in activities and conversations. Getting in is followed by getting along, and getting along relies on the ability of the researcher to develop research relationships with individuals based on respect and reciprocity, what we sometimes call “rapport.”

Rapport The establishment of rapport is often talked about as both an essential element in using participant observation as a tool as well as the principal goal of participant observation. Villa Rojas wrote about his and his collaborators’ field research in the Mayan region of Mexico: “Our close contact with local people has always led to excellent rapport, the only basis on which really reliable information can be obtained” (Villa Rojas 1979, 59). With shifts in the theoretical understanding of fieldwork, the concept of rapport has become contested ground. In the classical ethnographic literature, rapport is discussed

14_107-Bernard.indb 266

6/9/14 10:03 AM

8: Participant Observation

267

as a relationship of mutual trust and understanding between the researchers and the participants in research. It has been shorthand for “the threshold level of relations in fieldwork” (Marcus 1997, 86) that allows a researcher to communicate sufficiently as an insider to proceed. It has often embedded the notion of mutual understanding, and, in this sense, is sometime thought of as a state of interaction achieved when the researchers and participants come to share the same goals, at least to some extent. In other words, it is when members of the researched community and the researcher come to the point when each is committed to help the other achieve his or her goals; when participants participate in providing information for “the book” or study; and when the researcher approaches the interaction in a respectful and thoughtful way that allows the participants in research to tell their stories (van Willigen and DeWalt 1985). The narrative of building rapport recounted in Clifford Geertz’s (1973) essay on the Balinese cockfight, and recounted below, and a number of other narratives of building relationships in fieldwork, are interpreted by George Marcus as narratives of the “ironic entanglement of complicity with rapport” (Marcus 1997, 86). Marcus examines the complicit nature of the rapport-building process both in light of the power implications it poses and the role of the researcher. Conventional notions of rapport oversimplify the nature of the relationship between researchers and the subjects of research. In the Geertzes’ entry anecdote and Geertz’s later discussions, rapport becomes a complex act of complicity and one that the researcher comes to see as an opportunity from which he can benefit. Complicity in the rapport-building process can be seen simply as an innocent way to “break the ice,” or in a more sinister way, in which the researcher manipulates the relationship to conduct research. Taken to the most political end, a view of rapport as complicity could become a paralyzing moral dilemma, a position Marcus sees in the writing of Rosaldo (Marcus 1997). An alternative role is one of “collaboration,” (i.e., researchers and the subjects of research are seen as in dialogue in the construction of an ethnographic work as collaborators). Although not exactly what Marcus had in mind, the rhetoric of collaboration has become embedded in more participatory approaches to qualitative research in which the subjects of research are seen as “participants” rather than the older notion of “informants” (e.g., Morse 1991; S. Schensul et al., this volume). The critical understanding of the relationship between the researcher and the object of research that Marcus highlights is acknowledged by many contemporary ethnographers. Most have stepped away from the extreme positions of reflexivity and critique of the idea of rapport that make ethnographic research ethically/morally difficult, adopting a reflexive approach to the conduct of research that attempts to establish a more collaborative relationship. However, the notion of rapport, in its more complex understanding, seems to me to continue to be a useful shorthand for the relationships between researchers and the people in the communities they research. I believe that most researchers continue to work to develop a relationship of mutual understanding that allows the researcher to, at least partially, be accepted as an insider. In this more nuanced sense, rapport is built in much the same way as any other personal relationship. It is built over time. It requires that the researcher put effort into learning appropriate behavior in a setting; showing respect for people in a setting; being a good and careful listener; and being ready to reciprocate in appropriate ways. It

14_107-Bernard.indb 267

6/9/14 10:03 AM

268

Kathleen Musante

means that if the researcher expects the people they talk with to tell the “truth,” at least as they see it, the researcher must also be prepared to tell the truth.

Reciprocity Reciprocity has long been seen as a component of establishing rapport in fieldwork. Engaging in reciprocity may mean sharing personal goods and providing services (e.g., transportation) as appropriate. Rabinow (1977) found it easier not to have a car during fieldwork than to respond to constant requests for rides. Malinowski (1961 [1922]) and Denton (1970) both found they had to have sufficient supplies of treats and tobacco to meet the demands of the members of the communities they were researching. Reciprocity also includes sharing information. It certainly includes telling the truth when the researcher is asked about the research, his or her goals in research, or his or her own life story. Although answering personal questions can be uncomfortable, it is important to answer questions about religious beliefs and practices, values, opinions, and so on truthfully when asked directly. If you feel that you have to lie, how can you expect the participants in your research to be frank and open with you? However, the researcher can put their opinions and personal information in the most neutral way possible. The ethnographer often has to address the twin questions posed by the community: “What’s in this for you?” and “What’s in it for us?” People want to know what you will gain from this project. What’s in this for the researcher is often not an easy question to answer. Honesty is the only viable approach. Since most of us will not experience large direct economic gain from any project, the researcher can almost always reply that there is no real money involved. When substantial income can accrue from the study, the researcher may wish to consider sharing royalties with the community in an appropriate way. We know of writers who have funded scholarships for community students, contributed to community development projects, construction of parks, libraries, and the like. We generally answer that question in terms that are understandable in the local setting. For example: “We will write a book that will tell our colleagues what life is like here. This will help us finish our degree,” or “get a job,” or “keep our job,” and so on. Sometimes the answer is, truthfully, that we will be able to use the information to help find support for community projects, or the development of better policies. The response to “What’s in it for us?” can be equally problematic. Collings (2009) discusses this issue in a report on ongoing research with Canadian Inuit. As part of a staged approach to gaining official, community and individual agreement to carry out the research, he attended a community meeting to present his proposed project: I began by explaining the project’s aims, methods, and expected results. Not 3 minutes went by before a hand was in the air. What kinds of benefits will we see from this research? What good will your research be to us? Ooooh! I was excited on hearing a question I expected, even hoped, to hear. Trying not to sound too smug, I launched into my prepared answer: A database on subsistence production and food networks would be invaluable for community-based decisions on wildlife management and hunter support programs. I had hardly started in on my explanation when I noticed a woman tap the questioner on the shoulder. She whispered but

14_107-Bernard.indb 268

6/9/14 10:03 AM

8: Participant Observation

269

loudly enough that everyone in the room could hear her. “It says on page 3 he is going to give us lots of money.” The questioner turned back to me. “OK, it’s good then.” There were no more questions. The letters of support were ready the next day. (p. 139)

It is not always the case that there will be any monetary or other gain for the community. Our ethical standards and our institutional review boards require that we make that clear up-front as well. However, reciprocity often includes sharing research results with the community. This is essential to participatory research, and is especially important if the community can make good use of the information.

Breaking Through Many fieldworkers find that they can point to a single event or moment in which the groundwork for the development of true rapport and participation in the setting was established (e.g., Nader 1986; Stack 1996; Sterk 1996; Whyte 1996; Whyte and Whyte 1984). As I mentioned above, Clifford Geertz (1973) elegantly describes the event that allowed him and his wife to begin to gain acceptance as partial insiders in the Balinese village in which they worked. The Geertzes had been in the village for about a month, during which time the villagers treated them as though they were not there. They were rarely greeted, people seemed to look right through them, and people would move away when they approached. It was truly an anthropologist’s nightmare. Their breakthrough came as a result of a police raid on an illegal cockfight they were observing. Although the Geertzes could have stood their ground and presented the police with their credentials and permissions, they choose to run away with the rest of the villagers when the cockfight was raided. In Geertz’s words: On the established anthropological principle, “When in Rome,” my wife and I decided, only slightly less instantaneously than everyone else, that the thing to do was run too. We ran down the main village street, northward, away from where we were living. . . . About halfway down another fugitive ducked suddenly into a compound—his own it turned out—and we, seeing nothing ahead of us but rice fields, open country and a very high volcano, followed him. As the three of us came tumbling into the courtyard, his wife, who had apparently been through this sort of thing before, whipped out a table, a tablecloth, three chairs, and three cups of tea, and we all, without any explicit communication whatsoever, sat down, commenced to sip tea and sought to compose ourselves. (p. 415)

Moments later, when the police arrived, the Geertzes’ adopted host was able to provide a lengthy and accurate description of who they were, what they were doing in the village, and what permissions they had. In addition, he noted, the Geertzes had been in this compound all afternoon sipping tea and knew nothing about the cockfight. The bewildered police left. The Geertzes found that after that point they were enthusiastically incorporated into the community. An even more dramatic account of a single event that led to fuller participation is provided by Kornblum (1996) who was called on to stand with “his” gypsy family when they were attacked by Serbians in their camp outside Paris. In a moment of crisis, he became one of them. Sterk (1996) studied prostitutes in New York in the mid-’80s just as the magnitude of the AIDS epidemic had become clearer. She had

14_107-Bernard.indb 269

6/9/14 10:03 AM

270

Kathleen Musante

spent several days “on the stroll” in New York, but none of the prostitutes would speak with her. Finally, one challenged her, and she explained that she was from Amsterdam and wanted to know something about prostitution in the United States. Her knowledge of the life was challenged, and apparently, she answered appropriately, as she was befriended by the prostitute, Ann. Stack (1996) found her acceptance into the community in an African American section of a southern city happened more slowly but was facilitated when her car broke down. When that happened, Stack decided not to fix it. Without a car, she was more like the people of the community in which she was working; she was less visible and more a part of the community. Lack of mobility resulted in her spending more time in people’s homes and led, she writes, to her eventual acceptance into the community. We found in our work in Temascalcingo that the quality of the information we were receiving improved after our return from a three-week trip to the United States to renew our visas and take a break from fieldwork (DeWalt and DeWalt 2010). In a number of these cases, the breakthrough in acceptance was achieved when the researchers provided evidence that their relationship with the community was important and serious to them; when they demonstrated a more than passing commitment to a community. We returned to the community of Temascalcingo against the expectations of community members who had assumed we would not. Clifford and Hillary Geertz acted like Balinese villagers when they could have acted like privileged foreigners. Kornblum risked violence to stand with the people with whom he was working. These dramatic examples are quite vivid; rapport generally is established slowly, simply by continuing to live with and interact with a group of people and acting in ways that show your commitment to the community.

Talking the Talk It is not only worth the time and effort to learn the local language, it is imperative for participant observation. Debra Picchi (1992) recounts how she relearned this lesson during fieldwork with the Bakairi Indians in central Brazil. She went to Brazil as an ecological anthropologist with interests in demography, modes of production, and labor organization. After several weeks in the community, she overheard a conversation in Portuguese about her. “She doesn’t speak well,” Cici, a young mother of two complained loudly to her friends. “I think she must be as stupid as those giant anteaters that wander around the jungle.” Domingas, a frail old women, answered Cici softly, “You ask too much. She has been with us only a few weeks. You cannot expect her to speak Bakairi as well as we do. Be patient and she too will be a human being.” Maisa, another young mother spoke up, “I don’t know. I’m worried. The Alemao5 really is not learning very fast at all. Geraldina told me that she tried to explain to her the story of how the jaguar copulated with a human to produce our people. Geraldina said she didn‘t think that the girl understood two words of what she told her.” (pp. 144–55)

Picchi reflected that she had not been studying the Bakairi language diligently, as the people with whom she had been working also spoke Portuguese, and she was, at that point, more interested in collecting demographic and production data.

14_107-Bernard.indb 270

6/9/14 10:03 AM

8: Participant Observation

271

However, she also realized that the conversation had been staged for her benefit, and carried out in Portuguese to make sure she understood. After further reflection, she realized they were telling her that she would not be considered a true human being in their world until she could communicate in their language; finally, the women were trying to tell her that a grasp of the language was a necessary precondition for an effective study of their tradition. Winchatz (2006) addresses the problems inherent in interviewing in a foreign language. She argues that it is essential to ethnographic research, but even ethnographers with strong language capabilities may find a time when their linguistic abilities are not sufficient to avoid misunderstanding. In her work in Germany, Winchatz found that with careful attention, the points at which she misunderstood the comments of the participant in her research became points of analysis leading to a deeper understanding, much like Agar’s (1996) notion of “breakdowns.” She encourages ethnographers to embrace moments of misunderstanding and use them to seek instruction from respondents. Similarly, Bernard (2011) also encourages ethnographers to use language learning and real (and sometimes feigned) cultural naiveté as tools for developing relationships and learning. Finally, Winchatz notes that the assumption of novice/ learner’s role while trying to communicate in a nonnative language shifts power to the respondent. Winchatz also notes that even when one is carrying out research in one’s “native” language, misunderstandings can arise, and, with careful attention, also provide points for further analysis. BEING AN OBSERVER

Although developing an effective participant role is critical to the method of participant observation, it is not a tool of research unless the participant is also an effective observer. Researchers do vary with respect to their attentiveness to detail and their ability to recall detail. Anyone, however, can improve the degree to which he or she attends to detail, remembers detail, and perhaps most important, records detail. At its most basic, observation is just that: The researcher explicitly and selfconsciously attending to the events and people in the context he or she is studying. It is not just a visual phenomenon, but includes all of the senses. It also includes selfobservation, both of the way in which the investigator experiences the setting as a participant, the particular values and biases she or he brings to the setting (reflexivity) (e.g., Hage 2010; Luhrmann 2010), and observation of the impact of the observer on the research setting. In a way similar to choosing and taking on a participant role, the participant observer self-consciously takes on the role of observer as well. In fact, the need to be observing, and observing effectively, is one of the strains early in fieldwork that can lead to the experience of culture shock. The most important skill the participant observer develops is the ability to attend to details. Effective observation means “seeing” or “heeding” as much as possible in any situation. Mapping, counting, and active listening are all part of the method and can serve as exercises to improve observation. Noting the arrangement of physical space, the arrangement of people within that space, the specific activities and movement of people in a scene, the interaction among people in the scene (and with the researcher), the specific words spoken, and nonverbal interaction, including facial expressions,

14_107-Bernard.indb 271

6/9/14 10:03 AM

272

Kathleen Musante

are critical to observation (Pelto and Pelto 1978; Whyte and Whyte 1984). Mapping the social scene and the spatial layout of living and working spaces, whether a corral, a meeting, or a garden, becomes not only a way to examine spatial arrangements in various venues and events, but also a tool to focus observation and a way to train the observer to note detail carefully and record it faithfully. Much of the time a researcher is participating in an event or activity, he or she is also engaged in active listening, that is, listening attentively, making mental (sometimes written) notes about the conversation(s), and sometimes being prepared to offer a prompt when an aspect of the conversation touches on material important to the researcher, but when the information being offered is not entirely clear to the researcher. For Winchatz (2010), active listening, or, in her words, “heeding” is the critical skill in participant observing. Winchatz describes heeding as “a way of paying close attention to the discourse of native cultural group members through the act of active and engaged listening” (p. 341). LIMITS TO OBSERVATION

The impact of the observer’s point of view on the observed, how it is observed, and how it is recorded is not limited to the method of participant observation. Niels Bohr, the physicist credited with establishing the theory of quantum mechanics, understood the critical importance of the idea that every observer observes a phenomenon from the place from which the observer observes. Furthermore, the observer has no idea what is happening when not connecting with the phenomenon in some way (Bohr 1963; Frisch 1967). Bohr also noted that the act of observation (because it includes the need to intersect with the phenomenon) always has an impact on the object observed and, of course, the observation itself.6 Bohr’s discussion of the nature of observation in particle physics anticipated a similar reexamination of the nature of observation in social science that became a dominant theme in the critique of ethnography and ethnographic representation in the last quarter of the twentieth century. Although social science has mostly backed away from an extreme relativist position and paralyzing debate about the nature of observation and outright rejection of the notion of objectivity, the result has been the now routine inclusion of reflexivity in the planning, conduct, and reporting of research. Being reflexive involves a careful examination by investigators of the “place from which they observe” and their impact on the people and phenomena observed to better understand the relationships among the observer, the observed, and the report of the observation. I am with Bohr (1963) (and many others) on this question. We should not be asking “Is the research/researcher biased?” Instead, we need to ask “How is the researcher biased?” The “place from which the observer observes” and the impact of the observer on the observed then become explicitly part of the process of analysis and the reporting of research. Personal characteristics of ethnographers (e.g., gender, age, sexual orientation, ethnic affiliation), training and experience (e.g., language ability, quality of training, etc.), and their theoretical orientation all have an impact on observation. All of us bring biases, predisposition, and hang-ups to the field with us and we cannot completely escape these as we view other cultures. Aunger (2004) treats these biases

14_107-Bernard.indb 272

6/9/14 10:03 AM

8: Participant Observation

273

as an empirical matter rather than a cause for despair. He recommends using multiple participant observers, including members of the culture being studied, to determine precisely how researchers’ positionality shapes ethnographic observations. The “place from which the observer observes” is influenced by a mix of a number of more subtle characteristics, experiences, and situations that exist in a particular time and place for a particular observer. The most obvious is the theoretical position of the researcher. However, one of the important contributions of theoretical discussions in social science over the past two decades has been the axiomatic acceptance of the ethnographer as an individual with a specific gender, race, class affiliation, and any number of other characteristics that will have had an effect the specific ways in which a person will approach a research project, experience the research setting, be experienced by participants, and the expectations and perceptions he or she will bring. The gender of the ethnographer has an impact on several areas of the ethnographic enterprise. An important influence relates to the experiences of the ethnographer during the field research. Women in the field have often been harassed and have become victims of violence in ways different from men (Warren 1988). Just as men are often barred from situations in which they can know the intimate worlds of women, women ethnographers are sometimes barred from important parts of the worlds of men. The reports of ethnographers, however, suggest that women may find it easier to gain access to some aspects of men’s lives than male ethnographers find it to gain access to the worlds of women (Nader 1986; Teman 2010; Warren 1988). Differential access to the lives of women resulted in generations of predominantly male-biased ethnography, which has often paid little heed to the lives and concerns of women. Several classic ethnographic debates are, at least in part, the result of the different vantage point of the ethnographer. The gender of the ethnographer, for example, may be partially behind the discrepancies in the reports of Mead (1928) and Freeman (1983) concerning the sexual lives of Samoan girls. The view of economic exchange in the Trobriand Islands that Malinowski (1961 [1922]) presented is enlarged and enhanced by the work of Weiner (1988), who focused more of her work on exchanges involving women. Catherine Lutz (1988) has written about the experience of being a woman in Ifaluk. She noted that the lives of men and women on Ifaluk are sharply divided. Men and women, husbands and wives, may spend very little time together. This is not to say that women do not have high social status in many domains of life on Ifaluk. It is a matrilineal society, in which women contribute strongly to the economy through control of agricultural production, but that world is highly gendered. Lutz, however, had anticipated that she would be able to achieve the “genderless” or “generalized gender” status that a number of women ethnographers reported in other settings (see Fleuhr-Lobban and Lobban 1986; Jackson 1986; Lederman 1986; Warren 1988). In fact, she found that she could not achieve this but was required to conform to the gender expectations of the community around her. Lutz recorded the event that finally convinced her to abandon the hope that she could create a role outside of the Ifaluk system of expectations. She wrote: On the first evening, Tamalkar [her fictive “father”] and my “mother,” Ilefagomar, gave me a first elementary primer on what I should and should not do; I should say siro (respect

14_107-Bernard.indb 273

6/9/14 10:03 AM

274

Kathleen Musante

or excuse me) when passing a group of seated people; I should use the tag mawesh (sweetheart) when addressing someone; I should crouch down rather than remain standing if others were sitting; and, Tamalekar emphasized, I should not go into the island store if there were more than two men inside. I was to consider myself their daughter, they said, and Tamelakar would from then on refer to me before others as his daughter. A week later, a toi, or “mass meeting,” of the island’s men was called; on hearing this, I said that I would like to see it and was brought over to the meeting site by a middle-aged man from the village. Tamelakar was already there. Seeing me, he anxiously asked, “Where are you going?” and looked both uncomfortable and displeased when he heard I was interested in observing the meeting. As direct requests are rarely refused, he did not respond, but waved me to sit off to the side by his relatives. With this and subsequent encounters, such all-male occasions soon lost their interest for me, and I spent the great majority of my time with women in cook huts, gardens and birth houses. (Lutz 1988, 36–37)

Lutz expected to be able to choose the role that she would adopt if Ifaluk and anticipated that she would “allow” (1988, 33) herself to be socialized in arenas in which she was interested. If one is to be successful as a participant observer, however, that is not always possible. A number of women, though, have successfully stepped outside the prescribed roles for women within a particular cultural setting. Women have been involved with research on agricultural production, or other economic activities in which both men and women might work, but the spheres of men and women are different. Allen (1988), for example, was able to study the tasks of both men and women in her Peruvian research, although she notes that her original entrance into the community was eased because she was accompanied by a male colleague. Some of the most successful and fascinating fieldwork is conducted by teams of men and women. Murphy and Murphy (1974) provided a view of Mundurucu society that was almost unique for its time in the way that it placed in counterpoint the perspectives of men and women. The Murphys were able to do this because they had simultaneous access to different events and to different individuals during the same events. Men and women have access to different settings, different people, and different bodies of knowledge. Another, more recent example is the work of Barry and Bonnie Hewlett (2008, 2010). My own experience has been that having a man and woman involved in fieldwork at the same time has provided a more balanced view of community life, of key relationships, and of the interaction of households and families than we would have had if we had worked alone. FIELD NOTES (OR: IF YOU DIDN’T WRITE IT DOWN, THEN IT DIDN’T HAPPEN)

Field notes are the primary method of capturing data from participant observation. Researchers can audio- or video-record more formal interviews and events to record words and behaviors for later analysis, and record more formally the results of response to formal elicitation, time allocation, input and output of energy, and so on. However, the writing of field notes is virtually the only way for the researcher to record the observation of day-to-day events and behavior, overheard conversations, and casual interviews that are the primary materials of participant observation. I would argue that the writing of field notes is the method of participant observation. The construction of

14_107-Bernard.indb 274

6/9/14 10:03 AM

8: Participant Observation

275

field notes as detailed chronological narratives of the activities and conversations the ethnographer engaged in during a day of fieldwork was as important to the method that Malinowski described as his description of how he interacted and participated with the members of the community in which he lived. Field notes are fundamentally different than the transcribed narratives elicited by the direct questions at the researcher’s house (stereotypically on the veranda of his or her house). Reviewing a day’s events and writing up the notes are critical for constructing a research narrative that focuses on the analysis of events, behaviors, and speech within the social context. If we argue, as I have above, that participant observation is a method that encourages critical examination of the researcher’s ideas by careful observation of the phenomena under study through participation, then the recording of observation in a careful and detailed way is essential for the analytic steps that follow. Before 1990, relatively little had been written about the nature of field notes and how anthropologists record observation. The sixth edition of Notes and Queries (Seligman 1951), devotes only one and a half pages to a discussion of “descriptive notes” as one of four essential types of documentation. (The other three are: maps, plans, and diagrams; texts; and genealogical and census data.) Notes and Queries suggests three kinds of notes, which are still relevant today: records of events observed and information given (in which the researcher takes time to interview or converse with participants as events take place); records of prolonged activities and ceremonies (in which interview is not feasible); and, following the practice of contemporary ethnographers, a set of chronological, daily notes, which the committee called a journal (but that is distinct from the personal diary which a number of ethnographers keep [e.g., Malinowski 1967]). Pelto (1970) and Pelto and Pelto (1978) provide approximately two pages of discussion of field notes in each edition of their book on Anthropological Research. Most of this space is devoted to examples of the level of detail (high) that they see as desirable in recording field notes. Since 1990, however, a number of articles and chapters and several excellent volumes devoted to the recording of field notes have been published (Bernard 2011; Emerson et al. 2011; Sanjek 1990c). There are several important points to recognize about field notes and their relationship to the participant observation method. The first is that observations are not data unless they are recorded in some fashion for further analysis. Even though it seems after a few months in the field that common events and their variations will remain indelibly etched in the researcher’s mind for all time, memory is unfortunately more fleeting and less trustworthy than that. We lose the fine detail of observations and conversation all too quickly. The admonition in Notes and Queries that “It is unwise to trust to memory; notes should be written as soon as possible” (Seligman 1951, 51) is still relevant today. Participant observation is an iterative process, and, as we have noted, part of what occurs is the development of an analytic understanding of meanings, events, and contexts by the researcher. Sanjek (1990b) points out that the researchers who pioneered the method—Malinowski and Mead—knew this. They continually read and reread field notes, searching for things they did not understand, or about which they felt they had incomplete information (what Agar refers to as “breakdowns” [1986]), so as

14_107-Bernard.indb 275

6/9/14 10:03 AM

276

Kathleen Musante

to direct the flow of subsequent conversations or interviews. Mead and Malinowski recorded both observations and reflections on their fieldwork experience in field notes and personal diaries. If the researcher’s daily reactions to events and contexts are not recorded, it will be virtually impossible to reconstruct the development of understanding and to be able to review the growing relationship between the researcher and study participants in a manner that allows for reflexivity at the end of the process. The second point is that field notes are simultaneously data and analysis. They should be the careful record of observation, conversation, and informal interviews carried out on a day-by-day basis by the researcher. At the same time, the notes are a research product, constructed by the researcher. This inherent contradiction evident in field notes is part and parcel of the continuing discussions surrounding the nature of anthropological inquiry and the nature of ethnography. James Clifford addresses this question directly in his chapter Notes on (Field)notes (1990). Field notes are, at least, one step removed from an observation. They are a construction of the ethnographer and are part of the process of analysis. As one of Jackson’s (1990) ethnographer respondents said about field notes: “Each anthropologist knows it is a dialectic. The informant creates it; you create it together. There must be a tremendous sense of responsibility in it, that is, a sense of political history, one version” (p. 14). Clifford (1990) notes that the view of the field note as “pure inscription” (i.e., pure recording) cannot be sustained. Speaking of description (thick or otherwise), he writes: “Ethnography cannot, in practice, maintain a constant descriptive relationship to cultural phenomena. It can maintain such a relationship only to what is produced in field notes” (p. 68). Jot notes, those notes taken at the time of the observation, are a bit more immediate. However, Jackson’s respondents say that some people may become suspicious when the ethnographer takes notes, and a few people who are “otherwise friendly may never tolerate the practice.” Jackson (1990) reports that a number of the ethnographers she interviewed found that taking field notes in front of participants was uncomfortable and objectifying. However, others found that participants were insulted when notes were not taken, suggesting that what they had to say was not important enough to record. Evans-Pritchard also addressed this concern in his 1973 essay on fieldwork. Whyte (Whyte and Whyte 1984) went to relatively great lengths to avoid taking notes in front of participants in his research in Cornerville. To gain accurate maps of interaction in a men’s club, he would go to the bathroom or home to record events. Philippe Bourgois (1995, 1996), on the other hand, openly taped both interviews and conversations as he hung out with crack dealers in El Barrio in East Harlem. Participants in his study even joked about “how the book was coming” (1995, 27), an experience many of us have had with participants, who may even be anxious to see the finished work. There are several types and formats for notes taken during fieldwork and any number of ways of keeping notes and managing them (Bernard 2011; Emerson et al. 2011; Mead 1930; Sanjek 1990b). Researchers following different theoretical traditions approach and value field notes differently. However, all of the systems share several basic elements. Researchers jot down words and phrases, even whole sentences in a notebook during the course of a day or event (jot notes or scratch notes); they write fuller, more detailed

14_107-Bernard.indb 276

6/9/14 10:03 AM

8: Participant Observation

277

notes during more quiet time, reflecting on the day’s events (field notes); they review their notes regularly and write notes on their notes (analytic notes or meta-notes); they keep a diary or journal; they keep a calendar or log of events, listing activities in which they participated, events they observed, and people with whom they spoke. They even have some “notes” in their heads that never actually get written down (head notes). All of these acts of recording sum up to the body of stuff we call field notes. Head notes and diaries are increasingly being incorporated into analysis as part of a reflexive ethnography and as a means of incorporating the emotional response of the ethnographer and other tacit information into the analysis (Davies 2010b; Hage 2010; Hastrup 2010; Jackson 2010; Luhrmann 2010). Contemporary technology has altered the ways in which we record field notes. Mead typed her notes on a manual typewriter, in triplicate, and read and reread them making marginal notes on sheets of paper. When I first started field research (1970), I took jot notes in a small notebook and wrote field notes by typing on codeable note cards (In-Decks). Later, computers allowed for electronic text files that were searchable and eventually codeable. With the development of reliable voice recognition software, it is far easier to record longer jot notes and even longer field notes while in the scene using a digital voice recorder and transforming the file into text. Computer-assisted qualitative data analysis software (CAQDAS) allows for the easy electronic coding and retrieval of codes. They now all include the ability to review and code video and audio files. I do not recommend coding the audio files of recorded field notes directly as a usual practice. From my point of view, it hampers analysis of the field notes because it is still easier to manage and code a text file than an audio file. Voice recognition software makes it easy to transcribe audio recordings of the researcher’s own voice, but it is not yet possible to automate transcription of interviews with multiple voices (but see Park and Zeanah 2005). ETHICAL CONCERNS

The ethics of ethnographic research will be addressed in greater detail elsewhere in this volume. However, participant observation as a method raises unique ethical questions as compared with other methods. While it may be clear to most individuals who are being interviewed using structured interview schedules, or whose interviews are being recorded in some way, that the information resulting from these activities will be used in the research carried out by the fieldworker, the activities carried out during participant observation are less clearly so. The fieldworker is traveling alongside community members, participating in events, work, leisure activities, and hanging out. Community companions will probably not be aware that the researcher will faithfully record an account of these events as soon as possible, and that this will form a data core for analysis. Even if a researcher makes it clear that he or she will write a book or report on his or her experiences, participants may not realize that what they share as gossip during informal conversations will form part of this report. We want our companions to forget, for a time at least, that we are outsiders. We want them to become so comfortable with us as community participants that they will share insights and information that only insiders would know. This is the strength of the method. Thorne has summarized a number of the casually deceptive positions taken by ethnographic researchers in the past, which she sees as circumventing the issues of

14_107-Bernard.indb 277

6/9/14 10:03 AM

278

Kathleen Musante

consent and disclosure (Thorne 1980). Therefore, as Punch (1994), quoting Ditton points out, participant observation is inevitably unethical “by virtue of being interactionally deceitful” (Ditton 1977, 10). We have all been impressed by the degree to which people will suspend a conversation or interaction to remind us that the topic about which they are speaking is important, and to make sure that we will put it “in the book.” But it is more often the case that the people forget that casual interactions may form part of the data to be used in analysis. Several of the researchers mentioned earlier in this chapter carried out some research covertly. Webb, when she began trying to understand the poor of London and took on the role of rent collector to do so, did not reveal that she would write about the experience of the people with whom she interacted. The Adlers did not reveal to the drug dealers with whom they were hanging out that they would be writing an ethnography that included them. Johnson et al. (2006) comment that their roles as active participants would probably not be considered ethical today. On the other hand, Lugosi (2006) reviews several of these issues involved in covert research, arguing that concealment is sometimes necessary to carry out research. One of the most important points Lugosi makes in his review, is that, as I noted above, when the method is participant observation, it may be impossible to conduct the research completely overtly. It may be necessary to think in terms of a continuum of covert to overt research, with different points along the line raising different ethical concerns. As Marcus (1997) suggests, the establishment of rapport and, to use Lugosi’s (2006, 545) term, “rapport management,” can be thought of as inherently deceitful. Researchers approach participants, visit their homes, take part in daily activities, give gifts, all with at least some intent to engage people to participate in the research process. Lugosi carried out research on hospitality and hospitality exchange in bars and other venues. During the research period, he became the manager of one of his principal research venues. He describes a process through which he communicated his research in increasingly more overt ways to the patrons of the bar over time, a sort of gradual reveal of the research agenda. However, Bourgois openly recorded events and conversations throughout his research with drug dealers in New York. That he was writing a book was never concealed from the people with whom he interacted. In my experience, openness regarding the research enterprise has rarely compromised the collection of information. There certainly have been times when the quality of the conversation shifted when I turned off the recorder, but for me keeping the notebook open is critical to making the research nature of the relationship clear. The Statement on Ethics: Principles of Professional Responsibility adopted by the American Anthropological Association in 2012 (American Anthropological Association 2012) calls on researchers to be transparent and open about their research but is not prescriptive concerning what needs to be revealed and when. At the same time, it stresses the right of individuals to know that they are the subjects of research and the need for transparency regarding the purposes and nature of research. This requirement reflects the broader principle of informed consent in ethnographic research (Murphy and Dingwall 2007). Researchers using participant observation are obligated continually to assess the ethical issues raised during research (see Fluehr-Lobban, this volume). Ethnographers have often taken the position that providing anonymity for communities and individuals is sufficient to protect them. However, today it is increasingly

14_107-Bernard.indb 278

6/9/14 10:03 AM

8: Participant Observation

279

difficult to even suggest that the identities of communities will remain hidden. Punch recommends that we carry out research with this powerful method, but “think a bit first” (1994, 95). Although this is probably good advice, because we will continue to do participant observation, the responsibility the ethnographer bears “to do no harm to our informants” (American Anthropological Association 2012) is great and Punch’s admonition appears to trivialize it. It is the ethnographer’s responsibility not only to think a bit first, but to make conscious decisions on what to report and what to decline to report, based on careful consideration of the ethical dimensions of the impact of information on those who provide it, and the goals of research. Whyte (Whyte and Whyte 1984) reminds us that much of our work can be put to unintended use. It is important to emphasize that ethical questions surround not only the published information we provide but also relate to our field notes. First, there is the ethical question concerning the public taking of notes. Taking notes openly, at least part of the time, reinforces for participants that what is being done is research. The participants in Bourgois’s crack research knew he was writing a book, and his open taping of interviews and even conversations on the street made it clear to participants what was on the record and off. In fact, they sometimes suggested that he turn on his recorder when he turned it off. Openly taking notes, or recording events, alleviates some of the concern that participants lose awareness that they are participating in research. The second issue concerns preserving anonymity for participants identified in field notes. Not only is there the potential that field notes can be subpoenaed, governmental funding organizations such as NSF and NIH are becoming increasingly concerned with research integrity and falsification of data in funded research. They are demanding that primary data be available for review by others. Universities such as my own are developing policies concerning research integrity of nonfunded, as well as funded, research in which data, including field notes, would be available for inspection by internal review boards (IRBs) should allegations of a breach of research integrity be alleged. Most IRBs for the use of human subjects currently require that all questionnaires and transcribed interview data be stored without names, but field notes have always constituted a gray area. With computerization, however, it becomes relatively easy to assign code names or numbers to participants and use these from the outset or use the global search-and-replace facilities of word processing programs to expurgate real names from field notes. The research integrity policy statement for the University of Pittsburgh, for example, now suggests that real names be expurgated from field notes against the possibility that they will be requested by others. Ethnographers need to be aware of the implications of relationships and obligations that they incur in the field. Many anthropologists enter into fictive kinship relationships to find a place in a community. Catherine Lutz (1988) became a “daughter” to fit into Ifaluk society. Assuming this kind of relationship has a number of implications and adds a series of responsibilities for the anthropologists that should be carefully considered before the identity is accepted. Becky Ross, whose story forms one of the ethical case studies published by the AAA (Cassell and Jacobs 1997), was “adopted” into a family as a daughter and granddaughter. When it became necessary to assume the responsibilities of a true granddaughter and care for an elderly “grandfather” she did it, at personal expense (loss of precious field time), until the biological grandchildren were available to take over.

14_107-Bernard.indb 279

6/9/14 10:03 AM

280

Kathleen Musante

Limits to Participation A final issue relates to what we might call limits to participation. There are some dramatic cases of the need to establish limits to participation because engaging in these behaviors may be illegal, dangerous to the personal health of the ethnographer, or both. Philippe Bourgois (1995, 1996), became quite involved with the drug dealers with whom he was working, although he abhorred the violence and other activities in which they engaged. There are also many accounts of ethnographers being confronted with whether to engage in romantic and/or sexual involvements with members of communities they study (see Kulick and Willson 1995). Deciding how much to participate or not to participate in the life of the people being studied is not an easy decision. In addition, there are often occasions during which the ethnographer faces a difficult decision about whether or not to intervene in a situation. During his research with the Yanomama of South America, Kenneth Good (1991) came across a situation in which a group of teenage boys and three older women were engaged in a tug-of-war. A woman whom he had befriended earlier was in the middle. Good ascertained that the teenagers were trying to drag the woman off to rape her while the old women were trying to protect her. I stood there, my heart pounding. I had no doubt I could scare these kids away. They were half-afraid of me anyway, and if I picked up a stick and gave a good loud, threatening yell, they’d scatter like the wind. On the other hand, I was an anthropologist, not a policeman. I wasn’t supposed to take sides and make value judgments and direct their behavior. This kind of thing went on. If a woman left her village and showed up somewhere else unattached, chances were she’d be raped. She knew it, they knew it. It was expected behavior. What was I supposed to do, I thought, try to inject my own standards of morality? I hadn’t come down here to change these people or because I thought I’d love everything they did; I’d come to study them. (Good 1991, 102–3)

Good decided to do nothing but wrote that this was a turning point in his integration into the community. A month later, Good did intervene in a similar situation (1991, 104–5). Every ethnographer sooner or later faces dilemmas like these that become difficult ethical issues (Rynkiewich and Spradley 1976; Scheper-Hughes 1996). We may appeal to cultural relativism or to the role of “objective observer” to avoid intervention in situations like those faced by Good. In cases, however, in which we see the people with whom we are working being exploited, subjected to violence, or damaged in some other way, it is increasingly difficult to justify not intervening. Nash concluded that the world should not be seen as simply a laboratory in which we carry out our observations but rather a community in which we are “coparticipants with our informants” (1976, 164). She used this as an argument for working to try to help the tin miners she was studying in Bolivia fight for their rights. Scheper-Hughes (1996) argues even more strongly that the role of the ethnographer includes activism. She describes how she chose to intervene in the punishment of several young boys caught stealing in a South African village. She intervened to take an accused boy to the hospital to save his life after his punishment at the hands of villagers, even though her research was, in part, on the outcomes of popular justice.

14_107-Bernard.indb 280

6/9/14 10:03 AM

8: Participant Observation

281

THE COSTS OF PARTICIPATION: CULTURE SHOCK

One of the hallmarks of using participant observation as a key technique in an unfamiliar cultural context is the experience of what has come to be known as “culture shock” (Bock 1970; DuBois 1951; Oberg 1954; Ward et al. 2001). Classic descriptions of culture shock refer to it as the culmination of unease felt by the participant observer as a result of not being able to successfully operate in a new cultural setting; getting all the cues wrong; not being able to anticipate proper behavior; dealing with behaviors of others that are, to the home culture of the fieldworker inappropriate, shocking, dirty, immoral, or just plain different, but are perfectly acceptable within the context of the community in which the fieldworker finds him- or herself (Bock 1970; Oberg 1954; Ward et al. 2001). Culture shock surely includes what we would call homesickness, and in many cases, the loneliness of leaving all loved ones behind for a while. But culture shock is more than homesickness and loneliness. The demands of trying to conduct research successfully in a very different cultural context exact an additional toll. Oberg (1954, 2) discussed culture shock as “an occupational disease of people who have been suddenly transplanted abroad. Like most ailments it has its own etiology, symptoms, and cure.” He attributes culture shock to the anxiety that accompanies the loss of familiar signs and symbols (cues) of social intercourse when individuals move into different cultural contexts. Most of the cues of social intercourse are subtle and may be unconscious. The sojourner7 is a “fish out of water” (Oberg 1954). As a result of frustration and anxiety, the sojourner experiences the new culture as culturally “bad” and the home culture is romanticized as “good.” Ward et al. (2001) identify a series of cultural differences in the way that people communicate that appear to be related to the feeling of being a fish out of water for the sojourner. These include differences in etiquette, such as the way in which questions are phrased and requests are refused; ways of resolving conflict, such as the differences in the ways in which people in collectivist and individualist cultures negotiate; nonverbal communication, such as the mutual gaze, the degree of bodily touching, and the appropriateness of gestures; rules and conventions, such as punctuality and reciprocity; and forms of address. All of these have been shown to vary across cultural settings and to contribute to dis-ease in communications. Davies (2010a, 2010b) writes about the specific impacts of research based on “immersion” in a fieldwork setting. In contrasting the experience of the casual sojourner— tourist or otherwise—and researchers who immerse themselves in the fieldwork setting, he notes that: [the] purification of unpalatable experience is not possible for anthropologists. They reside in the field unprotected and subject to whatever chance thrusts upon them. Their dissonance is not managed by a corporate other, but by their private and spontaneous selves. . . . And whatever distance traversed to reach the field, the same dilemma confronts them—for as soon as concepts of “home” and “field “are understood in psychological rather than geographical terms “home” and “field” become whatever one experiences them as. (Davies 2010a, 90)

Davies uses the concept of dissonance to refer to what he calls the “conflict between self and environment” experienced when first entering the field and later when leaving

14_107-Bernard.indb 281

6/9/14 10:03 AM

282

Kathleen Musante

it. Dissonance is very close to the notion of culture shock discussed by others. Davies outlines a series of outcomes related to dissonance for fieldworkers. This includes psychological withdrawal from interaction in the field, which can be fatal to research; oscillation or emotional labiality; complete immersion, which, for Davies is similar to “going native” and abandoning the observer role; mourning, which can result a separation from people and relationships in the field; and altered perception, in which through a shift in perception the researcher becomes closer to the emotional situation of the participants in research and can result in a closer relationship between the researcher and the subjects of research, but may set up a period of reverse dissonance when the researcher returns home (Davies 2010a). Compared to other sojourners, the researcher is always “on” (i.e., playing the role of being the participant observer and never able to escape, always having to be alert as an observer, when the natural human role is to relax into being a participant). The fieldworker is not just trying to live in a new setting, but also feels acutely the need to be accomplishing the task of research in this setting—a granting agency is paying for the work and expects a report, an adviser is asking for field notes, an employer is looking for the data, the researcher’s degree and career may be on the line, and the researcher is paralyzed with loneliness and anxiety. Rosalie Wax describes her entrance into a Japanese resettlement camp in 1943. After an arduous two days of travel, Wax arrived at the Gila camp outside Phoenix in heat that she found “incredible.” Alone, in the camp with little preparation for the field experience Wax finally found her room (“cell”) by process of elimination from a number of other rooms. She writes: It contained four dingy and dilapidated articles of furniture: an iron double bedstead, a dirty mattress (which took up more than half the room), a chest of drawers, and a tiny writing table—and it was hotter than the hinges of Hades. Since there was no one around to ask where I might get a chair, I sat down on the hot mattress, took a deep breath, and cried. I was too far gone to be consciously aware that I was isolated or to wonder why I had left a beautiful and comfortable university town to stick myself in this oven of a concentration camp. Like some lost two-year-old I only knew that I was miserable. After a while, I found the room at the end of the barracks that contained two toilets and a couple of wash basins. I washed my face and told myself that I would feel better the next day. I was wrong. (Wax 1971, 67–68)

Wax described the weeks of ostracism, despair, and perceived failure that followed her arrival at Gila. Few would speak to her, and when she did have conversations, she frequently insulted her potential informants. She was clearly not trusted; lonely beyond belief; ready to leave at any moment; and sure that the entire project was a failure. All the while, the supervisor of the project she was supposed to be carrying out was demanding data. She writes: “Every time I returned to my stifling room after a series of futile ‘interviews,’ I just sat down and cried. I took long walks in the desert in temperatures of 110°–120°, and cried” (Wax 1971, 72).

Coping with Culture Shock Wax eventually found a way to cope with the situation and to make some progress with her research while she became acculturated to life in the camp and gained the trust

14_107-Bernard.indb 282

6/9/14 10:03 AM

8: Participant Observation

283

of the residents. She moved from her original room to one even less comfortable in the ruder barracks to which the Japanese were assigned. She avoided the staff. She enlisted the help of another social scientist in the camp and found someone to introduce her to several people from whom she took Japanese lessons as a ruse to get to know them better. Finally, she devised a set of “red herring” studies. These were plausible concrete data studies, which, while they had little to do with her real research project, were innocuous enough to be nonthreatening to respondents. The red herring studies allowed her to meet a number of people in a role they could easily understand. And she says “meanwhile I was gradually pushed into the role of a willing learner . . . friendly respondents now began to instruct me in some of the less confidential aspects and attitudes of center life as well as the rudiments of Japanese etiquette” (Wax 1971, 77). Mead (1970a) also suggests the new fieldworker begin with more structured data collection to be able to show progress through the difficult first weeks. Even while describing the plight of the new and lone fieldworker, she also notes that a person alone in the field may, in fact, be incorporated more quickly into daily life, learning the language more quickly, perforce socializing with villagers. Perhaps the most important strategy for coping with culture shock is to realize that the experience of intense anxiety and hopelessness is temporary. As Oberg noted to the expatriate women of Rio so many years ago, “There is a great difference in knowing what is the cause of your disturbance and not knowing” (1954, 4). Following Oberg and more contemporary culture shock researchers (Ward et al. 2001), knowing what to expect and understanding that there are cultural differences is one of the key strategies for coping. “Return culture shock” is also a problem. Wax (1971) writes that when she was beginning work on Doing Fieldwork she asked a young colleague recently returned from fieldwork in Melanesia what was the most important thing to tell students who had never been to the field. He replied: “Tell them to keep a personal diary in addition to their field notes and tell them that returning to your own society can be as difficult as trying to enter a strange one” (Wax 1971, 172–73). Davies (2010a) discusses how shifts in perception that take place in immersive fieldwork, when incorporated into analysis, make the tacit explicit and have an impact on perceptions when coming home. He argues that the process of fieldwork and shifts in perception make coming home, no matter how far home is from the field, an additional process of adaptation to what has now become a strange place. Rabinow found his return to New York after fieldwork in Morocco presented him with a “maze of slightly blurred nuance, that feeling of barely grasped meaning which had been my constant companion in Morocco [that] overtook me once again. But now I was home” (1977, 148). Davies argues that the familiarization with a viable set of alternative cultural dispositions in immersion fieldwork fundamentally challenges conceptions of self and the social world that sponsored them. Reintegration into the home context rechallenges the new self. The ability to harness “reverse culture shock” in analysis can become a tool in a deeper, more subjective ethnography. CONCLUSION

Participant observation as a technique of fieldwork has been a constitutive element of ethnographic research since the beginning of the twentieth century and has been a

14_107-Bernard.indb 283

6/9/14 10:03 AM

284

Kathleen Musante

distinguishing characteristic for anthropology compared with other social sciences. It may be essential to the conduct of cultural anthropology. Although tied to functionalist theoretical approaches early in the century, the reliance on participant observation and the recording of chronologically oriented, descriptive field notes, which also include the incorporation of the ethnographer’s thoughts and reactions, laid the groundwork for much of the theoretical development autochthonous to anthropology. That the descriptions of the research enterprise provided by Malinowski in 1922, Bourgois in 1995, and Boellstorff in 2010 can appear so similar suggests that the method, while not atheoretical, is so closely tied to a relatively unchanging theoretical core in anthropology as to provide the basis for a wide range of theoretical development around that core. The movement into more reflexive ethnographic writing has resulted in a quantum increase in the number of accounts of the fieldwork experience presented by ethnographers from a number of different theoretical approaches. The result is a continuing demystification of the process of fieldwork and ethnographic writing. Making explicit the process of participant observation allows the reader to better understand the information presented by the ethnographer. Narrative ethnography (Tedlock 1991) and personal accounts of field experience also provide the opportunity for new researchers to begin to anticipate problems, identify alternative strategies, and begin to craft their personal approaches to participant observation early in the fieldwork experience. The approach to training in ethnography common a generation ago—in which each new ethnographer is told to go out and reinvent anthropology methodologically and sink or swim on their own ability to do so—constitutes the worst form of intellectual elitism. Fortunately, we have gone beyond this. Approaches that emphasize the “observation of participation” are quite useful and important, but I see these as complementary to the use of participant observation as a means of collecting verifiable, reliable data concerning human behavior. We are not “objective” measuring devices. We can, however, use participant observation in conjunction with other methods to serve anthropology as a scientific pursuit. That is, I see reflexivity as a beginning point rather than as an end to ethnography (see Aunger 2004). We need to be aware of who we are, understand our own perspectives, and understand and interpret our interactions with the people we study. Once we have done that, we can strive to determine whether there are regularities in human behavior. Participant observation raises many important ethical issues for ethnographers. These ethical issues include the problem of establishing “limits to participation”—should ethnographers engage in illegal behaviors, should they establish intimate relationships with people in the field, and should they take up the causes of the people we study? Among the ethical challenges to the method is finding the most workable mix between participation and observation. There is the problem of maintaining the anonymity of the people that we study, often for their own protection. There are no easy answers to these concerns; they require careful thought throughout the research process. NOTES

1. Lindeman’s participant–observer is closer to what we now generally refer to as a key informant. The role of the participant–observer is to provide insight into the thinking of a group and the individuals that make it up. The role of the participant–observer is distinct from the

14_107-Bernard.indb 284

6/9/14 10:03 AM

8: Participant Observation

2. 3.

4. 5. 6.

7.

285

role of the “outside observer,” who is the researcher. In work with Hader, he elaborates on the role of participant–observer, as compared with direct observation carried out by an outside observer, suggesting ways to identify and train participant observers. The Adlers did not reveal that they were researchers during this study. Covert research, although still advocated by some, would be now be considered unethical by most researchers. I have carried out fieldwork in an essentially “commuting” situation and as a resident of the community. For me, there is no doubt that greater understanding comes from living in the community. In Palace, gods and wizards are highly regarded members of the online Palace community. Almao—literally, German—is used by the Bakairi to refer all non-Brazilian outsiders. The film version of the Michael Frayn play Copenhagan, which examines the nature of observation through the interaction of Bohr and his former colleague, Werner Heisenberg, during World War II, is an excellent introduction to the discussion of objectivity. Oberg used the term “sojourner” to describe people, like researchers, who are in a different cultural context for a limited period of time and expect to return home at the end of a defined period. The term has been picked up in the more general literature to refer to such travelers as international students, businesspeople, diplomats, and some missionaries, as compared with short-term travelers such as tourists and permanent travelers such as immigrants and refugees (Ward et al. 2001).

REFERENCES

Adler, P. A. 1985. Wheeling and dealing: An ethnography of an upper-level drug dealing and smuggling community. New York: Columbia University Press. Adler, P. A., and P. Adler. 1987. Membership roles in field research. Newbury Park, CA: Sage. Agar, M. 1986. Speaking of ethnography. Beverly Hills, CA: Sage. Agar, M. 1996. The professional stranger: An informal introduction to ethnography, 2nd ed. San Diego: Academic Press. Allen, K. 1988. The hold life has: Coca and cultural identity in an Andean community. Washington, DC: Smithsonian Institution Press. Alter, J. S. 2000. Knowing Dil Das: Stories of a Himalayan hunter. Philadelphia: University of Pennsylvania Press. American Anthropological Association (AAA). 2012. Statement on ethics, principles of professional responsibility. Washington, DC: American Anthropological Association. http://www .aaanet.org/coe/Code_of_Ethics.pdf (accessed December 11, 2012). Appadurai, A. 1981. Gastro-politics in Hindu South Asia. American Ethnologist 8: 494–511. Arhem, K. 1994. Antropologins mening: En introduktion. In Den antropologiska erfarenhetem, ed. K. Arhem, 9–37. Stockholm: Carlsson Bokforlag. Atkinson, P., and M. Hammersley. 1994. Ethnography and participant observation. In Handbook of qualitative research, ed. N. K Denzin and Y. S. Lincoln, 248–61. Los Angeles: Sage. Aunger, R. 2003. Reflexive ethnographic science. Walnut Creek, CA: AltaMira. Avenarius, C. B. 2004. The structure of constraints: Social networks of immigrants from Taiwan. Cologne: Kölner Universitäts-Publikationsserver. Becker, H. 1970. Sociological work: Method and substance. Chicago: Aldine. Behar, R. 1993. Translated woman: Crossing the border with Esperanza’s story. Boston: Beacon Press. Behar, R. 1996. The vulnerable observer: Anthropology that breaks your heart. Boston: Beacon Press. Bentley, M. E., G. H. Pelto, W. L. Strauss et al. 1988. Rapid ethnographic assessment: Applications in a diarrhea management program. Social Science & Medicine 27: 107–16.

14_107-Bernard.indb 285

6/9/14 10:03 AM

286

Kathleen Musante

Bernard, H. R. 2011. Research methods in anthropology: Qualitative and quantitative approaches, 5th ed. Lanham, MD: AltaMira. Bock, P. K. 1970. Culture shock: A reader in modern cultural anthropology. New York: Knopf. Boellstorff, T. 2010. Coming of age in Second Life: An anthropologist explores the virtually human. Princeton, NJ: Princeton University Press. Bohr, N. 1963. Essays, 1958–1962, on atomic physics and human knowledge. Volume 3, The philosophical writings of Niels Bohr. New York: Interscience Publishers. Bourgois, P. 1995. In search of respect: Selling crack in El Barrio. Cambridge: Cambridge University Press. Bourgois, P. 1996. Confronting anthropology, education and inner-city apartheid. American Anthropologist 98: 249–65. Briggs, J. 1986. Kapluna daughter. In Women in the field: Anthropological experiences, 2nd ed., ed. P. Golde, 19–44. Berkeley: University of California Press. Brymer, R. A. 1991. The emergence and maintenance of a deviant sub-culture: The case of hunting/poaching sub-culture. Anthropologica 33:177–94. Brymer, R. A. 1998. Hanging out with the good ’ole boys, gangsters, and other disreputable characters. In Doing ethnography: Fieldwork settings, ed. S. Grills, 3–18. Thousand Oaks, CA: Sage. Cassell, J., and S. E. Jacobs. 1997. Handbook on ethical issues in anthropology. Special publication of the American Anthropological Association. http://www.ameranthassn.org/sp23.htm (accessed December 15, 2012). Clifford, J. 1990. Notes on (field)notes. In Fieldnotes: The makings of anthropology, ed. R. Sanjek, 47–70. Ithaca, NY: Cornell University Press. Clifford, J. 1997. Routes: Travel and translation in the late twentieth century. Cambridge, MA: Harvard University Press. Collings, P. 2009. Participant observation and phased assertion as research strategies in the Canadian Arctic. Field Methods 21: 133–53. Constable, N. 2003. Romance on a global stage: Pen pals, virtual ethnography, and “mail order” marriages. Berkeley: University of California Press. Coy, M. 1989. Introduction. In Apprenticeship: From theory to method and back again, ed. M. Coy, xi–xv. Albany: SUNY Press. Crapanzano, V. 2010. “At the heart of the discipline”: Critical reflections on fieldwork. In Emotions in the field, ed. J. Davies and D. Spencer, 55–78. Stanford, CA: Stanford University Press. D’Andrea, A. 2007. Global nomads: Techno and new age as transnational countercultures in Ibiza and Goa. New York: Routledge. Davies, J. 2010a. Disorientation, dissonance, and altered perceptions in the field. In Emotions in the field: The psychology and anthropology of fieldwork experience, ed. J. Davies and D. Spencer, 79–97. Stanford, CA: Stanford University Press. Davies, J. 2010b. Introduction: Emotions in the field. In Emotions in the field: The psychology and anthropology of fieldwork experience, ed. J. Davies and D. Spencer, 1–34. Stanford, CA: Stanford University Press. Davies, J., and D. Spencer. 2010. Emotions in the field: The psychology and anthropology of fieldwork experience. Stanford, CA: Stanford University Press. Denton, R. K. 1970. Living and working with the Semai. In Being an anthropologist, ed. G. D. Spindler, 85–112. New York: Holt, Rinehart, and Winston. Desjarlais, R. 1992. Body and emotion: The aesthetics of illness and healing in the Nepal Himalayas. Philadelphia: University of Pennsylvania Press. Desjarlais, R. R., and C. J. Throop. 2011. Phenomenological approaches in anthropology. Annual Review of Anthropology 40: 87–102.

14_107-Bernard.indb 286

6/9/14 10:03 AM

8: Participant Observation

287

DeWalt, K. M., and B. R. DeWalt. 2010. Participant observation: A guide for fieldworkers. Palo Alto, CA: AltaMira. Ditton, J. 1977. Part-time crime. London: Macmillan. DuBois, C. 1951. Culture shock. Presented at the Mideast Regional Meeting of the Institute of Internatioanl Education. Chicago, November 28. Dumont, J.-P. 1991. The headman and I: Ambiguity and ambivalence in the fieldworking experience. Long Grove, IL: Waveland. Emerson, R. M., R. I. Fretz, and L. L. Shaw. 2011. Writing ethnographic fieldnotes, 2nd ed. Chicago: University of Chicago Press. Erickson, F. 2011. A history of qualitiative inquiry in social and educational research. In Sage handbook of qualitative research, ed. N. K. Denzin and Y. S. Lincoln, 43–59. Los Angeles: Sage. Evans-Pritchard, E. E. 1973. Some reminiscences and reflections on fieldwork. Journal of the Anthropological Society of Oxford 4: 1–12. Finnström, S. 2008. Living with bad surroundings: War, history, and everyday moments in northern Uganda. Durham, NC: Duke University Press. Firth, R. 1985. Degrees of intellegibility. In Reason and morality: ASA Monographs #24, ed. J. Overing, 29–46. London: Routledge. Fluehr-Lobban, C., and R. Lobban. 1986. Families, gender and methodology in the Sudan. In Self, sex and gender in cross-cultural fieldwork, ed. T. L. Whitehead and M. E. Conaway, 188–93. Urbana: University of Illinois Press. Freeman, D. 1983. Margaret Mead in Samoa: The making and unmaking of an anthropological myth. Cambridge, MA: Harvard University Press. Frisch, O. R. 1967. Niels Bohr. Scientific American 216: 145–48. Geertz, C. 1973. The interpretation of cultures: Selected essays. New York: Basic Books. Gewertz, D., and F. Errington. 2010. Cheap meat: Flap food nations in the Pacific. Berkeley: University of California Press. Glaser, B. G., and A. Strauss. 1967. The discovery of grounded theory: Strategies for qualitative research. Chicago: Aldine. Good, K. 1991. Into the heart: One man’s pursuit of love and knowledge among the Yanomama. New York: Simon and Schuster. Green, J., ed. 1978. Zuni: Selected writings of Frank Hamilton Cushing. Lincoln: University of Nebraska Press. Green, J., S. W. Green, and F. H. Cushing. 1990. Cushing at Zuni: The correspondence and journals of Frank Hamilton Cushing 1879–1884. Alburqueque: University of New Mexico Press. Grills, S. 1998. An invitation to the field: Fieldwork and the pragmatist’s lesson. In Doing ethnography: Fieldwork settings, ed. S. Grills, 3–18. Thousand Oaks, CA: Sage. Guimaraes, M. J. L. 2005. Doing anthropology in cyberspace: Fieldwork boundaries and social environments. In Virtual methods: Issues in social research on the Internet, ed. C. Hine, 141–56. Oxford: Berg. Hage, G. 2010. Hating Israel in the field: On ethnography and political emotions. In Emotions in the field: The psychology and anhropology of fieldwork experience, ed. J. Davies and D. Spencer, 129–54. Stanford, CA: Stanford University Press. Hastrup, K. 2010. Emotional topographies: The sense of place in the far north. In Emotions in the field: The psychology and anthropology of fieldwork experience, ed. J. Davies and D. Spencer, 191–211. Stanford, CA: Stanford University Press. Hecht, T. 1998. At home in the street: Street children of northeast Brazil. Cambridge: Cambridge University Press.

14_107-Bernard.indb 287

6/9/14 10:03 AM

288

Kathleen Musante

Hewlett, B. L., and B. S. Hewlett. 2008. A biocultural approach to sex, love, and intimacy in central African foragers and farmers. In Intimacies: Love and sex across cultures, ed. W. R. Jankowiak, 39–64. New York: Columbia University Press. Hewlett, B. S., and B. L. Hewlett. 2010. Sex and searching for children among Aka foragers and Ngandu farmers of central Africa. African Study Monographs 31: 107–25. Hine, C. 2000. Virtual ethnography. London: Sage. Hine, C., ed. 2005. Virtual methods: Issues in social research on the Internet. Oxford: Berg Publishers. Hinsley, C. 1983. Ethnographic charisma and scientific routine: Cushing and Fewkes in the American Southwest, 1879–1893. In Observers observed: Essays in ethnographic fieldwork, ed. G. W. Stocking Jr., 53–69. Madison: University of Wisconsin Press. Holy, L. 1984. Theory methodology and the research process. In Ethnographic research: A guide to general conduct, ed. R. F. Ellen, 1–12. London: Academic Press. Horowitz, R. 1996. Getting in. In In the field: Readings on the field research experience, 2nd ed., ed. C. D. Smith and W. Kornblum, 41–50. Westport, CT: Praeger. Jackson, J. 1986. On trying to be an Amazon. In Self, sex and gender in cross-cultural fieldwork, ed. T. L. Whitehead and M. E. Conaway, 263–74. Urbana: University of Illinois Press. Jackson, J. 1990. I am a fieldnote: Fieldnotes as a symbol of professional identity. In Fieldnotes: The makings of anthropology, ed. R. Sanjek, 3–33. Ithaca, NY: Cornell University Press. Jackson, M. 2010. From anxiety to method in anthropological fieldwork: An appraisal of George Devereux’s enduring ideas. In Emotions in the field: The psychology and anthropology of fieldwork experience, ed. J. Davies and D. Spencer, 35–54. Stanford, CA: Stanford University Press. Johnson, J. C. 1982. Of strangers and outsiders: Ethnography and serendipity in coastal communities. Paper presented at the annual meeting of the American Anthropological Association, Washington, DC, December 1–4. Johnson, J. C. 1998. Research design and research strategies. In Handbook of methods in cultural anthropology, ed. H. R. Bernard, 131–70. Walnut Creek, CA: AltaMira. Johnson, J. C., C. B. Avenarius, and J. Weatherford. 2006. The active participant-observer: Applying social role analysis to participant observation. Field Methods 18: 111–34. Katz, J. 1988. Seductions to crime: Moral and sensual attractions in doing evil. New York: Basic Books. Kawulich, B. B. 2011. Gatekeeping: An ongoing adventure in research. Field Methods 23: 57–76. Kluckhohn, C. 1943. Bronislaw Malinowski 1884–1942. Journal of American Folklore 56: 208–19. Kluckhohn, F. R. 1940. The participant-observer technique in small communities. American Journal of Sociology 46: 331–43. Kornblum, W. 1996. Introduction. In In the field: Readings on the field research experience, 2nd ed., ed. C. D. Smith and W. Kornblum, 1–7. Westport, CT: Praeger. Kraidy, M. M. 1999. The global, the local, and the hybrid: A native ethnography of glocalization. Critical Studies in Mass Communication 16: 456–76. Kulick, D., and M. Willson, eds. 1995. Taboo: Sex, identity, and erotic subjectivity in anthropological fieldwork. London: Routledge. Leach, E. 1957. The epistemological background to Malinowski’s empiricism. In Man and culture: An evaluation of the work of Bronislaw Malinowski, ed. R. Firth, 119–37. New York: Harper Torchbooks. Lederman, R. 1986. The return of redwoman: Field work in Highland New Guinea. In Women in the field: Anthropological experiences, 2nd ed., ed. P. Golde, 359–87. Berkeley: University of California Press.

14_107-Bernard.indb 288

6/9/14 10:03 AM

8: Participant Observation

289

Lindeman, E. C. 1924. Social discovery: An introduction to the study of functional groups. New York: Republic Publishing Company. Lohman, J, D. 1937. Participant observation in community studies. American Sociological Review 2: 890–97. Lugosi, P. 2006. Between overt and covert research: Concealment and disclosure in an ethnographic study of commercial hospitality. Field Methods 12: 541–61. Luhrmann, T. 2010. What counts as data? In Emotions in the field: The psychology and anthropology of fieldwork experience, ed. J. Davies and D. Spencer, 212–38. Stanford, CA: Stanford University Press. Lutz, K. 1988. Unnatural emotions: Everyday sentiments in a Micronesian atoll, their challenges to Western theory. Chicago: University of Chicago Press. Maginn, P. J. 2007. Negotiating and securing access: Reflections from a study into urban regeneration and community participation in ethnically diverse neighborhoods in London, England. Field Methods 19: 425–40. Malinowski, B. 1961 [1922]. Argonauts of the Western Pacific. New York: Dutton. Malinowski, B. 1967. A diary in the strict sense of the term. New York: Harcourt, Brace and World, Inc. Malinowski, B. 1978 [1935]. Coral gardens and their magic. New York: Dover. Marcus, G. E. 1995. Ethnography in/of the world system. Annual Reviews in Anthropology 24: 95–117. Marcus, G. E. 1997. The uses of complicity in the changing mise-en-scène of anthropological fieldwork. Representations Issue 59: 85–108. Marcus, G. E. 1998. Ethnography through thick and thin. Princeton, NJ: Princeton University Press. Marcus, G. E., and M. J. Fischer. 1986. Anthropology as cultural critique: An experimental moment in the human sciences. Chicago: University of Chicago Press. McCall, G. G., and J. L. Simmons. 1969. Issues in participant observation: A text and reader. Reading, MA: Addison-Wesley. Mead, M. 1928. Coming of age in Samoa. New York: William Morrow. Mead, M. 1930. Social organization of Manu’a. Honolulu: Bernice P. Bishop Museum. Mead, M. 1969. Social organization of Manu’a, rev. ed. Honolulu: Bernice P. Bishop Museum. Mead, M. 1970a. The art and technology of field work. In Handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 246–65. New York: Columbia University Press. Mead, M. 1970b. Field work in the Pacific islands 1925–1967. In Women in the field: Anthropological experiences, ed. P. Golde, 323–63. Chicago: Aldine Press. Morse, J. M. 1991. Subjects, respondents, informants, and participants? Qualitative Health Research 1: 403–6. Muller, K. 2006. Zuni-made man: Frank Hamilton Cushing and the power of Zuni material culture. Presented at the annual meeting of the American Studies Association, Oakland, CA, October 12–15. Mullings, L. 2000. African-American women making themselves: Notes on the role of black feminist research. Souls: A Critical Journal of Black Politics, Culture and Society 2: 18–29. Mullings, L., and A. Wali. 2001. Stress and resilience: The social context of reproduction in central Harlem. New York: Kluwer Academic/Plenum Publishers. Mullings, L., A. Wali, D. McLean et al. 2001. Qualitative methodologies and community participation in examining reproductive experiences: The Harlem birth right project. Maternal and Child Health Journal 5: 85–93. Murphy, E., and R. Dingwall. 2007. Informed consent, anticipatory regulation and ethnographic practice. Social Science & Medicine 65: 2223–34.

14_107-Bernard.indb 289

6/9/14 10:03 AM

290

Kathleen Musante

Murphy, Y., and R. F. Murphy. 1974. Women of the forest. New York: Columbia University Press. Nader, L. 1986. From anguish to exultation. In Women in the field: Anthropological experiences, 2nd ed., ed. P. Golde, 97–116. Berkeley: University of California Press. Nash, J. 1976. Ethnology in a revolutionary setting. In Ethics and anthropology: Dilemmas in fieldwork, ed. M. A. Rynkiewich, M. A. James, and P. Spradley, 148–66. New York: John Wiley and Sons. Oberg, K. 1954. Culture shock. Presented at the Women’s Club of Rio de Janeiro, Brazil, August 3. Park, J., and A. Zeanah. 2005. An evaluation of voice recognition software for use in interviewbased research: A research note. Qualitative Research 5: 245–51. Pelto, P. J. 1970. Anthropological research: The structure of inquiry. New York: Harper and Row. Pelto, P. J., and G. H. Pelto. 1978. Anthropological research: The structure of inquiry, 2nd ed. New York: Harper and Row. Perusek, D. 2012. Cancer, culture, and individual experience: Public discourse and personal affliction. Ethos 40: 476–506. Picchi, D. S. 1992. Lessons in introductory anthropology from the Bakairi Indians. In The naked anthropologist: Tales from around the world, ed. P. R. DeVita, 144–55. Belmont, CA: Wadsworth Publishing. Punch, M. 1994. Politics and ethics in qualitative research. Thousand Oaks, CA: Sage. Rabinow, P. 1977. Reflections of fieldwork in Morocco. Berkeley: University of California Press. Rashid, S. F. 2007. Accessing married adolescent women: The realities of ethnographic research in an urban slum environment in Dhaka, Bangladesh. Field Methods 19: 369–83. Riemer, J. W. 1977. Varieties of opportunistic research. Urban Life and Culture 5: 467–78. Rynkiewich, M. A., and J. P. Spradley. 1976. Ethics and anthropology: Dilemmas in fieldwork. New York: John Wiley and Sons. Sanjek, R. 1990a. The secret life of fieldnotes. In Fieldnotes: The makings of anthropology, ed. R. Sanjek, 187–270. Ithaca, NY: Cornell University Press. Sanjek, R. 1990b. A vocabulary for fieldnotes. In Fieldnotes: The makings of anthropology, ed. R. Sanjek, 92–121. Ithaca, NY: Cornell University Press. Sanjek, R., ed. 1990c. Fieldnotes: The makings of anthropology. Ithaca, NY: Cornell University Press. Schensul, J., and M. D. LeCompte. 2013. Essential ethnographic methods: A mixed methods approach, 2nd ed. Walnut Creek, CA: AltaMira. Scheper-Hughes, N. 1992. Death without weeping. Berkeley: University of California Press. Scheper-Hughes, N. 1996. The primacy of the ethical: Propositions for a militant anthropology. Current Anthropology 36: 409–40. Scrimshaw, N. S., and G. R. Gleason, eds. 1992. RAP: Rapid assessment procedures. Boston: International Nutrition Foundation for Developing Countries (INFDC). Scrimshaw, S., and E. Hurtado. 1987. Rapid assessment procedures for nutrition and primary health care: Anthropological approaches to improving programme effectiveness. Tokyo: The United Nations University. Seligman, B. Z. 1951. Notes and queries on anthropology, 6th ed. London: Routledge & Kegan Paul. Shostak, M. 1983. Nisa: The life and words of a !Kung woman. New York: Vintage Books. Singleton, J. 1989. Japanese folkcraft pottery apprenticeship. In Apprenticeship: From theory to method and back again, ed. M. W. Coy, 13–30. Albany: SUNY Press. Slim, H., and J. Mitchell. 1992. The application of RAP and RRA techniques in emergency relief programmes. In RAP: Rapid assessment procedures, ed. N. S. Scrimshaw and G. R Gleason, 251–57. Boston: International Nutrition Foundation for Developing Countries. Spradley, J. P. 1980. Participant observation. New York: Holt, Rinehart and Winston.

14_107-Bernard.indb 290

6/9/14 10:03 AM

8: Participant Observation

291

Stack, C. 1996. Doing research in the flats. In In the field: Readings on the field research experience, 2nd ed., ed. C. D. Smith and W. Kornblum, 21–25. Westport, CT: Praeger. Sterk, C. 1996. Prostitution, drug use and AIDS. Westport, CT: Praeger. Stocking, G. W., Jr. 1983. The ethnographer’s magic: Fieldwork in British anthropology from Tylor to Malinowski. In Observers observed: Essays in ethnographic fieldwork, ed. G. W. Stocking, Jr., 70–120. Madison: University of Wisconsin Press. Strauss, A. L., and J. Corbin. 1990. Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage. Strauss, A. L., and J. Corbin. 1998. Basics of qualitative research: Techniques and procedures for developing grounded theory. Newbury Park, CA: Sage. Susser, I. 2009. AIDS, sex, and culture: Global politics and survival in southern Africa. Oxford: Wiley-Blackwell. Taylor, J. S. 2008. On recognition, caring, and dementia. Medical Anthropology Quarterly 22: 313–35. Tedlock, B. 1991. From participant observation to the observation of participation: The emergence of narrative ethnography. Journal of Anthropological Research 47: 69–94. Teman, E. 2010. Birthing a mother: The surrogate body and the pregnant self. Berkeley: University of California Press Thorne, B. 1980. “You still takin’ notes?” Fieldwork and problems of informed consent. Social Problems 27: 284–97. Throop, C. J. 2012. On the varieties of empathic experience: Tactility, mental opacity, and pain in Yap. Medical Anthropology Quarterly 26: 408–30. doi:10.1111/j.1548-1387.2012.01225.x. Tsing, A. L. 1993. In the realm of the diamond queen. Princeton, NJ: Princeton University Press. Tsing, A. L. 2005. Friction: An ethnography of global connection. Princeton, NJ: Princeton University Press. Van Maanen, J. 1988. Tales of the field: On writing ethnography. Chicago: University of Chicago Press. van Willigen, J. 1989. Gettin’ some age on me: Social organization of older people in a rural American community. Lexington: University Press of Kentucky. van Willigen, J., and B. R. DeWalt. 1985. Training manual in policy ethnography. Washington, DC: American Anthropological Association. Villa Rojas, A. 1979. Fieldwork in the Mayan region of Mexico. In Long-term field research in social anthropology, ed. G. Foster, T. Scudder, E. Colson, and R. Kemper, 45–64. New York: Academic Press. Wanat, C. L. 2008. Getting past the gatekeepers: Differences between access and cooperation in public school research. Field Methods 20: 191–208. Ward, C., S. Bochner, and A. Furnham. 2001. The psychology of culture shock. Philadelphia: Routledge. Warren, C. A. B. 1988. Gender issues in field research. Newbury Park, CA: Sage. Wax, R. H. 1971. Doing fieldwork. Chicago: University of Chicago Press. Weatherford, J. M. 1986. Porn row. New York: Arbor House. Webb, B. P. 1926. My apprenticeship. London: Longmans, Green and Co. Webb, S., and B. P. Webb. 1902. Problems of modern industry. London: Longmans, Green and Co. Webb, S., and B. P. Webb. 1932. Methods of social study. New York: Longmans, Green and Co. Weiner, A. B. 1988. The Trobrianders of Papua New Guinea. New York: Holt, Rinehart and Winston. Whyte, W. F. 1996. Doing research in Cornerville. In In the field: Readings on the field research experience, 2nd ed, ed. C. D. Smith and W. Kornblum, 73–85. Westport, CT: Praeger.

14_107-Bernard.indb 291

6/9/14 10:03 AM

292

Kathleen Musante

Whyte, W. F., and K. K. Whyte. 1984. Learning from the field: A guide from experience. Beverly Hills, CA: Sage. Winchatz, M. R. 2006. Fieldworker or foreigner?: Ethnographic interviewing in nonnative languages. Field Methods 18: 83–97. Winchatz, M. R. 2010. Participant observation and the nonnative ethnographer: Implications of positioning on discourse-centered fieldwork. Field Methods 22: 340–56. Winegar, J. 2012. The privilege of revolution: Gender, class, space, and affect in Egypt. American Ethnologist 39: 67–70. Zahle, J. 2012. Practical knowledge and participant observation. Inquiry 55: 50–65.

14_107-Bernard.indb 292

6/9/14 10:03 AM

CHAPTER 9

Behavioral Observation Raymond Hames and Michael Paolisso

The goal of direct observations of behavior is to produce quantitative descriptions of behavior in a variety of natural settings. This approach is relatively new and uncommon in anthropology compared to behavioral biology, where researchers cannot rely on informant accounts of behavior. As noted by Johnson and Sackett (1998), what people actually do on a daily basis and who they interact with are not accurately reported in standard ethnographic literature or textbooks, and this is especially true of the activities of women and children. To a large extent, the development of behavior observations has been initiated and developed by researchers who are ecologically and economically oriented. This is not surprising, given the importance of quantitative behavioral data (e.g., hours worked in various tasks) required for adequate descriptions and tests of hypotheses in those fields. Recently, the scope of behavioral research has broadened to examine non-economic topics such as child and caretaker interactions, doctor–patient dynamics, and visiting patterns. Some literature exists on techniques of direct behavior observation. Early synthetic reviews are found in naturalistic and laboratory studies of animal behavior or ethology such as a classic paper by Altmann (1974) and Martin and Bateson’s (1993) textbook that focus on psychology and ethology. In anthropology, methodological summaries and ethnographic uses of behavior observations are found in Johnson and Sackett (1998), Borgerhoff Mulder and Caro (1985), Hames (1992), and Paolisso and Hames (2010). Strictly speaking, direct observations are made by a researcher on a subject or subjects. However, in some instances in which the observer and subject are the same, indirect observations are important tools that can produce reliable quantitative data on naturalistic behaviors (Paolisso and Hames 2010). These methods include forms of traditional time diaries and the day reconstruction method (DRM) (Kahneman et al. 2004). The major challenges to these indirect approaches are inaccurate recordings based on faulty or biased recall. Recent methodological advances in real-time data entry using experience sampling methods and ecological momentary assessment (ESM/ EMA), combined with smartphones and data entry software, may revolutionize indirect behavior observations (Trull and Ebner-Priemer 2009). Direct behavior observation methods are used to collect quantitative data on a variety of behaviors to better describe them in terms of frequency, duration, or periodicity and to test hypotheses that make predictions about behavior under a variety of conditions. Behavioral observations are not designed to be substitutes for standard ethnographic techniques such as participant observation, interviews, surveys, or 293

14_107-Bernard.indb 293

6/9/14 10:03 AM

294

Raymond Hames and Michael Paolisso

experiments. Furthermore, direct and systematic observation can be time consuming. Nevertheless, a number of advantages to the method make it useful in a variety of situations. Some of these advantages have little to do with the utility of collecting behavior observations per se. For example, in community-based research, direct observation requires an ethnographer to get out in the community on a regular basis and become familiar with all households and all segments of the community, including children and the elderly. By following direct observation protocols, the ethnographer gains a greater understanding of the community geography, social patterns, and daily movements by the people who live there. Direct observation also enhances rapport by giving households individualized attention through regular visits and gives an ethnographer the opportunity to discover rare or interesting practices that otherwise would have gone undetected. The more direct advantages revolve around gaining accurate descriptions of behavior. In traditional participant observational research, one quickly begins to form beliefs on who does what, how often, and under what conditions, sometimes leading to a confirmation bias. That is, once a belief is established, exposures to counterexamples fail to extinguish it and rare confirmations serve as reinforcements. This implicit knowledge often strongly colors qualitative ethnographic reporting. From our own personal experiences, impressions gained through participant observations of the division of labor, association patterns, and child care responsibilities often turned out to be incorrect after we analyzed our time allocation data—sometimes wildly so. So, even if behavior observation is not a major research protocol, data gathered through systematic observation are likely to enhance the general quality of ethnographic reporting. A powerful case for using time allocation in comparative research is made by Johnson and Behrens (1989) on the topic of paternal care of children. Johnson and Behrens show that categories developed by Human Relations Area Files (HRAF) researchers on normative patterns of behavior are not consistent with observed patterns of behavior in time allocation studies taken from a robust sample of societies. They further argue that cross-cultural analysis of time allocation data is a more powerful way to test comparative theory. The example they give is of HRAF codes that rate fathers as dedicated or indifferent caregivers. High ratings seem to be based on ethnographic descriptions of fathers who take delight in playing with and caring for their children. However, quantitative data show that in such cultures, fathers spend no more than five to seven minutes per day interacting with infants and toddlers, which suggest that normative generalizations poorly represent paternal care. A good example of this change in the comparative approach is the recent spate of time allocation research on the grandmother hypothesis, which is the idea that a grandmother’s alloparental efforts and economic support of her daughter is adaptively crucial (Scelza 2009) and alloparental care by other close kin is common (Kramer 2010). BEHAVIOR OBSERVATIONS IN RELATION TO OTHER FORMS OF DATA COLLECTION

The uses of behavior observations (direct and indirect) depend on the kinds of questions or hypotheses posed. Johnson and Sackett (1998) note that behavior observation

14_107-Bernard.indb 294

6/9/14 10:03 AM

9: Behavioral Observation

295

is an expensive research method and, following Whiting and Whiting (1973), they recommend that it only be used when other methods of gathering behavioral data cannot yield the precision necessary to answer research questions. We think the Whitings’ view is extreme because it raises many interesting questions that cannot be answered without the quantitative data afforded by behavior observations. Interviewing is simply not up to the task. This is starkly revealed in Vitzthum’s (1994) comparison of the accuracy of recall versus direct observation in nursing for an Andean people. We need more of these comparative studies to help us understand the circumstances under which recall is an adequate method. A great strength of behavior observations is that you collect data on a large number of individuals; this allows you to examine how, for example, age and sex may account for variation in behavior. More to the point, use of the method depends on whether the research question requires behavior observations. As mentioned, economic and ecological research in anthropology have historically emphasized use of quantitative data on labor inputs (Hames 2010), and economic researchers made some of the earliest contributions to behavior observations (e.g., Richards 1939). Investigations of time allocation and division of labor in relation to cultural evolution (Minge-Klevana 1980), changes in individual productive activities across the lifespan (Bock 2002), and recent historic shifts in the division of household labor (Bianchi et al. 2000) have been powerfully illuminated by behavior observations. It is hard to imagine accurately answering these questions without recourse to behavior observations. Another area that emphasizes behavior observations is child care and alloparenting and human development (see, e.g., the collection of chapters in Hewlett and Lamb 2005) and life history theory (Kramer 2010). Finally, certain theoretical approaches such as behavioral ecology frequently demand behavior observations because the models employed require quantitative measures of somatic, reproductive, and nepotistic efforts that are often operationalized as behaviors that have costs in terms of time, energy, or risk gained through behavior observations (Winterhalder and Smith 2000). TYPOLOGY BEHAVIOR OBSERVATION TECHNIQUES: SAMPLING AND RECORDING RULES

Any researcher seriously considering collecting behavior observations should fully understand the different kind of data produced, the costs and benefits of each, methodological challenges, and, most importantly, which technique yields the best kind of data, given the hypotheses to be tested or the descriptions required. There are a variety of different ways in which behavior observations can be accomplished. In the simplified figure below (Figure 9.1), basic types of behavior observations are a combination of a recording rule and a sampling rule. The sampling rule determines whether a group scan or an individual (or focal follow) will be observed. The recording rule specifies how the observations will be measured. The model represented in Figure 9.1 is derived from studies employing direct observations. Studies using indirect observations or self-reports create data that can be incorporated into this scheme. For example, the day reconstruction method is a kind of continuous focal observation, while ecological momentary assessment is a kind of instantaneous focal observation. Both are discussed below.

14_107-Bernard.indb 295

6/9/14 10:03 AM

296

Raymond Hames and Michael Paolisso

Figure 9.1. Basic observational techniques.

SAMPLING RULES

While observing behavior, one selects between sampling a single individual (a focal person or a focal follow) or scan sampling, which is a sample of a group of individuals. Focal sampling is commonly used in conjunction with continuous recording. In fact, it is very difficult to use continuous recording for more than one individual at a time. However, scan observations are possible if they are limited to closely interacting dyads such as a parent and offspring (Fouts and Lamb 2005; Ivey 2000; Ivey et al. 2005) or spatially delimited groups, such as a group involved in a ritual, a shouting match (Flinn 1988a, 1988b), or cooperative agricultural labor (Hames 1987). A solution to the problem of attempting to record more than several individuals on a continuous basis is to use video recording, followed by a review of the video to carefully extract data for analysis, as in Takada’s (2005) work on mother infant interactions among !Xun foragers (see also Peregrine et al. 1993). An alternative strategy of observing a group would be to code the behavior very generally. For example, if a group of people were gathered for a ritual performance, one could record some of them as watching and others performing instead of more precisely coding behavior as watching while talking, waiting to perform while watching, or performing a specific ritual act. Focal sampling in conjunction with instantaneous recording is desirable if a researcher is interested in a detailed characterization of a behavior or a behavior complex. In such a situation, one might make an instantaneous observation every three to five minutes on the subject. This is the approach taken by Bird and Bliege Bird (2005) to collect details of hunting behavior in terms of measuring travel, search, rest, successful and unsuccessful pursuits, tools used, processing, and return. There is some inconsistency in the literature with the terms scan sampling and spot checks by ethnographers working in a traditional village field setting. In the spot check procedure pioneered by Johnson (1975), a researcher visits houses at random hours of the day and in random order and records the behavior of those present in an attempt to sample all village members. In some instances, subjects were engaged in solo or noninteractive behaviors, but in other instances they engaged in interactive behaviors such as conversation, grooming, or child care. If non-household members are observed, their behavior is also recorded. These spot checks are really scan samples of all those

14_107-Bernard.indb 296

6/9/14 10:03 AM

9: Behavioral Observation

297

present at a household at a particular instant. Most sampling rules are dependent on whether one preselects an individual or group to observe. An alternative approach is called behavior sampling. Researchers interested in a specific set of behaviors such as aggression and reconciliation record such behaviors in a group setting whenever they occur without preselecting an individual. This approach is used to measure rare but important behaviors and may be used in conjunction with scan sampling (see Martin and Bateson 1993, 51, for further details). RECORDING RULES

There are two types of recording rules: continuous (state) versus instantaneous (event). The chief difference between continuous and instantaneous recordings is that in continuous recording, one sequentially records the behavior of an individual for a predetermined length of time. Continuous recording is a moment-to-moment shorthand description of a behavior stream. For example, one might visit a house and watch a resident prepare a meal and sequentially record all the steps from beginning to end. In contrast, in event or instantaneous recording, one records the behavior of an individual at the instant they are first observed at predetermined intervals. For example, when entering a house you might decide to record the behavior of all individuals in the house the moment you encounter them, wait five minutes, and then again record what all are doing. Alternatively, in a focal follow you might decide to instantaneously record the behavior of only one individual once every five minutes. The easiest way to conceptualize the differences between the two is that continuous recording is like a video, whereas instantaneous recording is like a photo. Each technique has its strengths and weakness in terms of detail, recording accuracy, costliness, and obtrusiveness. Thus, the kind of data gained through continuous versus instantaneous recording are fundamentally different. Instantaneous recording leads to the collection of “dimensionless” data. Each datum is a classification of the kind of behavior observed and perhaps the intensity of the behavior. It is dimensionless because no information is collected on duration, frequency, or other useful dimensions of behavior. Continuous recording is used most commonly by psychologists who focus on very well-defined and restricted patterns of behavior, often in a laboratory setting. A coarser form of continuous recording is also used in time diary (Robinson and Godbey 1999) and DRM research. In DRM protocols, an individual is asked to reconstruct a serial ordering of the major episodes of the day, then they are asked to note the start and end times and answer a variety of questions (e.g., they select from a list of 16 activities) about what went on in each episode (Kahneman et al. 2004). Instantaneous recording, as its name implies, is the instantaneous recording of a behavior, a snapshot, if you will. While it lacks the depth of complexity of continuous recording, it is exceptionally versatile and designed to quickly gather behavior observations on a large number of individuals in a variety of contexts. We describe these two approaches in greater detail.

Continuous Recording Continuous recording is a moment-to-moment recording of a stream of behavior over a short time scale, typically no more than a half-hour. In practice, continuous

14_107-Bernard.indb 297

6/9/14 10:03 AM

298

Raymond Hames and Michael Paolisso

recording is limited to the fine-grained analysis of specific kinds of behaviors or behavior complexes such as parent–child interactions, patterns of attachment (Roggman et al. 1994), communicative and social interactions (Pettit and Bates 1989), and classroom behavior (Pianta et al. 2007). The kind of data gained through continuous versus instantaneous recording are remarkably different. Using continuous recording, one can collect nearly anything of interest in a behavioral stream such as: • Frequency: how frequently a behavior occurs within a particular time period • Intensity: how energetically or forcefully the behavior is acted. There are numerous field and laboratory experimental studies (e.g., Durnin and Passmore 1967; Hipsley and Kirk 1966; Montgomery and Johnson 1976) that can be adapted to one’s field data to create good estimates of caloric expenditures of effort. • Latency: the period of time prior to the onset of a behavior • Duration: how long a behavior lasts • Sequence: the ordering of behaviors through time or in relation to external contexts

Despite the richness of the data collected through continuous recording, there are a number of limitations to this technique. The first is that is extremely intrusive because the observer is minutely recording the behavior of a subject. Some subjects may find such close attention unnerving and distracting. It can affect how behaviors are initiated and their duration compared to how the behavior unfolds if the subject is unobserved. For example, a subject may alter the way he or she responds to a fussy infant knowing the observer may not approve or appreciate a slow or irritated response, which would occur if the subject were unobserved. Coding the behavioral stream can be difficult. Knowing when a behavior begins or ends or whether it is really a subcomponent of an ongoing behavior presents problems. Obviously, the coding scheme needs to be derived from numerous practice sessions so that coding issues can be worked out with clear operational definitions of all behaviors and transitions. Most research using continuous recording such as that in developmental psychology, medicine, and consumer behavior is restricted to laboratories or in highly structured setting such as classrooms (Pianta et al. 2009) or medical examination rooms (Stange et al. 1998). In such settings, there is reduced intrusiveness and researchers invariably have a highly structured set of codes to classify behavior. Video recordings are often employed to carefully analyze behavior (Peregrine et al. 1993). Finally, a major problem with continuous recording is that the number of different individuals representing various age and gender classes tends to be limited because of the time required to use this method, making generalization across a population difficult. As noted below, being able to include a well-sampled cross-section of a community is a major strength of instantaneous recording.

Instantaneous Recording Instantaneous recording leads to the collection of “dimensionless” data. Each datum is a classification of the kind of behavior observed and perhaps the intensity of the behavior. It is dimensionless because no information is collected on duration, frequency, or other useful dimensions of behavior. Ethnographers working in small traditional agricultural villages or among hunter–gatherers have been exceptionally active in using

14_107-Bernard.indb 298

6/9/14 10:03 AM

9: Behavioral Observation

299

instantaneous recording (see http://www.yale.edu/hraf/publications_body_complete publist.htm#Time%20Allocation%20Series for a list of computer databases from this research). Typically, researchers visit households at random hours and immediately record the behavior of those present. If one or more individuals are absent, the ethnographer asks a household member where the missing individuals are and what they are doing. Procedures for randomizing visits and the order in which household are visited and other adjustments to produce candid recordings are detailed in Johnson (1975), Borgerhoff Mulder and Caro (1985), Betzig and Turke (1985), and Hames (1992). One-zero recording (sometimes known as behavior sampling) is another kind of instantaneous recording used infrequently by ethnographers but extensively in animal behavior. Good examples of ethnographic use are found in classroom settings (Smith 1985) and among Hadza hunter–gatherers (Marlowe 2004) on the frequency and type of paternal care of infants and children. The basic procedure is to count whether or not (hence one-zero) a behavior occurs within a limited time frame. For example, one might plan a 10-minute observation period and decide to check whether the behavior occurred at 30-second intervals. If the behavior was expressed within a 30-second interval, it is recorded. If the behavior of interest is not expressed, a zero is recorded. So long as the sampling intervals are short (e.g., sampling every minute versus every five minutes), this method produces reasonably accurate measures (Martin and Bateson 1993, 55). It is identical to an instantaneous recording rule because duration measurements are not made. Furthermore, who is sampled is determined by who expresses the behavior. A major advantage of this method is that it can be used in a group setting such as a classroom or social gathering of more than a dozen individuals. Finally, behavior sampling is useful for recording behaviors that are rare but significant (e.g., disputes). One might assume that since instantaneous recording only produces counts of behavior, such a technique might be very limiting. However, counts can be legitimately transformed into real-time measures under certain conditions and with certain reasonable assumptions. For example, if one samples behavior during waking hours, say a 14hour day, and one knows that 15% of observations were in food preparation activities, then one could reasonably conclude that 2.1 hours per day were spent in this activity. (Of course, this assumes that this behavior only occurs during the daylight-sampling period; see Scaglion [1986] on nighttime sampling).

Direct and Indirect Observations: A Role for Self-Reports We can distinguish between two types of behavior observations. In the classic approach developed in ethology, behavioral ecology, and allied field studies such as primatology, the observer and the subject are not the same individual. This approach is used in traditional ethnographic research in a small community setting (Johnson and Sackett 1998). But when humans are the focus of research, they can act as observer and subject by recording their own behavior. Traditional time diaries (Paolisso and Hames 2010) fit this definition and they have a variety of implementations. Given the difficulties of using direct observation in modern settings, we argue that time diary or self-report techniques should be more fully explored and implemented by anthropologists. Significant methodological innovations have been

14_107-Bernard.indb 299

6/9/14 10:03 AM

300

Raymond Hames and Michael Paolisso

made using smartphones and other electronic devices. We believe that these subjectdependent methods are extremely important for researchers working in modern settings where observer-dependent methods may be time consuming (reducing sample size), intrusive (inhibiting natural behavior), or unwelcome. In traditional time diary research, study participants provide a description of the day’s activities along with an assignment of the approximate starting and stopping times for each activity, recorded either in free format or in fixed intervals (Stinson1999). Data from time diary studies have been used to advance a wide range of social science research, including trends and gender differentials in housework (Bianchi et al. 2000), parental time with children (Sandberg and Hofferth 2001; Sayer et al. 2004), and leisure activities (Jacobs and Gerson 2004; Robinson and Godbey 1999; Schor 1991). Other time diary studies have investigated trends in TV viewing, Internet use, civic involvement (Putnam 2000; Sayer 2001), and religious participation (Presser and Stinson 1998). In time diaries, the most commonly used recall period is the previous 24 hours. More distant recall periods are also possible, such as last week or month, although they are used much less often than the 24-hour recall for good reason. The general rule is that the more distant the recall period, the more general and inaccurate will be the recall. A nice demonstration of this phenomenon is made by Wutich (2009) in her comparison of time diary, prompted recall, and free recall methods on water use in a Bolivian squatter community. Today, most studies use an ordinal listing (earliest to latest) of activities rather than a clock. There are two formats for these listing. First, beginning at a specified time (e.g., 4:00 a.m. or “the time you woke up”), the respondent is asked to list the activities she or he engaged in for specified increments of time during the day, for example, every 15 minutes. Alternatively, past activities can be elicited using a less-structured ordinal listing. Rather than prompting behavioral recall for small time intervals, respondents are asked to list the activities they undertook in the order they completed them. In this approach, the beginning and ending times are dictated by the respondents’ reported length of time they spent in each behavior. There is less respondent burden in this approach, although the interviewer has less control over the recall process. The documentation of behavior for specified periods of time produces fine-grained data, if interviewers and respondents can manage the cognitive burden of recalling behavior in such small segments. Piloting is critical to determine the optimal time interval, which should be the smallest possible one that guides respondents through the day’s activities without creating mental fatigue and loss of recall accuracy. Increased availability of the Internet, wireless communication, and inexpensive mobile computers and smartphones have helped usher in a number of new innovations in time diary research using an instantaneous recording rule. For example, experience sampling method (ESM) prompts study participants, with a pager beep or cell phone call, to ask them to record where they are, what they are doing, and even how they are feeling (see Csikszentmihalyi and Larson, 1987). ESM, in turn, has been modified to produce the day reconstruction method (DRM) that collects 24-hour recall data; study participants then review and discuss each recorded activity in terms of the feelings they experienced. This approach is a kind of generalized continuous recording, whereas ESM is instantaneous.

14_107-Bernard.indb 300

6/9/14 10:03 AM

9: Behavioral Observation

301

In this hybrid approach, DRM is intended to reproduce the information that would be collected by asking about experiences and feelings in real time, thus reducing respondent burden and disruption of normal activity (Kahneman et al. 2004). On the issue of respondent burden, we are unsure that DRM is less burdensome than ESM because DRM takes 45–70 minutes to complete the form (Kahneman et al. 2004, 1777), while ESM reports take one–three minutes to complete (Stone and Shiffman 2007, 240). However, the burden in DRM may be less given that the subject controls when he or she must respond. The obvious downside of DRM is that it relies on informant recall, although the way the protocol is structured may reduce recall error. Also closely related to ESM is a broader approach known as ecological momentary assessments (EMA), an approach widely used in behavioral medicine (see Trull and Ebner-Priemer [2009, 457] for a comparison of the two approaches). ESM researchers seek to collect information about individuals’ behavioral states at specific moments during their daily lives in their natural environments (see Shiffman et al. 2008; Stone and Shiffman 1994). The approaches for EMA data collection use either event-based recordings to capture behavior and context for specific events/activities or time-based prompted assessments (usually randomly scheduled) to capture subjects’ experiences during those events (Shiffman 2009). Recent implementations of EMA include the collection of physiological states using a variety of biosensors integrated with smartphones (Gaggioli et al. In press). Finally, in anthropology DeCaro et al. (2012) employ what they call the daily life architecture that combines experience-sampling methods with a structured diary approach. Several times a day, in response to beeps or cell phone calls, participants reconstruct their daily experiences, adding to information they have recorded, on a PDA or notebook, earlier in the day. The major problem with some early time diary approaches is that long-term memory is less accurate than short-term memory. Johnson and Sackett (1998) use an example of a study of university faculty who were asked to estimate the number of hours they worked per week. We agree that this sort of question will not yield reliable information for the reasons given by Johnson and Sackett. However, new implementations of self-reports where individuals are instructed to record their behavior using a restricted set of choices after being alerted by a smartphone or timer on a PDA avoids the memory problem by providing a near real-time recording of behavior. A number of recent studies using PDAs and cell phones convince us that subject responses are likely to produce accurate observations (Shiffman et al. 2008; Stone et al. 2007). To a significant degree, accuracy is dependent on the study participant’s willingness to implement the study protocol. If there is strong compliance, reinforced by an understanding by the participant of the goals and objectives of the study, this new generation of recall studies, assisted by technology, may provide a cost-effective and less-intrusive approach to collecting information on behaviors and associated cognitive and emotional states. We believe that such approaches closely mirror instantaneous observations employed in traditional observer–subject approaches and avoid many of the pitfalls seen in traditional time diary approaches such as recall problems and imprecise descriptions of behavior explored in more detail below.

14_107-Bernard.indb 301

6/9/14 10:03 AM

302

Raymond Hames and Michael Paolisso

METHODOLOGICAL ISSUES

A number of methodological issues require further attention in direct and indirect behavior observations to improve reliability and validity; we highlight two that revolve around reactivity and accuracy.

Observer Presence and Subject Reactivity Independently of how it is implemented, direct observation is intrusive and, as mentioned above, it poses difficulties in an urban setting where privacy is valued or expected. This is because an observer could enter a house at unpredictable hours to make observations to create a random set of observations. How much it will affect what is observed will depend on a variety of factors, such as the culture’s rules about what should be hidden or freely seen in everyday life, the rapport the observer establishes with study participants, and whether instantaneous or continuous recording is used. Analysis by Harvey et al. (2009) shows that subject reactivity in observational approaches does exist but it does not bias study results and they suggest ways to minimize the effect. Instantaneous observations in a traditional ethnographic setting where the researcher is more or less a member of an intimate community does not pose the problems it would in modern setting where the observer is not a community member and is not known outside his or her role as detached scientific observer. A way to get around the problem of instantaneous scans in an urban setting might be “block sampling,” as Behrens (1981) did in a Shipibo village of widely scattered households. In this approach, a researcher would enter a home and record an observation of all present household members every five to 10 minutes during a two- or fourhour block of time (see also Gurven and Kaplan 2006). In contrast, time diaries and other forms of indirect observations are much less intrusive. Typical time diary and DRM protocols encourage subjects to serially piece together their reports in the evening at a time of their choosing, making it minimally intrusive. So-called beeper studies, such as those used in EMA research, are probably in the middle ground—less intrusive than direct observation but more so than time diaries and DRM. Typically, subjects must quickly respond to unpredictable calls, pages, or other cues. Before the widespread use of cell phones this might have been intrusive, but with cell phone use being the norm it is probably much less intrusive than in the past. Informant Accuracy The major weakness of self-report of behavior is accuracy. Inaccurate reports can result from a variety of causes such as cultural and reputational factors, biased recall, poor informant training, culturally sensitive behaviors, and lack of informant commitment to the project. These problems extend to any research using surveys or interviews where informants must generate responses to queries. There is a large literature addressing such problems (e.g., Bernard et al. 1984), and researchers are advised to learn how to avoid pitfalls and enhance the accuracy and reliability of informant responses. Shiffman et al. (2008) provide a comparison of recall versus EMA in informant accuracy. In some cases, the two methods are comparable; in other cases, they are not. The problem is that the kind of data collected through EMA research focuses on internal states of interest to behavioral medical researchers such as perceptions of pain, levels of

14_107-Bernard.indb 302

6/9/14 10:03 AM

9: Behavioral Observation

303

anxiety, tiredness, and other medically relevant psychological and physiological measures. Whether or not one is more or less accurate than another for everyday behaviors, such as shopping, television viewing, or social interaction is unclear. Aside from Vitzthum (1994) mentioned earlier, few studies have attempted to compare direct observations to informant recall, although there are examples from medical research (Stange et al. 1988) and factory ergonomic research (BarrieraViruet et al. 2006). In these studies, a third party, a nurse, records the interaction between a doctor and patient. After the interaction, the patient and doctor fill out a questionnaire asking them about the procedures performed, questions asked, responses given, and so on. The interview information is then compared to the nurse’s notes, referred to as the “gold standard,” to determine how accurately what is recalled by the doctor and the patient matches the nurse’s observations. More studies comparing different methods are merited.

Coding How to classify behavior, or coding, presents a variety of challenges that include standardization for comparisons, how to characterize behavior in terms of its function or structure, standard codes, simultaneous behaviors, and others discussed below. That said, one should categorize behavior in ways that will allow the descriptive and theoretical goals of one’s research to be satisfied. Considerable research has been done on behavior observations leading to a multiplicity of standardized coding schemes. When possible, it is always a good idea to make one’s coding scheme compatible with others’ to permit comparisons and meta-analyses. STRUCTURAL VERSUS FUNCTIONAL CODES

There are two radically different approaches to coding. In one instance, a code could be a long-hand description of the various bodily movements that compose a behavior. Such codes are referred to as structural (Hames 1992) or physical descriptions (Borgerhoff Mulder and Caro 1985). For example, one could describe a person standing near a counter with a potato in one hand and a peeler in the other while the peeler moves across the potato and removing peels. That is a structural description because it highlights the physical movement that composes the behavior. Such an approach often requires a multivariate description such as stance, limb movement, and tool use. In short, it is similar to a long-hand description. Alternatively, one could use a functional code and call the behavior “peeling potatoes,” which describes the purpose or consequence of the behavior. Because of their simplicity and intuitiveness, functional codes are used most commonly. However, in certain instances, structural codes are necessary. In studies of infant care (Borgerhoff Mulder and Milton 1985), for example, the structural components of parent–child interactions often require very detailed descriptions of parental movements, including sensing an infant’s distress and handling of the child, and the child’s response. Furthermore, the function of some kinds of behavior must initially at least be described structurally because we may not know the significance of their function. Betzig and Turke (1985) distinguish between observed versus inferred behavior in instantaneous recording. For example, if a person is resting with hoe in hand while

14_107-Bernard.indb 303

6/9/14 10:03 AM

304

Raymond Hames and Michael Paolisso

others are working in the garden, the observed behavior would be resting but the inferred behavior is hoeing. It is inferred because the individual had gone there to hoe, had done some hoeing, but was resting at the moment observed. Is it justifiable to classify the behavior as hoeing or weeding because rest breaks are part of the process of this complex behavior? Betzig and Turke recommend that one take a two-tiered approach and code for observed and inferred behavior. STANDARD CODES

In the sociological study of Western societies, many use coding schemes based on the structure developed by Szalai (1972). These codes are arranged into mutually exclusive behavior groups that cover all aspects of human behavior. They commonly include personal care, employment, education, domestic work, child care, purchasing goods and services, voluntary work, social and community activities, recreation and leisure, and travel. However, we believe that a hierarchical scheme is much superior. The ATUS (American Time Use Survey) coding system (http://www.bls.gov/tus/ lexicons.htm) of the U.S. Bureau of Labor Statistics (http://www.bls.gov/tus/) uses a hierarchical structure, classifying reported activities into 17 major categories, with two additional levels of detail in each category. ATUS coders assign a six-digit classification code to each diary activity (rather than the three-digit code commonly used in some time-use surveys). The first two digits represent the major activity categories; the next two digits represent the second-tier level of detail; the final two represent the third—the most detailed level of activity (Shelley 2005), resulting in 400 detailed activities. The website http://www.bls.gov/tus/lexicons.htm contains the ATUS codes and instructions that are the result of the efforts of numerous researchers to develop a robust and general scheme. For those interested in more closely examining the codes, a data extract builder (ATUS-X) has been designed to make it easy for users to create data files that contain the time use, personal characteristic, and household characteristic variables they want, thus making the data more accessible to a broader audience. Finally, ATUS-X is a project dedicated to making it easy for researchers to use archived ATUS data. ATUS is an ongoing time diary study with freely available data archives from 2003 to the present. Under the leadership of Allen Johnson, researchers who had collected data employing instantaneous recording contributed to the establishment of a time allocation digital database of 13 small-scale societies and a coding scheme along with cultural descriptions (Johnson and Sackett 1998). The scheme is hierarchical, with 10 major categories and numerous subcategories especially geared for small-scale forager, agricultural, and pastoral groups (Johnson and Sackett 1998, 325–28; for a list of time allocation databases developed in this project, see http://www.yale.edu/ hraf/publications_body_completepublist.htm#Time%20Allocation%20Series). SIMULTANEITY

People can engage in more than one activity at a time. Therefore, the problem of what we call simultaneity is problematic in setting up a coding scheme. In the modern context, one can talk on a phone while preparing a meal. In a more traditional situa-

14_107-Bernard.indb 304

6/9/14 10:03 AM

9: Behavioral Observation

305

tion, encountered by those who study child care activities, are activities such as nursing a child while making a basket (Lamb and Hewlett 2005). There is no easy solution to this problem. One approach would be to record all behaviors with or without priority. To assign a particular behavior as the primary behavior, the researcher would have to apply a consistent rule for such a decision. In the examples above, environmental context may be a reasonable guide. In those examples, the primary activities would be food preparation (being in the kitchen is the context) and basket making. Johnson and Sackett (1998, 327) discuss the pros and cons of various solutions to this problem. CODES BEYOND SIMPLE BEHAVIOR

A typical data entry line in a behavior observation typically consists of the subject and date and time of the observation along with the location. How behavior varies with time and day of week or season and place is of obvious interest because such factors fundamentally affect behavior. For example, Hames (1987) used a location variable to measure how much time individuals spent laboring in their own gardens and those of their co-resident neighbors and the sharing of meals (Hames and McCabe 2007). Other researchers such as Sugawara (1988) used the location variable to document inter-camp visiting among !Kung foragers, and Ohtsuka et al. (2004) documented gender-based difference in the exposure to environmental toxins. Below, we describe some interesting uses to which behavior observations have been put that go beyond the recording of activities in space and time coordinates to gain fuller and more complex characterizations of behavior.

Internal States One can record a subject’s internal feelings. As noted earlier in experiential sampling and the newer DRM approach (Kahneman et al. 2004), a subject is asked to report his or her subjective state while they are engaged in a particular behavior (Chick 1994; Csikszentmihalyi and Larson 1987). Such internal states include whether they are bored, excited, focused, tired, impatient, and so on, as well as their behavior. Such information can be collected whether one is following instantaneous or continuous recording protocols. In psychology, Fisher and colleagues (Fisher et al. 2011) asked students to depress a counter (for golf) every time they thought of sleep, food, or sex. At the end of the day, they recorded the count and reset the counter. Finally, Trull and Ebner-Priemer (2009, 460) predict that biosensors to monitor physiological states such as heart rate and galvanic skin response will become common as their cost and obtrusiveness decrease. Food Transfers In egalitarian societies, individuals are expected to share food resources outside of the family, but how often it is done and to whom resources are given and received is difficult to measure. Initial research into this area (e.g., Henry 1951) revolved around watching food entering a camp and then observing distributions to individuals. This approach is fraught with a variety of sampling biases and is time consuming. Kaplan (Kaplan et al. 1984) used instantaneous observations to document food distributions among the foraging Ache. In the course of instantaneous recording, any time Kaplan

14_107-Bernard.indb 305

6/9/14 10:03 AM

306

Raymond Hames and Michael Paolisso

observed someone eating, he identified the kind of food eaten, asked who gave it, and who produced it. Through this information, he was able to richly describe foodsharing networks among the Ache.

Social Interactions Charting patterns of social interaction would seem to be an excellent way of revealing the social and communicative organization of any group. Who a person spends time with, how frequently they interact, and what sorts of activities they interact in are basic ways to understand the social nature of any group or personal network. Yet few researchers have bothered with this kind of analysis. Early on, Allen and Orna Johnson (1976) measured the amount of time husband and wife dyads among the Machiguenga spent together in a variety of activities. Sadly, although such research is essential to understanding the nature of the marital bond and household organization it has never been replicated. Mark Flinn (1988a, 1988b) did two interesting studies on mate and daughter guarding in a Trinidadian community, describing agonistic interactions between potential suitors of their wives and daughters. Sugawara (1984, 1988) mapped patterns of spatial interaction, conversation, grooming, and contact in a Kalahari San camp. Finally, Dunbar and Marriott (1997) recorded the topics of conversation among groups of students in pubs. Although conversation is one of the most common behaviors, and defines us as human, this study is the only one that bothers to classify conversational topics (e.g., gossip and work-related stories) in a naturalistic setting. Describing social interaction from an observational and naturalistic basis is a field ripe for investigation. The major exceptions to the dearth of studies on social interaction are the many studies of caretaker or alloparent interactions with infants and children. In these, researchers identify the children a subject is holding, feeding, grooming, or carrying. This sort of research has led to considerable understanding of the alloparenting networks that seem to be a fundamental aspect of social life in traditional societies (see, e.g., the collection in Lamb and Hewlett 2005). Computers and Software for Behavior Observation A number of software programs have been designed for data input by researchers or self-reports generated by subjects that run on computers, smartphones, and other devices capable of electronic data entry. We believe that those innovations allowing subject self-reports in real time represent an important methodological advance because they reduce problems of inaccurate reports and reactivity and permit frequent sampling and efficient data collection. Most of these programs not only permit data entry and management for both continuous and instantaneous recording, but also for observer training, interobserver agreement, data management, and data analysis as well as the integration of video and audio recording. Many of the products listed below represent long-term development of software that run on standard computer platforms and are designed for data entry by observers, not subjects, and most of the development has focused on continuous recording of behavior with the ability to integrate video and audio streams. However, the important innovation comes in hardware and software platforms, such as eCove and

14_107-Bernard.indb 306

6/9/14 10:03 AM

9: Behavioral Observation

307

Techneos Entryware, are designed for real-time self-reports. Some of the better-known vendors have Internet links to provide more details to interested parties. • EBASS or (Ecobehavioral Assessment Software Systems) at http://www.jgcp.ku.edu/~jgcp/ products/EBASS/ebass_descrp.htm is recommended only for those who have a need for instantaneous recording. The program is relatively inexpensive compared to the others, but only runs on Windows-based PCs. • Annotation at http://www.saysosoft.com/ is another relatively inexpensive program that runs only on Macs and can be programmed to record instantaneous or continuous behavior and can be integrated with video and audio recording. • eCove at http://www.ecove.net/ is an expensive multifaceted program with a variety of modules that runs on iPad, iPhone, Android operating systems, Windows PCs, and Apple computers. • Noldus at http://www.noldus.com/ is a software and hardware system. It is the richest in terms of features and appears to be used by more than any other package for behavior observation and is extremely well supported. Reviews of this package have been made by ethnographers (Ice 2004; Koster 2006). Unfortunately, it is exceptionally weak on instantaneous recording except through Pocket Observer, a bulky-looking, hand-held device with a raised keypad for relatively easy data entry. • Invivodata at http://www.invivodata.com/solutions/diarypro-ediary/ sells software and hardware for self-reports that can be run on a variety of hand-held devices. Importantly, they offer Internet recording of subject input and database management. • Entryware by Techneos Systems at http://www.techneos.com/sites/default/files/pdfs/ Entryware_product_info_sheet.pdf and Techneos Systems home page at http://www .techneos.com/ has an simple-to-set-up data entry program for self-reports and designed for use on smartphones, PDAs, and tablets with the ability to output the data to database and statistical programs.

The importance of products such as eCove and Entryware is that they are well designed for the development of easy-to-use data entry programs for self-reports (see Gravlee et al. [2006] for an example). While these software solutions have not been deployed extensively by ethnographers (for exceptions, see Gravlee et al. 2006 and DeCaro et al. 2012), they are widely used in educational, medical, and consumer fields, and we believe they have the potential to enhance the quality of self-reports. The software can be programmed to query subjects at random or fixed intervals. When a subject responds, a time- and date-stamped data-entry module opens, the subject responds by using drop-down lists to answer queries, and the saved file can be transmitted to a web site or emailed to the researcher. Invivodata offers web-based recording of subject reports that we believe to be an important innovation, especially when it supports a generic smartphone interface. Internet-based reporting coupled with “cloud” storage has obvious advantages over data stored in a personal electronic device. Data are securely stored and time stamped, subjects do not have to upload, and researchers can monitor subject compliance and behavior in near real time. CONCLUSION

Our goal in this chapter has been to review the theoretical and methodological progress that has been made in behavior observation research over the past few decades,

14_107-Bernard.indb 307

6/9/14 10:03 AM

308

Raymond Hames and Michael Paolisso

emphasizing anthropological contributions. In doing so, we have documented parallel efforts in allied behavior science fields such as animal behavior, behavioral medicine, and sociology. Behavior in naturalistic settings has always been important in the social sciences, but attempts to study behavior intensively have been slow to develop because of the perceived costliness of doing so and the comparative ease in asking informants to tell us what they do, an option unavailable to those who study nonhumans. To a large extent, the benefits of studying behavior have changed with the employment of theory demanding that behavior be studied in a quantifiable fashion to evaluate a variety of hypotheses. In this regard, Allen Johnson’s seminal introduction of spot checks (1975) has led to a proliferation of studies and methodological improvements in behavior observation. We need a new generation of behavioral research in anthropology using direct and indirect observations in creative ways and sharing methodological lessons learned. New technology from smartphone and web-based data entry holds the promise of reducing the problems of reactivity and costliness in direct observations and increasing the accuracy of indirect observation. While there have been steady methodological advances in behavioral medicine to increase the reliability and validity of indirect or subject-based observations, ethnographers have barely begun to grapple with these problems. Some lessons learned in behavioral medicine may be applicable to ethnographers, but great differences in what behaviors or internal states are studied may make some of the lessons learned inapplicable to the kinds of questions ethnographers ask. The information presented in this chapter covers a wide range of methodological strengths and weaknesses of behavioral research to date. We know a lot about how to observe and code behaviors, the strengths and weaknesses of existing approaches, and the associated methodological challenges. The widespread availability of cell phones capable of photos and videos, coupled with an almost global acceptance of sharing that information on the Internet, suggests that we are now more comfortable with recording ourselves and sharing that information. Our daily news is filled with these examples from the Arab Spring uprising, the Afghanistan war, and everyday rescues and natural disasters. This “new normal” on informal and public sharing of our actions and intents for behavioral research methods will create new research opportunities and challenges. For example: How can this almost global willingness, requirement in some cases, to capture and share behavior be linked with the approaches we described above? If individuals or groups are already actively documenting an event or behavior, how can behavior research integrate and complement this public documentation and use? How are ethical, moral, and research limits of observation research evolving in response to this new global and public-driven sharing of behavior? We are increasingly recording behaviors (on security and traffic cameras, Twitter, Facebook, etc.). On a global scale, individuals, organizations, and governments are collecting, interpreting, and acting on new forms of documented behaviors. The implications of these streams of behavior information for anthropological studies of behavior are fertile grounds for conceptual and methodological work. Future behavioral research in anthropology can build on a solid conceptual and methodological foundation and benefit from new technologies to collect and involve

14_107-Bernard.indb 308

6/9/14 10:03 AM

9: Behavioral Observation

309

study participants. This future research can also be collaborative with and contribute to new public-driven collection and interpretation of behaviors. It is a very exciting time to be an anthropologist engaged in behavioral research. REFERENCES

Altmann, J. 1974. The observational study of behavior. Behaviour 48: 1–41. Barriera-Viruet, H., T. Sobeilh, N. Daraiseh, and S. Salem. 2006. Questionnaires vs. observational and direct measurements: A systematic review. Theoretical Issues in Ergonomics Science 7: 261–84. Behrens, C. A. 1981. Time allocation and meat procurement among the Shipibo Indians of eastern Peru. Human Ecology 9: 189–220. Bernard, H. R., P. Killworth, D. Kroenfeld, and L. D. Sailer. 1984. The problem of informant accuracy: The validity of retrospective data. Annual Review of Anthropology 13: 495–517. Betzig, L., and P. Turke. 1985. Measuring time allocation: Observation and intention. Current Anthropology 26: 647–50. Bianchi, S., M. Milkie, L. Sayer, and J. Robinson. 2000. Is anyone doing the housework? Trends in the gender division of household labor. Social Forces 79: 191–228. Bird, D. W., and R. Bliege-Bird. 2005. Martu children’s hunting strategies in the Western Desert, Australia. In Hunter-gatherer childhoods: Evolutionary, developmental & cultural perspectives, ed. M. E. Lamb and B. S. Hewlett, 129–46. New Brunswick, NJ: Transaction. Bock, J. 2002. Learning, life history, and productivity. Human Nature 13: 161–97. Borgerhoff Mulder, M., and T. Caro. 1985. The use of quantitative observation techniques in anthropology. Current Anthropology 26: 232–62. Borgerhoff Mulder, M., and M. Milton.1985. Factors affecting infant care in the Kipsigis. Journal of Anthropological Research 41: 231–62. Chick, G. 1994. Experience sampling in anthropological research. CAM: Cultural Anthropology Methods Newsletter 6(2): 4–6. Czikszentmihalyi, M., and R. Larson. 1987. Validity and reliability of the experience-sampling method. Journal of Nervous and Mental Disease 178: 526–36. DeCaro, J., E. DeCaro, and D. Ashley. 2012. Investigating the social ecology of daily experience using computerized structured diaries: Physical activity among Mexican American young adults. Field Methods 24: 328–34. Dunbar, R. I. M., and A. Marriott. 1997. Human conversational behavior. Human Nature 8: 231–46. Durnin, J., and G. Passmore. 1967. Energy, work, and leisure. London: Heinemann. Fisher, T. D., Z. T. Moore, and M.-J. Pittenger. 2011. Sex on the brain?: An examination of frequency of sexual cognitions as a function of gender, erotophilia, and social desirability. Journal of Sex Research 49: 69–77. Flinn, M. 1988a. Mate guarding in a Caribbean village. Ethology and Sociobiology 9: 1–29. Flinn, M. 1988b. Parent-offspring interactions in a Caribbean village: Daughter guarding. In Human reproductive behavior, ed. L. Betzig, M. Borgerhoff Mulder, and P. Turke, 189–200. Cambridge: Cambridge University Press. Fouts, H. N., and M. E. Lamb. 2005. Weaning emotional patterns among the Bofi foragers of Central Africa: The role of maternal availability and sensitivity. In Hunter-gatherer childhoods: Evolutionary, developmental & cultural perspectives, ed. M. E. Lamb and B. S. Hewlett, 309–21. New Brunswick, NJ: Transaction. Gaggioli, A., G. Pioggia, G. Tartarisco, G. Baldus, D. Corda et al. 2013. A mobile data collection platform for mental health research. Personal and Ubiquitous Computing 17: 241–51.

14_107-Bernard.indb 309

6/9/14 10:03 AM

310

Raymond Hames and Michael Paolisso

Gravlee, C. C., S. N. Zenk, S. Woods, Z. Rowe, and A. J. Schulz. 2006. Handheld computers for direct observation of the social and physical environment. Field Methods 18: 382–97. Gurven, M., and H. Kaplan. 2006. Determinants of time allocation across the lifespan: A theoretical model and an application to the Machiguenga and Piro of Peru. Human Nature: An Interdisciplinary Biosocial Perspective 17: 1–49. Hames, R. 1987. Relatedness and garden labor exchange among the Ye’kwana. Evolution and Human Behavior 8: 354–92. Hames, R. 1992. Time allocation. In Evolutionary ecology and human behavior, ed. E. A. Smith and B. Winterhalder, 203–36. Chicago: Aldine de Gruyter. Hames, R. 2010. Production decisions and time allocation. In Society and environment: Methods and research design, ed. I. Vaccaro, E. A. Smith, and S. Aswani, 35–56. Cambridge: Cambridge University Press. Hames, R., and C. McCabe. 2007. Meal sharing among the Ye’kwana. Human Nature 18: 1–21. Harvey, S. A., M. P. Olortegui, E. Leontsini, and P. J. Winch. 2009. They’ll change what they’re doing if they know that you’re watching: Measuring reactivity in health behavior because of an observer’s presence: A case from the Peruvian Amazon. Field Methods 21: 3–25. Henry, J. 1951. The economics of Pilaga food distribution. American Anthropologist 53: 187–219. Hewlett, B. S., and M. E. Lamb. 2005. Emerging issues in the study of hunter-gatherer children. In Hunter-gatherer childhoods: Evolutionary, developmental & cultural perspectives, ed. M. E. Lamb and B. S. Hewlett, 3–18. New Brunswick, NJ: Transaction. Hipsley, E. H., and N. E. Kirk. 1966. Studies of dietary intake and the expenditure of energy by New Guineans. Vol. 147. Noumea, New Caledonia: South Pacific Commission (Technical paper no. 147). Ice, G. 2004. Technological advances in observational data collection: The advantages and limitations of computer-assisted data collection. Fields Methods 16: 352–75. Ivey Henry, P. K. 2000. Cooperative reproduction in Ituri forest hunter-gatherers: Who cares for Efe infants. Current Anthropology 41: 857–66. Ivey Henry, P. K., G. A. Morelli, and E. Z. Tronick, ed. 2005. Child caretakers among Efe foragers of the Ituri forest. New Brunswick, NJ: Transaction. Jacobs, J., and K. Gerson. 2004. The time divide. Cambridge, MA: Harvard University Press. Johnson, A. 1975. Time allocation in a Machiguenga community. Ethnology 14: 301–10. Johnson, A., and C. Behrens. 1989. Time allocation research and aspects of method in crosscultural comparison. Quantitative Anthropology 1: 313–34. Johnson, A., and O. Johnson. 1976. Male/female relations and the organization of work in a Machiguenga community. American Ethnologist 2: 634–48. Johnson, A., and R. Sackett. 1998. Direct systematic observation of behavior. In Handbook of methods in cultural anthropology, ed. H. R. Bernard, 301–30. Walnut Creek, CA: AltaMira. Kahneman, D, A. Krueger, D. Schkade, N. Schwarz, and A. Stone. 2004. A survey method for characterizing daily life experience: The day reconstruction method. Science 306: 1776–80. Kaplan, H., K. Hill, K. Hawkes, and A. M. Hurtado. 1984. Food sharing among Ache foragers of Eastern Paraguay. Current Anthropology 25: 113–15. Koster, J. 2006. The use of the Noldus Observer 5.0 and Psion handheld computer in a remote fieldwork setting. Field Methods 18: 430–36. Kramer, K. 2010. Cooperative breeding and its significance to the demographic success of humans. Annual Review of Anthropology 39: 417–36. Lamb, M. E., and B. S. Hewlett. 2005. Hunter-gatherer childhoods: Evolutionary, developmental and cultural perspectives. In Hunter-gatherer childhoods: Evolutionary, developmental &

14_107-Bernard.indb 310

6/9/14 10:03 AM

9: Behavioral Observation

311

cultural perspectives, ed. M. E. Lamb and B. S. Hewlett, 407–15. New Brunswick, NJ: Aldine Transaction. Marlowe, F. 2004. Who tends Hadza children? In Hunter-gatherer childhoods: Evolutionary, developmental, and cultural dimensions, ed. M. E. Lamb and B. S. Hewlett, 177–99. New Brunswick, NJ: Aldine Transaction. Martin, P., and P. Bateson. 1993. Measuring behavior. Cambridge: Cambridge University Press. Minge-Klevana, W. 1980. Does labor time decrease with industrialization? A survey of timeallocation studies. Current Anthropology 21: 279–98. Montgomery, E., and A. Johnson. 1976. Machiguenga energy expenditure. Ecology of Food and Nutrition 6: 97–105. Ohtsuka, R., N. Sudo, M. Sekiyama, C. Watanabe, T. Inaoka, and T. Kadono. 2004. Gender difference in daily time and space use among Bangladeshi villagers under arsenic hazard: Application of the compact spot-check method. Journal of Biosocial Science 36: 317–32. Paolisso, M., and R. Hames. 2010. Time diary versus instantaneous sampling: A comparison of two behavioral research methods. Field Methods 22: 357–77. Peregrine, P., D. R. Drews, M. Norlth, and A. Siupe. 1993. Sampling techniques and sampling error in naturalistic observation: An empirical evaluation with implications for cross-cultural research. Cross-Cultural Research 27: 232–46. Pettit, G. S., and J. E. Bates. 1989. Family interaction patterns and children’s behavior problems from infancy to 4 years. Developmental Psychology 25: 413–20. Pianta, R., J. Belsky, R. Houts, and F. Morrison. 2007. Opportunities to learn in America’s elementary classrooms. Science 315: 1795–96. Pianta, R., K. La Paro, and B. Hamre. 2009. Classroom Assessment Scoring System™ (CLASS®). Baltimore: Brookes Publishing. Presser, S., and L. Stinson. 1998. Data collection mode and social desirability bias in selfreported religious attendance. American Sociological Review 63: 137–45. Putnam, R. 2000. Bowling alone. New York: Simon and Schuster. Richards, A. 1939. Land, labour, and diet in Northern Rhodesia. London: Oxford University Press. Robinson, J., and G. Godbey. 1999. Time for life: The surprising ways Americans use their time. University Park: Pennsylvania State University. Roggman, L, J. H. Langlois, L. Hubbs-Tait, and L. A. Rieser-Danner. 1994. Infant day-care, attachment, and the file drawer problem. Child Development 65: 1429–55. Sandberg, J., and S. Hofferth. 2001. Changes in children’s time with parents: United States. Demography 38: 423–36. Sayer, L. 2001. Time use, gender and inequality: Differences in men’s and women’s market, nonmarket and leisure time. College Park: University of Maryland Press. Sayer, L., S. Bianchi, and J. Robinson. 2004. Are parents investing less in children? Trends in mothers’ and fathers’ time with children. American Journal of Sociology 107: 1–43. Scaglion, R. 1986. Importance of nighttime observations in time allocation studies. American Ethnologist 13: 537–45. Scelza, B. 2009. The grandmaternal niche: Critical caretaking among Martu Aborigines. American Journal of Human Biology 21: 448–54. Schor, J. 1991. The overworked American: The unexpected decline of leisure. New York: Basic Books. Shelley, K. 2005. Developing the American time use survey activity classification system. Monthly Labor Review June: 31–35. Shiffman, S. 2009. Ecological momentary assessment (EMA) in studies of substance use. Psychological Assessment 21 (Special Section: Psychopathology in Daily Life: Using Ecological Momentary Assessment Methods), 486–97.

14_107-Bernard.indb 311

6/9/14 10:03 AM

312

Raymond Hames and Michael Paolisso

Shiffman, S., A. Stone, and M. Hufford. 2008. Ecological momentary assessment. Annual Review of Clinical Psychology 4: 1–32. Smith, P. K. 1985. The reliability and validity of one-zero sampling: Misconceived criticisms and unacknowledged assumptions. British Educational Research Journal 11: 215–20. Stange, K., S. Zyzanski, T. Fedirko Smith, R. Kelly, D. Langa, S. Flocke et al. 1998. How valid are medical records and patient questionnaires for physician profiling and health services research?: A comparison with direct observation of patient visits. Medical Care 36: 851–67. Stinson, L. 1999. Measuring how people spend their time: A time-use survey design. Monthly Labor Review (August): 12–19. Stone, A. A., and S. Shiffman. 1994. Ecological momentary assessment (EMA) in behavioral medicine. Annals of Behavioral Medicine 16: 199–202. Stone, A. A., S. Shiffman, and A. Atienza. 2007. The science of real-time data capture: self-reports in health research. Oxford: Oxford University Press. Sugawara, K. 1984. Spatial proximity and bodily contact among the central Kalahari San. African Studies Monographs 3: 1–43. Sugawara, K. 1988. Visiting relations and social interactions between residential groups of the Central Kalahari San: Hunter-gatherer camp as a micro-territory. African Studies Monographs 8: 173–211. Szalai, A. 1972. The use of time. The Hague: Mouton. Takada, A. 2005. Mother-infant interactions among the !Xun: Analysis of gymnastic and breastfeeding behaviors. In Hunter-gatherer childhoods: Evolutionary, developmental & cultural perspectives, ed. M. E. Lamb and B. S. Hewlett, 123–44. New Brunswick, NJ: Aldine Transaction. Trull, T. J., and U. W. Ebner-Priemer. 2009. Using experience sampling methods/ecological momentary assessment (ESM/EMA) in clinical assessment and clinical research: Introduction to the special section. Psychological Assessment 21: 457–62. Vitzthum, V. 1994. Suckling patterns: Lack of concordance between maternal recall and observational data. American Journal of Human Biology 6: 551–62. Whiting, B., and J. Whiting. 1973. Methods for observing and recording behavior. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 282–315. New York: Columbia University Press. Winterhalder, B., and E. Smith. 2000. Analyzing adaptive strategies: Human behavioral ecology at twenty-five. Evolutionary Anthropology 9: 51–72. Wutich, A. 2009. Estimating household water use: A comparison of diary, prompted recall, and free recall methods. Field Methods 21: 49–68.

14_107-Bernard.indb 312

6/9/14 10:03 AM

CHAPTER 10

Person-Centered Interviewing and Observation Robert I. Levy and Douglas W. Hollan

Person-centered ethnography is a term used by Robert LeVine (1982) to refer to anthropological attempts to develop experience near ways of describing and analyzing human behavior. An effort is made to represent human behavior and subjective and intersubjective experiences from the point of view of the acting, intending, and attentive subject, and to actively explore the emotional saliency and motivational force of cultural beliefs, symbols, and structures, rather than to assume such saliency and force (see Hollan 2001). Such an approach, utilizing the forms of interviewing and observation we discuss in this chapter, enables one to investigate, in a fine-grained way, the complex interrelationships between individuals and their social, material, and symbolic contexts. This has been an issue of central importance to social theorists since Aristotle and has, if anything, been gaining significance over the last 15 years. For example, many contemporary anthropologists and other social scientists have become reengaged with issues of agency, subjectivity, intersubjectivity, and consciousness and how these play out in fieldwork and ethnography (Borneman and Hammoudi 2009; Davies and Spencer 2010), in empathy (Hollan and Throop 2011; Throop and Hollan 2008), in postmodern and late capitalism (Biehl et al. 2007, DelVecchio Good et al. 2008; Ortner 2005), in the variety of ways societies “individualize” or not in a global context (Bauman 2001; Beck and Beck-Gernsheim 2001; Yan 2009), and in morality and ethics (Faubion 2011; Lambek 2010; Robbins 2004; Zigon 2008), to name just a few. All of these issues, and many others, can be illuminated by using person-centered interviewing and observation. As Hollan mentions in a review written in 2001: Person-centered ethnography has persisted and will continue to develop not because it lends itself to the study of certain narrow, particular, historically and culturally bound issues, but because it offers a powerful way of grounding social, psychological, and even biological theories of human behavior in the lived experience of real people. (p. 62)

This is a methodology of very broad scope and should be useful not only to researchers in all four subfields of anthropology but to those in the other social science fields as well, including education (Stevens et al. 2008). How are community members constituted by their contexts? To what degree and in what way are they at least partially autonomous individuals, engaged in a dynamic, sometimes coercive, sometimes enabling interplay, with a context that is in some way separate from and alien to them? What is the nature and location of such constructs as “self,” “identity,” “agency,” “cognition,” and the like, in different kinds of 313

14_107-Bernard.indb 313

6/9/14 10:03 AM

314

Robert L. Levy and Douglas W. Hollan

communities? How are the phenomena relevant to these constructs differentially formed, stabilized, and located in the interplay between public and private spheres? What, in a particular place and in relation to a particular person, is to be thought of as a communal (sociocultural) form; what as an intrapersonal (psychological) form? When is this traditional dichotomy misleading, so that some synthesizing concept of behavioral locus and integration might be warranted? What do these various locations of local influences on thought, emotion, morality, and action have to do with social process and with the implications of history, stability, and change—and, indeed, with the idea of freedom? Such questions about the kinds of behavior people experience to be in the scope of their own self-conscious willing and agency and those they do not—encouraged, shaped, or imposed from without—are both raised and answered by person-centered interviewing and observation. We will be concerned in this chapter with certain techniques for investigating these relations of individual and context. A number of preliminary cautions and assertions are necessary. Our focus is on the relatively neglected anthropological study of individuals. These methods are not attempts to study individuals primarily in or within themselves (for, say, the purposes of some comparative personality theory, or some issue in general psychology, or to humanize an ethnography by appending life stories to it). Rather, they are attempts to clarify the relations of individuality, both as output and input, to its sociocultural and material contexts. The implication is that the study of individuals is an essential component of adequate social theory. We assume that the local forms of individuality and individual–social transactions have to be studied empirically in each particular setting and for each kind of problem (although we will also argue for a concept of types of settings for some purposes) and that we cannot thus use some a priori generic model of “anthropological man and woman.” The interviews we will be centrally concerned with are not just samples of discourse—not one kind of local discourse among others, not just narratives, not just life histories or autobiographies. They are in part such things, but they are conducted in an attempt to attenuate and disrupt ordinary and conventional patterns of social discourse. In so doing, we hope to elicit behavior that moves beyond role-determined surface scripts to suggest hidden or latent dimensions of the organization of persons and of the sociocultural and material matrix and their interactions. We do this to investigate the cognitive and emotional saliency of social and cultural models of behavior, rather than to assume such saliency. For example, by observing when an interviewee actually becomes elated, anxious, hopeful, frightened, and so on while discussing certain topics or experiences, we can assess if and when such displayed affect and behavior conforms to conventional models and scripts of behavior and if and when it does not. Identification of such congruencies and gaps between cultural models or scripts and experience is essential for theorizing how individuals and contexts are related and coconstitute one another. We are not only concerned with how the experiences of individuals may vary by cultural context. We are also interested with how they may vary by social, political, and economic structural and typological features such as size and density of population, social complexity, degree of embeddedness in larger global systems, history of colonial and postcolonial relations, and so on that may shape the relations between

14_107-Bernard.indb 314

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

315

public and private forms and transcend differences in cultural contexts per se. Thus, it may be that person–context relations in all relatively small, face-to-face communities are more similar than their differences in history and culture would otherwise suggest. For programmatic notes on structural and typological contrasts, see Levy (1989, 1994). We are, of course, aware of the ideological and methodological disagreements that some will have with the “cookbook” way in which we have necessarily written this chapter. However, we believe that the potential distortions implicit in some of our methods—including the differential power and status of the interviewer and the everpresent dangers of ethnocentrism and self-interest—are of the sort that are at least partially correctable when identified, brought to consciousness, and analyzed. Person-centered interviewing and observations are not made up of standard techniques such as those used by scientific technicians to ensure what they imagine to be valid and reliable (i.e., easily replicable) results. The interviewing and observing discussed here are rather more like learning how to perform a musical score. You have to know how to do it (or something very like it) in some important sense beforehand; you have to already be a musician. You have to make use of social, psychological, and probably evolutionary skills of social knowing and interpersonal interaction that you bring to the score. This means that none of what follows is to be followed mechanically. It is rather to be taken as a series of examples, of guides for the mobilization and education of a preadapted understanding. These methodological suggestions are no more mechanical and positivistic than is a musical score for skilled performers. And just as not everyone can go from the score to a satisfactory performance, not everyone can grasp skillfully (or is interested in) person-centered aspects of social behavior. (For recent discussions of the role of empathy in fieldwork and in human life more generally, see Hollan 2012; Hollan and Throop 2011; and Throop and Hollan 2008.) This outline of procedures has another problem. Particular practices will be suggested and related to particular topics of inquiry and categories of analysis. But these topics and categories are tentative and may well be changed as a result of trying to use them. They are often crudely approximate and are rough starting places, which may be undermined, revised, or changed altogether as a result of the investigations they generate. We use these guidelines merely to get the interview process started. We will be concerned here mostly with interviewing in person-centered ethnographic studies and only tangentially with other types of person-centered studies. However, interviewing must always be conducted in conjunction with studies that elucidate significant aspects of the social, cultural, and material contexts of the people to be studied, determine the relevant issues and topics to be covered in the interviews, and make the interview materials and questions intelligible to potential interviewees. Person-centered interviews must also be augmented by special or supplemental studies dictated by the investigator’s more particular research interests and developing sense of significant problems. For example, if you are particularly interested in mourning behavior, you will be asking questions not only about how funeral ceremonies are organized and managed, but also about experiences of loss and grief throughout the life course and about beliefs about the afterlife. If you are particularly interested in mental health issues, you will be asking questions not only about how

14_107-Bernard.indb 315

6/9/14 10:03 AM

316

Robert L. Levy and Douglas W. Hollan

mental and emotional problems are identified, labeled, and treated, but also about how interviewees experience stress and anxiety on a day-to-day basis and about what stages of life they find most challenging and why (Bernard 2011; Bradburn and Sudman 1979; Briggs 1986; Douglas 1985; Fontana and Frey 1994; Garrett and Zaki 1982; Gorden 1975, 1992; Holstein and Gubrium 1995; Hyman and Cobb 1975; Kahn and Cannell 1957; Kluckhohn 1945; McCracken 1988; Merton et al. 1956; Payne 1951; Quinn 2005; Richardson et al. 1965; Rubin and Rubin 1995; Spradley 1979; Werner et al. 1987; Whyte 1960, 1984). To the extent that person-centered interviews engage the interviewee as an informant (i.e., a knowledgeable person who can tell the anthropologist–interviewer about culture and behavior in a particular locale), these interviews are similar to other types of interviews discussed in the social science literature. However, person-centered interviews also engage the interviewee as a “respondent,” as an object of systematic study and observation in him- or herself. The interviewer observes and studies the interviewee as he or she behaves in the interview setting, as he or she reacts or responds to various probes, questions, and topics (Devereux 1967; Fink 2007; Greenson 1967; Pittinger et al. 1960; Rogers 1942; and Sullivan 1954). Contemporary examples of studies based on person-centered interviewing in anthropology include Anderson-Fye 2004; Csordas 1994; Hanssen 2002; Herdt and Stoller 1990; Hollan 2005c; 2008, 2011, 2012, 2013; Hollan and Wellenkamp 1994, 1996; Kendall-Taylor 2009; Kracke 1978; Lester 2011; LeVine 2010; Levy 1973, 1977, 1990; Obeyesekere 1981, 1990; Parish 1994, 1996, 2008; and Simon 2009. It is the balanced combination of informant and respondent modes of interviewing that is characteristic of person-centered interviews and that distinguishes them from most other types of interviews. THE INTERVIEWEE AS INFORMANT AND RESPONDENT

There is a significant difference between saying to a Tahitian interviewee something like “Please describe for me exactly how and why supercision (a penis-mutilating rite of passage) is done by Tahitians,” and asking him “Can you tell me about your supercision?” “What happened leading up to it?” “What happened that day?” “Did it change your life in any way?” “How?” “What did you think and feel about it then?” “What do you think and feel about it now?” The first question uses the interviewee as an informant, as an expert witness (albeit with a limited and special perspective) about some community belief or practice. The second set of questions treat the interviewee as a respondent, as an object of study in him- or herself; it explores what he or she makes of the belief or procedure. The relation between the two sets of answers is directly informative (see Levy 1973, 117–22) and illuminates the emotional impact of supercision on males in a way that the informantbased cultural description does not, so that the two types of description complement each other, both for social and psychological interpretations. Person-centered interviewing moves back and forth between the informant and the respondent modes. A remark of a young woman informant: “I felt very shy and embarrassed at that time” might be followed by a respondent-type probe: “Tell me more about how you felt,” or by the informant-type questions: “What do girls usually feel

14_107-Bernard.indb 316

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

317

under those circumstances?” “How do they usually act?” “If they don’t feel or act like that, what do people think?” These oscillations between respondent and informant modes illuminate the spaces, conflicts, coherences, and transformations, if any, between the woman-in-herself (either in her own conception, or in the interviewer’s emerging one) and aspects of her perception and understanding of her external context or life-world. Notice that if we stay at the descriptive and phenomenal levels, this unpacking of phenomena is not too difficult. However, problems arise when we try to theorize these separations and relations between person and context. We cannot in any satisfactory way simply equate private behavior with the mental or personality and non-private behavior with culture or society. The tensions, overlaps, and apparent polarities between person and context that person-centered interviewing reveal present challenges and opportunities in theorization to each investigator. A closely related problem is how we judge that the interviewee is giving objectively valid information as an informant. This determination, always tentative, starts with an interpersonal skill that we use in all conversation. It is based on the interviewer’s sense of consistency, knowledge of other people’s reports, and sense of the informant’s biases and peculiarities—which, when taken account of properly, sometimes make the informant a more valid reporter than a more “normal” informant, paradoxically. Person-centered interviewing generates a field of new phenomena, of reports and behaviors, that are then subject to interpretation. We will first consider the procedures of the interview process and note potential sources of distortion and misinterpretation arising from them before discussing how the interviews are to be interpreted and analyzed. PROCEDURES FOR CONDUCTING PERSON-CENTERED INTERVIEWS

Person-centered interviews utilize a combination of informant- and respondentdirected questions and probes. A probe is an intervention to elicit more information, not necessarily in the form of a question. An encouraging grunt may be a probe, as is a verbal directive like: “Tell me more about that.” Questions and probes may be relatively focused or closed (“Exactly what did he say that made you feel like that?”) or relatively vague or open ended (“What was going on?”). Open-ended probes are purposely ambiguous. This leaves the respondent a relatively wide choice of responses, dictated less by the exact answer to a factual question than by, presumably, some private concern or orientation. The interviewer tries to widen the constraints of the question as much as is relevant to what he or she is interested in at the moment. “Tell me about your childhood,” is more open than “Tell me about the people around you in your childhood family,” which is more open than “Tell me about your parents,” which is more open than “Tell me about your mother,” which is more open than “Tell me about how your mother disciplined you.” (We will return to open and closed interventions.) The specific choices and emphases in the contents of responses as well as the form of the responses made by the respondent within the relatively wide constraints of the probe are important material for analysis.

14_107-Bernard.indb 317

6/9/14 10:03 AM

318

Robert L. Levy and Douglas W. Hollan

WHEN TO BEGIN

Several factors determine when person-centered interviewing should begin.

Linguistic Competence Person-centered interviewing depends, of course, on adequate linguistic competence. Many anthropologists do not achieve enough fluency in the local language to do person-centered interviewing. It is not only harder to shape questions for respondent probes than for informant probes (it is harder to be properly vague in a foreign language than properly precise), but the linguistic nuances in the respondent’s discourse that convey personal information are often meaningful variations of the standard language. Interviewers have to know enough of the standard language to realize that the respondent is talking in a peculiar or significant way. Some anthropologists do not work in the native language. When informants are bilingual, anthropologists may work in the areal lingua franca, the national language (pidgin, trade Malay, Nepali etc.), or the former colonial language. When an informant is monolingual in some local language, the anthropologist might use bilingual assistants. Any of these translation shortcuts, if necessary, will systematically affect the respondent’s behavior, if only obscurely. Different aspects of identity, morality, selfpresentation, and mind are organized around and evoked by the use of each of the two (or more) languages the respondent uses. The local language is almost certainly most closely related to people’s private realms of experience and to their personal organization. It may be productive and theoretically interesting to work with informants in two (or several) languages for comparison, but it is preferable for person-centered interviews to be conducted in the respondent’s core or primary language. Much of the language forms used during respondent interviewing are different from many aspects of the language that the anthropologist learns for public discourse and informant interviewing. So it is often necessary to work out the vocabulary and phrasing of questions and probes with the help of a patient informant. It is also very helpful to try some pilot interviews with people who will not be major respondents. That way, if things go wrong, it will not compromise essential studies. (It is sometimes a good idea to do these interviews and the linguistic preparation with someone who lives outside the community; this minimizes community members’ expectations about the interviews.) The most helpful way to improve interviewing skills is to study closely the errors and misunderstandings revealed by recordings of interviews. Understanding Local Culture and Context Just as interviews will drift from public to private realms (below), so does the anthropologist’s understanding. It takes considerable general knowledge about a place and its people before we can begin to understand the presence and significance of private variants and transformations of local cultural and social forms. The fieldworker must, therefore, function as a general ethnologist before turning to person-centered studies, so relatively long periods of fieldwork are required. Trust and Respondents’ Motivations In the United States, people are often willing to discuss personal matters freely with experts who are strangers and to do so in many public arenas. Elsewhere, though,

14_107-Bernard.indb 318

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

319

where “face” and reputation are essential, a person does not discuss potentially reputation-threatening private worlds with someone unless that person trusts the other considerably. This is particularly true in relation to the anthropologist who will be living in the community and who will be in daily contact with many different local people, including the interviewee’s superiors and enemies. Much close observation of and contact with the anthropologist (insofar as that observation and contact leads to trust) is necessary before local people will consent to being relatively frank respondents. We must ask at this point why interviewees engage in interviews at all. It is easier to grasp—especially in small communities where anxieties about public reputation require much self-censorship and self-control or in communities where political positions may be full of danger—why people would not want to participate. But it has been our experience in a range of communities that almost all people we asked (representing a cross-section of their communities) were willing to become and remain respondents. Early in the anthropologist’s stay, people (unless they have good reason to keep silent) are motivated to engage superficially, at least, with the anthropologist and his or her interviews through politeness, hospitality, compassion for the naive and confused fieldworker, and curiosity. In some kinds of communities, there is a sense that it is prestigious to be chosen as an expert. In many communities, the respondent assumes that an exchange relation is being set up and that some sort of obligation on the part of the interviewer will result. Essentially, a long-lasting relationship of mutual obligation is being entered into, initiated by the respondent’s generosity. Thus, overt quid-pro-quo offers of payments “settling” the interviewer’s obligation are disturbing and often insulting to the respondent. In some communities we have worked in, it would have been a serious violation of the conventions of exchange to offer money at the start (i.e., to hire respondents). We gave substantial gifts or reciprocal services to respondents (sometimes, as in Nepal, to poor respondents, gifts of money), usually at the end of the interview series. We also found that we had assumed some very long-term obligations to informants and their families. In many contemporary, Westernized communities, informants and respondents become professionalized, and, in exchange for their time and expertise, anthropologists may hire them, as they would any other kind of expert. Such interviews may be problematic unless respondents become engaged for the sorts of personal reasons we will note below. All interviews provide some kind of immediate psychological reward for interviewees. They understand themselves to be taken seriously by someone they may perceive as having high status. They are selected over their peers and treated as an expert. Highstatus interviewees, on the other hand, may consider helping the anthropologist to be part of the responsibility (and validation) of their status. Such motivations may operate in the beginning phases of a respondent interview, but the interviewee soon realizes that something unusual is going on. He or she is not being treated for the most part as a cultural expert, nor in his or her public role, but as an individual, and in a rather deep and comprehensive way. In contrast to informant-directed interviews, which may be successful no matter whether the reward is intrinsic or extrinsic, the respondent interviewee must develop additional motivations for the interview to proceed properly. It has been our experience

14_107-Bernard.indb 319

6/9/14 10:03 AM

320

Robert L. Levy and Douglas W. Hollan

in diverse communities that insofar as respondent-centered interviews are conducted properly (in ways we shall try to specify), they become, when people slowly grasp what is going on, deeply meaningful to interviewees. More often than not, they express regret at the conclusion of the interview series, no matter how lengthy this has seemed to the interviewer. Why should this be? It seems that the opportunity to talk about, express, share and, in fact, construct aspects of one’s private world in conditions where this proves safe—away from immediate external communal censorship, in an interaction with someone who is sympathetic, nonjudgmental, disinterested, and trustworthy—is of considerable value for many, perhaps most people. This is particularly the case in small, face-to-face communities (or those under some sort of rigid social control) where it may be difficult to express, articulate, or explore private worlds, even with friends or family. Obviously, these are the very communities in which self-exposure is most dangerous. In these, and in all communities, it is essential to protect rigorously the reputation (and sometimes life and liberty) of individuals. Clearly, this is a profound responsibility for the anthropologist, who must protect the respondent through guarding and disguising field materials (e.g., through writing some things in a coded language) and resist any temptation to gossip. There may be some topics that are so sensitive and dangerous that they should not even be asked about, and certainly not recorded in ways that could possibly implicate the interviewee. Political participation in many countries is one such matter, as are behaviors defined as crimes, and any behavior the exposure of which would cause a significant threat to reputation. On other occasions, it may be safe to talk about some subjects so long as no recordings are made and written notes are coded or disguised in some other manner. LOCATION OF INTERVIEWS

To maximize private responses, it is important to interview the respondent apart from his or her family, friends, and acquaintances. The presence of others automatically shifts behavior and discourse toward public behavior and socially proper responses. The search for a private place for interviewing is one of the most difficult, sometimes impossible, matters in many communities. In many places, isolated, private, and secret conversations (even among friends and/or spouses, the latter being the case in many traditional Indian families) are deeply suspect as being an escape from communal or familial controls. The problem is deepened in the case of cross-gender interviews. In some communities, a man can interview a woman only if the woman brings a baby or young child with her. In others, such interviews are possible if the interlocutors have some sort of (and the proper kinds of) fictive kinship relation, a relationship that is often given to outsiders in kin-organized societies to facilitate social relationships. INTERVIEW TOPICS

The subject matter of an interview series consists of discussions of as wide a range as possible of the issues in the interviewee’s life that seem important to him or her and/or to the interviewer. Most of these will concern, on the one hand, the important forms and patterns of life found in a particular community (such as, e.g., being an untouchable, or a wife in an arranged marriage, or about life lived in the narrow constraints of

14_107-Bernard.indb 320

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

321

a tiny village), and, on the other hand, significant matters of more or less general human concern. It is especially important and interesting to explore what is at stake for the respondent in his or her daily life, from the respondent’s point of view (Kleinman 2006; Kleinman and Kleinman 1991; Wikan 1990). These accounts of aspects of lives are approached first descriptively, then with an emphasis on the interviewee’s personal experience and interpretations—that is, sometimes in the informant mode, sometimes in the respondent mode. Person-centered interviews consist, then, of topics of inquiry that are approached and dealt with in certain ways. We will consider the topics first and aspects of how to deal with them in later sections. Inquiries into specific topics generate stories. “This is the day-by-day life of us untouchables.” Those stories are directly meaningful in themselves, but the form in which they are told also yields a good deal of general information about aspects of personal organization, local context, and person–context relations that cut across many different topics. It is important to note that the list of interview topics we discuss here is selected and biased toward our own research interests (see Levy [1973, 509–11] and Hollan and Wellenkamp [1994, 227–29, 1996, 205–6] for more complete lists of interview topics). Other interviewers in other settings must develop their own set of topics, although it may be useful to cover most of the topics we discuss below for comparison with other studies and because they may prove to be valuable at a later time. Our list of topics attempts to highlight and probe the personal pole of person-context interactions. In actual interviews, both poles have to be probed. There are several reasons for our present emphasis. Relevant social and cultural patterns vary in different kinds of communities. The problems addressed in a caste system are different from those of a small, relatively egalitarian village, for example. A generalized list cannot deal with such variables. The topics of the personal pole, at least at the level of abstraction we are using here (identity, feeling), are perhaps more general than social variables. But the main reason for our present emphasis is that general aspects of context—economic forms, social class, gender, kinship structures, religious forms, modernization, state or governance structures, and so on—are the familiar subject matter of general anthropological inquiry and theory. The categories of the person–pole interacting with these public forms are relatively less familiar to anthropologists. Although many person-centered categories of inquiry and analysis may illuminate intrapsychic processes of various kinds, we understand them, as we have noted, as being open and interactional aspects of context–person relations. Conversely, at the social level, we would similarly expect such categories as “production” or “exchange” to be open and implicated in person–context relations. Many of the topics we discuss here are relevant to all respondents, but some will obviously have more relevance for some categories of people (women in their defining contrasts to men, high status versus low status, young versus old, and people with specialized roles—mothers, local leaders and religious leaders, etc.) than others. We can suggest only briefly what we mean by each of the 11 topic categories we discuss immediately below. For further understanding, it may be helpful to see how the categories are used in the literature we have cited. See especially Hollan (2000a, 2001) for broad discussions of much recent work in person-centered ethnography.

14_107-Bernard.indb 321

6/9/14 10:03 AM

322

Robert L. Levy and Douglas W. Hollan

Locating Information This cluster of issues locates individuals in the various significant components of their community’s social structure, both in their current activities and in their earlier life. It guides the inquiry of the first stages of the interviews and covers such matters as social roles, patterns of association, friends and enemies, economic status and activities, marital status, children—and whatever else may be socially and culturally significant for placing respondents in the local social network. Of importance here are not only specific relationships but generalized ones with classes of superiors and inferiors, allies and opponents, equals, people of the same and opposite sex, relatives and nonrelatives, people in the local community, and various kinds of outsiders. Not only is the individual to be located in his or her current life, but also developmentally in relation to what is taken to be significant past experience, particularly matters having to do with learning and with the developmental aspects of people’s present orientations and behavior. Here, we may include such things as patterns of life in the family of origin, parents’ and other caretaking adults’ roles and characteristics, sibling patterns, household caretaking and teaching practices, early economic activities, formal education, travel, and other putatively formative and socially differentiating matters. In societies that have rites of passage, it is important to explore memories and interpretations of the rites, and their effects, if any, on the respondent. In the early phases of the interview series, locating data are gathered principally in informant and conversational modes. One may ask a community leader what a leader does, how people achieve that status, and the like. It is only at some later point in the interview series that certain items in the locating data may be returned to in respondent mode. One might ask, for example, “How do you feel about being a community leader?” “What about that work makes you anxious?” “What is difficult about that work?” “What is rewarding about it?” “How do people judge and react to what you do?” It must be emphasized, in this case and in general, that these private responses to public roles may be deeply informative about such centrally social matters as, in this case, the dynamics and problems of authority and governance in a particular community. Patterns of Identification and Identity Formation The locating data and patterns of relationship are all related to the interviewee’s position in a historical and social nexus. This position is reflected and becomes a behavior-directing force, in part, in his or her identity, understood as the intersection of (aspects of) an individual and (aspects of) his or her communally provided social roles and social definition. Identity is reflected in the behavior-guiding orientation to the issue of “Who am I?” as formulated in different contexts, a conception of “identity” deriving largely from the work of Erik Erikson (1950, 1959, 1968). Answers such as “I am an aristocrat,” “I am myself,” “I am a member of such and such a family,” “I am a farmer,” “It depends on whom I am with,”1 and so on are all obviously of differential importance. These matters can be discussed directly with the respondent as well as inferred from other interview materials and from observations of the informant. In our experience, the direct question “What would you answer if I

14_107-Bernard.indb 322

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

323

asked ‘Who are you?’” addressed to someone who knows that you already know who he or she is, in some sense, produces very informative discussions. Ethnic, class, familial, gender, and other components of identity can be approached directly. Questions such as “What does it mean to be a Tahitian?” and “In what way are you like, or different from that?” directly probe aspects of categorization of self and, in defining contrast, others. The processes of identity formation can be approached through questions like “Who (an individual or class of people) did you want to be like when you were a child?” “In what ways?” “Why?” “At what age?” “Who did you wish not to be like?” One can trace the interviewee’s sense of developing consciousness of identity, and of shifting, multiple, or unified identities by discussing these issues directly—as well as by inferences from other interview materials. An important aspect of identity is locally constructed sexual identity. This entails definitions of the sexual Other (as all identity categories define classes of others). It is also closely related to and in part distinct from ideas and behaviors about physical sexuality and general orientations about bodies. Direct probes at sexual identity include the respondent’s definitions and evaluations of men and women and their lives, self-definition in relation to these general definitions, envy or disapproval of aspects of other sex and gender roles and characteristics, and advantages and disadvantages of being male or female. What traits do the interviewee sense in him- or herself to be masculine or feminine? Does he or she wish sometimes in fantasy to change sex or gender? All this overlaps with local and private forms, meanings, and attitudes about heterosexuality in contrast to homosexuality and other forms of sexuality.

Aspects of Self Quite different from the idea of identity is the local (and individual’s) conception about and the actual forms of self, if self is taken to be that subsystem of a larger mind that is the locus of “I,” of experientially autonomous action and, thus, of moral responsibility. An individual’s thoughts, feelings, and behaviors are variously sorted as belonging to self or to nonself. Respondents and/or their communities may variously explain nonself phenomena as emanating from some natural or supernatural source, alien to some essence of what it is to be a conscious and responsible individual. The concept of self is related to the idea of person, in that a community’s ideas about which members are socially responsible (persons) implies that they possess competent selves in one or another sense that must be locally investigated. Although the interviewer’s understanding of local ideas about selves and actual local structures of self are, for the most part, based indirectly on a range of miscellaneous interview materials, it is possible, usually late in the interview series, to ask about the respondent’s sense of which actions or thoughts are really his or hers, which just conventional actions, and which come from outside the self, and if from the outside, then, from where. Possession phenomena are of course centrally relevant to these issues (Hollan 1996, 2000a, 2000b; Levy 1996b). Such conceptions are related to the moral issue (below) of which thoughts and behaviors one and one’s associates are responsible for, and the somewhat different issue of which behaviors respondents believe they can or cannot control at some level of thought or action.

14_107-Bernard.indb 323

6/9/14 10:03 AM

324

Robert L. Levy and Douglas W. Hollan

Probes can be used to investigate the relation of ideas about self—its separation, blending, congruence, or antagonism—to the body or to some spiritual entity or to other parts of mind; one can investigate as well the stability or instability of the sense of self, the relation of self to different contexts, the relation of self to the local use of drugs and their effects (and to meditation and mystic experiences in some communities), the strength of the self in relation to internal and external pressures, the particular resources the self has to control itself, and so on. These latter issues are on the border of moral organization and conception.

Morality Related to the conception of the self as the particular focus within the larger mind of moral responsibility is a large set of issues related to aspects of moral evaluation and, especially, to self-control and attempts to control others that can be approached directly—as well as being implied and illustrated in discussions of many topics. We can explore respondents’ ideas about what constitutes good and bad behavior. Why is a particular behavior good or bad? What kinds of feelings and personal consequences are associated with doing good, doing bad? What kind of things is the respondent aware of wanting to do, but refrains from doing? Why? Ideas (and relevant episodes) bearing on what we would gloss as shame, guilt, sin, embarrassment, repentance, rectification, justice, and the like can be explored directly. The tensions between public ideas about moral matters and those of the respondent, and the relation of those tensions to his or her actions, are all important here. Anthropologists have always been interested in issues of morality and ethics (e.g., Parish 1994), but this interest has grown and become more focused in the last decade or so (Fassin 2012; Faubion 2011; Lambek 2010; Mattingly 2012; Robbins 2004; Zigon 2008). The Body A principal focus of private experience is the ways that people in various communities live in and through their bodies, their sense and evaluations of the functions and form of their own bodies and the bodies of others, and the relation of bodies to action, to a person’s value, to a person’s nature and constitution, and to self and identity. Various miscellaneous matters are relevant here, including body-centered sociocultural topics such as eating, cleanliness, pollution, excretion, menstruation, procreation, and physical sexuality. All of these can be explored and asked about directly. There is now an enormous literature on the anthropology of the body and the senses (Classen 1993; Csordas 1994, 2002; Desjarlais 2003; Geurts 2002; Howes 2003; Synnott 1993). Stress, Illness, and Healing Essential information about any systematically organized arena, particularly the personal and social forms we are concerned with here, is derived from how it breaks down, is defended from breakdown and is, if possible, ultimately restored. Various phenomena (locally interpreted in various ways) indicate problems (symptoms) in people’s internal adjustment and external adaptations and local ways of attempting to prevent, deal with, and relieve the corresponding discomforts and socially troubling behaviors.

14_107-Bernard.indb 324

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

325

These symptoms and healing and control techniques can be asked about directly and are important clues about local personal and social organization. Included here are such things as indications of the presence of and contexts for generating psychophysiological stress (e.g., anxiety, headaches, insomnia, sweating palms, poor appetite), mental disorders (as defined in local judgment, and/or in what seem to be some transculturally recognized form), and other symptoms of anxiety or depression, drinking and drugs, suicide, violence—and, in response to such things, local healing and problem-preventing practices.

Death Ideas, feelings, and responses concerning the respondent’s own death as well as the deaths of others are rich sources of information about self and identity, orientations to time, ideas of guilt and punishment, defense mechanisms, meanings of intimate relationships, aspects of aggression, and so on. These areas of investigation often show clearly the interplay between public doctrines and private orientations. Emotion As many studies (e.g., Levy 1984) have shown, it is possible to approach feeling and emotion directly, by exploring categories, terminology, experience, behavior, and evaluations of the respondent in relation to a range of what the interviewer considers (taking account of local definitions) to be the domain of emotions. Such an approach is supplemented by (and often clarifies) accounts of emotional responses in respondents’ stories about other matters and by direct observations of respondents’ emotional reactions during the interviews. Religion and the Supernatural In early psychological anthropology, certain areas of local culture were thought of as projections. These were areas that, because of a lack of material constraints (in contrast to the reality-constrained procedures of, say, farming or fishing), were thought to reveal deep psychological orientations. Foremost among these projective areas were religion and its associated symbols, rites, and rituals. The opposition between realistic and projective culture is problematic, but the exploration of the respondent’s understandings and uses of the culturally constituted realm of the supernatural is another rich source of psychocultural information. These personal understandings, uses, and experiences of the supernatural reveal meanings and functions of community religion that may be obscure at the level of cultural description. (See, for example, Levy’s [1990, 323–335] discussion of blood sacrifice in Bhaktapur and Hollan’s [1996] analysis of spirit possession in Toraja.) Fantasy, Creative Art, Dreams Our discussion about projective culture is also relevant to the relatively private creations of fantasy, plastic or literary art, and dreams, projective activities generated or, at least, arranged out of cultural pieces by the respondent him- or herself. Children The respondent’s discussion of his or her own children is a rich source of information on education, learning, morality, ideas of the self, and the like. Ideas about what

14_107-Bernard.indb 325

6/9/14 10:03 AM

326

Robert L. Levy and Douglas W. Hollan

children are like and how they should be dealt with illuminate a good deal of the respondent’s central orientations as well as provide clues to what formative experiences respondents had during their own childhoods (Levy 1996a). HOW THE INTERVIEWS ARE TO BE CONDUCTED

Topic areas suggest what issues and experiences the person-centered interviews might cover, but not how or in what manner they are to be conducted, or how to think about them. As we have noted, it is not only the particular topic areas that make these person-centered or respondent-oriented anthropological interviews, but also how the interviews are conducted. We now turn to a discussion of how the interviews should be carried out.

The General Stance of the Interviewer Perhaps the most basic technical matter in conducting interviews is the general stance of the interviewer toward the respondent. It closely resembles a familiar one in ordinary social interactions, the first phases of talking with a friend who has something important to discuss. We listen sympathetically and nonjudgmentally,2 drawing the friend out and trying to help him or her say what is on his or her mind. We don’t break off the friend’s report by changing the subject, by talking about ourselves, by prematurely advising or moralizing, or by demonstrating our cleverness. Our responsibility in this phase of listening is to (1) facilitate the friend’s communication; and (2) try to fully understand what he or she is trying to communicate and express. The conversation with the concerned friend then may move on to a second stage—that of trying to be helpful. If it moves on too quickly to the advice stage, the listener-friend is suspected of not doing the job rightly, of being too egoistic. For purposes of trying to understand the respondent, the anthropological interview maintains itself in this first phase. Being helpful is not our present concern (at least in our immediate role as ethnographers)—although we have argued that skillful listening and interviewing seems, in itself, to be helpful to the respondent. But just how does a friend listen properly and encourage his or her partner’s talking, thinking, and feeling? What are the interventions he or she uses? Here a discussion of specific elements of technique will be relevant; but we must first consider the basic problem of interview distortion. The Need for Self-Monitoring The validity and nonfactitiousness of loosely structured interviews depends almost entirely on the quality of the anthropologist’s interviewing, its constant correction through feedback from studying the interviews, and self-monitoring during the interviews. We are not concerned here with straightforward unskilled errors in technique (which can be corrected relatively easily). Rather, we are talking about systematic distortions in understanding that can profoundly bias an otherwise excellent methodology, errors that good technique and skillful theoretical rhetoric, in fact, obscure. We may roughly differentiate two overlapping sorts of distortion of understanding— cognitive and motivated. We will say most about the latter. Cognitive distortion is a misunderstanding of what is going on based on the interviewer’s prior intellectual and social experience with other kinds of people (than the

14_107-Bernard.indb 326

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

327

respondent) and with other kinds of contexts (than the respondent’s). At least some of this misunderstanding is correctable through ordinary learning based on what is often a pleasurable and rewarding perception of correctable error—the kind of learning that every anthropologist undergoes in a new field setting. This normal corrective learning is, in fact, the whole point of doing fieldwork. But distortion and misunderstanding may be motivated to serve some personal need of the interviewer (including guarding some deeply and centrally organizing cultural orientations). There is a significant difference here from simple ignorance. In the motivated case, the uncongenial true state of affairs may well be somehow known or intuited by the interviewer, stirring up defensive maneuvers and self-protective distortions, blunting or closing off threatening new understandings. These motivated processes tend to signal themselves. That is, if interviewers are attentive to their own sensations and responses, they can become aware of signs of the process of distortion while it is going on, although they may well be uncertain as to what, exactly, is going on. During an interview, the interviewer may be aware that something unusual or psychological is going on about which he or she is unclear—the unclarity being the point of the defensive operation. What the interviewer senses are abrupt and often mysterious alterations in his or her own feelings. The interviewer perceives an alteration of his or her state of concentration, alertness, or interest, and may feel anxious, bored, cloudy-minded, inattentive, irritated, excited, or whatever. The interviewer must struggle to understand what happened (either during the interview—in which case, a brief note of the response and of what seemed to be going on in the interview should be made—or later while listening to the recording of the interview). What was going on just prior to the disturbance? Was there a pattern to such feelings and alterations? Did they happen with certain kinds of respondents, with certain kinds of materials under discussion, with certain developments in the interviewer’s relationships with respondents? It is often difficult to grasp what is going on while an interview is taking place. Recordings (below) are invaluable for two main reasons. (1) Listening to the recordings will usually make evident what was occurring at the time of altered awareness. While listening to the recording, the same reaction is often generated again. (2) But the recordings have another kind of essential information on self-motivated distortion. They show the effects of these alterations on the interview process, for if the interviewer’s first response to uncongenial material is a matter of feeling and confusion, the next response is liable to be an action, distorting the interview and suppressing the disturbing material. There are various ways the interviewer may, more or less unconsciously, deflect the disturbing interview phenomena. For example, the interviewer may simply not hear what the respondent was trying to talk about, may intervene to change the subject or deflect it, or may signal his or her own anxiety, anger, or disapproval in some (often quite subtle) way. However they are produced, the distortions of the interview, and thus the understanding of what is going on, are usually evident when the recordings and transcripts are studied closely. In our experience in our own interviewing and in teaching interviewing, the (always embarrassing) awareness of these distorting responses and maneuvers may lead to significant self-corrections and learning.

14_107-Bernard.indb 327

6/9/14 10:03 AM

328

Robert L. Levy and Douglas W. Hollan

In the early days of psychological anthropology, psychoanalytically inclined anthropologists, following the claims of the earliest psychoanalytic anthropologists such as Geza Roheim, occasionally proposed that an extended psychoanalysis was necessary to avoid motivated distortions in fieldwork. Our experience has been, however, that short and intensive training (during one semester) in interviewing skills, focusing on practice interviews that are recorded for the purposes of corrective learning and feedback, serves this purpose. It alerts graduate students to problems in interviewing techniques, to habitually motivated distortions, and, particularly to the perception of the signs and symptoms of the personal needs and uneasiness that lead to distortion.

Guarding against Forcing a Wished-for Response It is essential that interviewers be particularly careful to be nondirective as they approach material that they sense they wish to make come out in a certain way, something, for example, that will strengthen an attractive developing hypothesis or bolster a cherished worldview. A kind of preemptive blandness is called for at these points. Guarding against Exaggerating the Fragility of the Interviewee In other sections, we comment on the sensitivity of certain topics that have to be recorded with care and sometimes avoided. Inexperienced interviewers sometimes, however, avoid topics of importance that they inappropriately fear are too sensitive. For many, this is a matter of an exaggerated fear of the fragility of the respondent in regard to emotionally loaded material that might stir up feelings of anxiety, anger, or sadness. Clearly, interviewers can be brutal, sadistic, and insensitive. Often, however, they will be too cautious, and, in so being (and by avoiding certain topics and by signaling back their anxiety to the interviewee, covertly entreating him or her to change the subject), close off the discussion of critically informative feelings, evaluations, concerns, and experiences. Listening for Ambiguity and Unpacking It Novice interviewers are often so anxious about avoiding silences and figuring out what the next question should be that they do not realize before listening to a recording of the interview when the interviewee has said something that is ambiguous enough that it requires further elaboration or explanation before its meaning becomes clear. Interviewers must listen carefully for such ambiguity, and, if possible, ask for clarification or expansion before the conversation flows to another topic. For example, the interviewer might say, “Just a moment ago you were talking about your feelings about divorce (or any other topic), could you tell me a little more about that (or any other subject)?” Opening Moves and Questions The interview series must start with topics and in a manner that seem natural to the respondent. Thus, we begin by asking about locating data—the usual ways of getting to know someone—and in an informant mode. Examples might be “Where were you born,” “Where do you live now,” “Describe your house to me,” “What do you do for a living,” “How do you spend a day,” and the like—developed subsequently in some

14_107-Bernard.indb 328

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

329

systematic way. The anthropologist asks for locating information, or information about the informant’s corner of public culture, treating the interviewee as an expert informant. Follow-up questions at this stage are to expand informant information, and questions about feelings and private interpretations are avoided. It is necessary at first to begin to map out via the informant mode the public world in which the informant exists, which is the world that provides the context for respondent studies. In the early phases, premature respondent-type interviewing would probably be unnecessarily disturbing to the interviewee. Shifts to the respondent mode come gradually over the set of interviews, as the interviewer approaches more personal target topics, asks questions probing personal feelings and ideas, and uses open-ended interviewing. The shift will begin when some mutual trust, sympathy, and understanding has been built up. The early phase allows the interviewee and interviewer time to get to know one another, at least in a preliminary way.

Interventions and Open and Closed Probes Open-ended interviews can be used in the early stages of investigating any sociocultural phenomena (e.g., to understand the range of a domain and the relative saliency of items in it—“Tell me about how people fish”), but such probes quickly lead to more focused, fact-finding questions. In person-centered, open-ended interviewing, there is a continuous back-and-forth movement between the open-ended probes and more specific questions. The interviewer’s responses in the interview may be called interventions. One purpose of interventions is simply to keep the interview going—grunts and nods signaling that the interviewer is listening and interested serve this purpose; without such signals, most interviews quickly dry up. But interventions—probes, questions, grunts, “Uh huh,” “Go on,” and the like—not only indicate attention, but are so placed as to guide the direction of the interview or encourage the continuation of some topic. As the introduction to this section notes, the ideal purpose of an open intervention (open in that it gives the interviewee maximum options in his or her response) is to encourage the interviewee to form the response as far as possible in terms of his or her personal organization. Closed interventions narrow the options for response, and often, but not necessarily, shift the response in the direction of objective (i.e., public) information. The openness or closedness of interventions is relative, on a continuum. Within the investigative arena of “mother” (all other areas having been for the moment closed off), “Tell me about your mother” is quite open; the interviewee must choose how to respond out of a large permissible universe of responses. “What did she look like,” is narrower, but there are still significant respondent choices. “What color hair did she have” is still narrower and a function of the interviewee’s descriptive competence, the interviewee now acting presumably mostly as an informant. Interviewees, at first, are often at least mildly surprised or uncomfortable with very open-ended questions that tend to violate the expectations of normal dialogue. Respondents, anxious about the unfamiliar mode of questioning or unsure of the intent or linguistic competence of the anthropologist, may ask something like “Just exactly what about my childhood do you want to know?” Usually, just repeating the question or slightly rephrasing it will produce a response, especially when interviewees are just

14_107-Bernard.indb 329

6/9/14 10:03 AM

330

Robert L. Levy and Douglas W. Hollan

checking to make sure that the anthropologist really meant to be so vague. Generally, however, even the most uncertain or timid interviewees learn to respond to openended probes as they become familiar with the interviewing format and as they come to trust that the interviewer really is interested in what they have to say. The interview is not a debate, nor, properly speaking, a dialogue. At first, the respondent may try to deal with the interview in the mode of two-person conversation familiar to him or her, and the interviewer must tactfully deflect this into the interview form. Respondents often pose polite questions to the interviewer such as “What do you think about this?” or “How do you folks do this?” Such questions can be temporarily deflected by a response such as “Let’s talk about this after the interview.” Insofar as the respondent’s question was dictated by politeness rather than curiosity, he or she will not bring it up after the interview, and the interviewer may not want to bring it up by him- or herself, at least until after the interview series is completed. While the discussion of an area is going on, the interviewer listens and signals that he or she is listening. The ability to listen attentively and accurately is critical to the interviewing process. This attentive listening not only helps the interviewer determine proper interventions, it also communicates to the respondent that the interviewer is interested in what he or she has to say. And it enables the interviewer to encourage the respondent to further elaborate on a particular topic by referring back to something already stated. Shy or fearful respondents who worry that they have nothing more of interest to say (or fear the interviewer is not hearing them properly) can be encouraged to go on if the interviewer says: “A while ago you mentioned X. Could you tell me more about what that was like, how it felt, etc.?” One of the hardest things for an inexperienced interviewer to tolerate is silence, the interviewee’s and his or her own. It is important to remember, however, that openended interviews are full of silences, as the respondent wonders whether it is all right to go on and continue to monopolize the discourse, pauses to choose where to go next, or struggles internally with what he or she is about to say. Interviewers need to learn to anticipate and to tolerate the silences. When the interviewer is convinced that a silence has persisted too long or that the respondent is having difficulty answering, he or she may say something like, “Yes?” or “What are you thinking?” or “Why are you hesitating?” If the interviewee’s discussion of some topic really seems at an end, the interviewer must move on. If the respondent has been describing an event, a situation, a social form, he or she may now be asked questions that move the discussion further into the personal realm—”How did you feel about that?” or “What is your own opinion about that?” or “Do you agree with what people say about it?” If the interviewer feels the topic is temporarily exhausted, he or she may move to related topics, or, if warranted, to some quite different area, excusing abruptness with a remark such as, “Let’s now talk about something else.”

Sensitive Topics Sometimes a topic that the interviewer senses may be uncomfortable for the respondent (e.g., matters related to the body and sexuality or to moral behavior) can be approached indirectly. One can, for example, ask first: “What do other people really do in these circumstances?” “What do wives really feel about their mothers-in-law?”

14_107-Bernard.indb 330

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

331

“Do young men go with prostitutes?” These discussions of others’ behavior prepare the interviewer and interviewee for the possible next probe: “What about you?” There is another way of approaching some of the potentially delicate topics related to body-centered states, sexuality, menstruation, and some aspects of stress and emotion. People in most places are accustomed to talking about illness. Discussing the history of illness and the physical symptoms of what are locally taken to be effects of illness can serve as a natural entry into discussions of other areas where physical states and bodily sensations are important.

Ending the Interview Series Ending the interview series requires as much tact and respect as beginning it. The respondent must be prepared for closing in previous interviews with, perhaps, a statement like: “There will just be a few more things we need to talk about, and then we’ll be finished with these interviews.” The interviews should never be ended abruptly or without the interviewee’s foreknowledge. A busy anthropologist who has been closely involved with a particular respondent for weeks might be tempted to suddenly drop such a respondent when beginning with a new one. This would violate local (and universal) understandings of relationship and exchange. The anthropologist has to find ways of keeping the relationship going and of showing that he or she still respects the respondent even after the formal interviews are over. NUMBER AND LENGTH OF INTERVIEWS

How many interviews are necessary in the study of a respondent? How long should each interview take? The distinction between informant and respondent is relevant in approaching these questions. An informant is asked to give systematically a set of verbal descriptions of local phenomena. This can be accomplished comparatively quickly. A respondent, on the other hand, is asked, in a sense, to react to a set of questions and probes. It is the shifting play of these reactions that provide the materials for a growing understanding of the respondent’s private world. And, although only the coherent discursive statements of an informant are useful, almost any verbal and bodily behavior of the respondent may prove informative. In contrast to an informant’s responses, those of a respondent do not get used up so quickly; they only begin to become noninformatively redundant and undemanding of further contextualizing after a large number of responses have been observed. Thus, it takes a number of interviews to discern patterns in the respondent’s responses. Finally, it will also take a certain number of interviews to cover the list of the target topics that the anthropologist has chosen to cover. The length of the interview series is an important technical tool in itself. The respondent’s personal world becomes more and more clarified as he or she comes to trust the interviewer and rapport deepens; as he or she gets some sense of what these peculiar interviews are about; as he or she returns with different perspectives to major private concerns; and as public role behavior begins to shift to private, more intimate modes of response. The depth, significance, and honesty of respondent responses tend to grow over the course of a series of interviews. Inexperienced interviewers fear that they will

14_107-Bernard.indb 331

6/9/14 10:03 AM

332

Robert L. Levy and Douglas W. Hollan

run out of questions, that the interviews will become sterile, that the respondent will become annoyed. This is not what happens in well-conducted interviews where understanding deepens, and where, as we have noted, the respondent usually regrets having the series discontinued. There are so many variables determining the length of the interview series that it is not possible to give more than very rough numbers. In our experience, depending on one’s goals, something between 6 (if goals or time are limited) and 20 or more interviews may be reasonable. For special purposes (to check, for example, the generality of some form among some class of respondents), 1 or 2 interviews may be sufficient. The question of the length of individual interviews is simpler—it is a question of how long the participants can pay close attention to each other. This is particularly problematic for the anthropologist who needs to be at his or her most alert. In our experience, it is difficult to be properly alert in an interview of more than an hour or so, although respondents’ limited availability often requires that the interviewer be prepared to conduct longer interviews at any time of the day or night. RECORDING AND TRANSCRIBING THE INTERVIEWS

Note Taking during the Interview It is best to keep note taking during the interview to a minimum. Writing may interfere with paying attention and it sends signals (often wrong ones) to the respondent that whatever he and she is doing or saying at the moment of note taking is more interesting than non-noted behaviors. This can skew the interview. Most notes about interviews will come from recall immediately after the interview and from the study of recordings. However, it may be necessary to note certain matters briefly for mnemonic purposes, notes that can then be expanded in writing after the interview. These notes may include: (1) important visual aspects of the respondent’s behavior (which, of course, would not be picked up by the audio recording); (2) significant events and shifts in the interviewer’s internal experience; and (3) guesses about some potentially illuminating and organizing hypothesis. These matters should be noted along with their context— that is, what seems to be happening in the interview at the moment they come to mind. They should be recorded as briefly as possible. Note Taking after the Interview Notes should be written as soon as possible after each interview. They should include a list of the topics covered in the interview, a list of new topics and questions suggested by the interview for later follow up, an impression of the significant behavior of the respondent, comments on the interactions between interviewer and respondent, comments on nonverbal behaviors, and comments on the interviewer’s reactions to various parts of the interview. The written notes should attend to all the matters that the interviewer thinks might be missed or obscured in the details of the audio recordings. Audio Recording If at all possible, interviews should be audio recorded. Recordings of well-conducted interviews contain an enormous amount of the kinds of information that allow interpretation of respondent–interviewer interactions and of the respondent’s personal

14_107-Bernard.indb 332

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

333

world. Although audio recordings obviously cannot capture the visual data of video recordings, we believe that respondents generally find them to be less distracting and obtrusive than video recordings and they do not pose as many problems in protecting anonymity and confidentiality. People who use video recordings have a special set of technical and methodological problems that are beyond the scope of this chapter. Many of the issues we discuss about audio recordings are relevant to video recordings, but there are obviously extensive additional problems of administration and analysis that are unique to video recordings. The ethics of video recordings are also different than those for audio recordings and need to be carefully considered in their own right. Listening to and transcribing recordings, particularly in the early stages of doing interviews, are excellent ways to improve language competence and become aware of linguistic and interviewing mistakes. Problems in pronunciation, vocabulary, grammar, and intervention phrasing can be spotted, and, if necessary, clarified in discussions with local people. Beyond helping improve linguistic and interviewing skills, recordings are essential for capturing the content and the form of interviews and are the only way (aside from video recordings) to capture the kinds of micro-patternings that are (as we shall see) essential phenomena for trying to understand the personal organization of respondents and their relations to their public cultures. Further, they are essential sources of information about the anthropologist’s conscious and unconscious biases and the linguistic and cultural blind spots that may critically distort interviews. Most novice interviewers fear that audio recording may disturb respondents. In our experience, if respondents agree to be interviewed, the recording device is not usually an additional disturbance. The respondent must be able to trust the interviewer that his or her privacy and interests will be rigorously protected. If trust has been established, the recorder does not usually make a difference, especially after the respondent has gotten used to it in the first interview or two. It is evident that audio recording of sensitive, embarrassing, or illegal activity is even more potentially compromising of the interviewee than are written notes. Because of that, it is sometimes necessary to turn off the recorder and make written (and coded) notes in discussing certain kinds of material. TRANSCRIBING THE RECORDINGS

Logging the Recordings It is difficult to find your way around in audio-recorded material, especially to relocate important material after the recording is first heard. As one listens to the recording, it is necessary to make a written log locating all potentially significant data by means of the recorder’s counter. This preliminary logging is still a fairly crude locating index of important materials—you don’t know what may be important until you have done later interviews with the respondent and with other respondents. Written Transcriptions The ideal way of working with recordings is to have written or typed transcriptions. These may be limited to those sections of the tape that the interviewer thinks are important or may include the entire interview—which is often of great value during later

14_107-Bernard.indb 333

6/9/14 10:03 AM

334

Robert L. Levy and Douglas W. Hollan

study of material. Levy used completely transcribed interviews in his work in Tahiti (1973) and Nepal (1990), as did Hollan and Wellenkamp when writing their two books on the Toraja of Indonesia (1994, 1996). The interviewer will almost certainly need help to make full written transcriptions of recordings because this requires a great deal of time and work. If at all possible, adequate money for the preparation of transcriptions should be written into the grant budget. Linguists studying in the area may welcome the chance to transcribe interviews as samples of local linguistic forms. In some places, there are people working as professional scribes who would find transcription related to their usual work. Sometimes, after returning from the field, you can find people from the language area who are studying or working abroad and who may be willing to do the transcriptions. In all these cases, obviously, the greatest care must be used not to compromise the identities of the respondents. All sensitive materials must be protected, and it is often better (if not required by human subjects protection) not to transcribe recordings that deal with sensitive subjects or illegal activities. Transcriptions prepared by others are really only first drafts that have to be corrected and augmented while the interviewer listens to the recording. Unless the transcriber is a linguist or has professional training, he or she will not be able to capture all of the interesting audible features that a researcher might want to examine. The untrained transcriber will clean up the recording, neglecting or omitting grammatical errors, stutters, abrupt changes, and hesitations in the flow of discursive speech, all of which are important for understanding the relation of the speaker to his or her utterance. The transcription must be corrected against the audio recording, the interviewer notating the presence of significant or interesting auditory features—features that reveal aspects of the state of the speaker and how a segment of discourse is intended by the speaker to be understood. These additions require the development of a code for indicating significant auditory forms and features. ANALYSIS OF THE INTERVIEWS

Types and Levels of Meaningful Phenomena The person-centered interview with its special topics and rhythms of open and closed questions generates the phenomena that the anthropologist must attend to and find ways of thinking about. We can sort some of the meaningful phenomena of the interviews into types and levels. See Quinn (2005) and Ewing (2006) for other excellent discussions along similar lines. CONTENT AS MEANINGFUL STORY

Interviews are full of stories recounted in the surface text. These are in part directly meaningful, but their meaning is modified significantly in the light of the formal aspects of the stories and their disruptions and in the qualities of the voice in which they are recounted. Because of our saturation in written texts and our habits of ordinary listening when we are considering speech as a report on reality (or as a story), we are most conscious of that part of a complex verbal communicative act that can be expressed as a written set of words—written in a certain way, with the hesitations and errors often treated as

14_107-Bernard.indb 334

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

335

“noise” and filtered out both in transcription, and, to a large degree, even in our hearing of the oral statement. This is what we usually think of as the text of a communication, of an utterance, of a conversation. Such texts were characteristic of many life history accounts in anthropology and provide a limited kind of information. There is, in fact, much significant material in the story so heard, the story-in-itself, for the relations of individuals to their culture. “I think ritual animal sacrifice is wonderful now, but when I was young I could feel the sacrificial knife against my own throat, and I thought, if adults could do that to animals, what would they do to me if they ever got angry at me” (see Levy 1990, 332ff). This is directly meaningful as a story. The meanings conveyed in the form of the story itself, in the disruptions of the story text and in the way it is told in its concrete telling, are attended to by the interviewer in a sort of intuitive and calibrating mode as he or she listens to the story—and are registered often as peripheral impressions. These peripheral features are often lost in conventional written transcriptions and in paraphrases of respondents’ stories. Different people are differentially sensitive to formal qualities of discourse. Not everyone is equally “psychologically minded” or intuitive about them, or formally trained in conversation analysis. It is, however, possible to become more sensitive to the form and quality of discourse, even without formal training. MEANINGS IN THE ORGANIZATION OF THE SURFACE TEXT

As we listen to a story, a statement, or an account, we often come to know more about it than what it seems to say on the surface. That extra knowing is often treated as being different from what we know from the surface discourse. We sense that “There is something strange about this,” or “Now I understand what he is trying to say,” or “He doesn’t really mean what he is saying.” Some of this intuited meaning is in the organization of the story. The way the story is told at the level of story (in contrast to interferences and prosodic phenomena that are not at the level of the story itself and that we deal with below) gives us much information. These formal characteristics of the story itself are facilitated by the freedom to proceed in storytelling, encouraged by open-ended interviewing. (Some of these story forms are culturally conventional, but others have to do—also or instead—with personal organization.) At the level of the story, we would look, among other things, at the following: What are the themes of the stories? Are there themes that are habitually introduced early in the discussions of a range of topics? Are there repeated or obsessive themes? What are the matters that are not talked about, if the context would have made talking about them expectable in some sense? Thus, in response to the relatively open query, “Tell me about your childhood family,” it would perhaps be significant if, say, the respondent said nothing about his or her father. How are the themes related in the discourse? How does the story flow from one thematic area to another? Is it possible that these flows show some logic of relationship and implication of different themes? DISTURBANCES OF THE SURFACE TEXT

Personal meanings play beneath the surface of the story and put pressure on the story, as shown in the organization of themes. Aspects of personal meaning also show

14_107-Bernard.indb 335

6/9/14 10:03 AM

336

Robert L. Levy and Douglas W. Hollan

themselves as disruptions of the ideal and culturally and linguistically correct surface form. These disruptions are of various kinds. They are often missed in ordinary listening, often going by too rapidly, and recordings are essential here. One of the classic kinds of disruption is slips of the tongue, where an error erupts into the surface. A Tahitian village informant, calling the missionary-introduced Christian god “they” instead of “he” (which she subsequently denied and was then embarrassed about when she heard it on the recording), seemed to suggest a submerged or implict personal proto-polytheistic sense of deity, a residue of old ideas and persisting family structure. Most disruptions are not slips of the tongue in which some unexpected meaningful content appears at the verbal or phonemic level. They are rather hesitations, stammers, interruptions, incompletions, or abrupt switches to a new direction. These suggest tensions and dynamics of organization that produce pressures on ideal discourse, which are often clarified by their locations in the story context. These disturbances must be indicated by some convention on the written transcriptions of the recordings. Another kind of disruption of ideal communication may be noted here. This is the matter of what the respondent makes of the interviewer’s probes and questions. (We assume for this purpose that the probes and questions are correct in terms of local rules for communication.) Those interventions that are ignored, misheard, or misinterpreted in comparison with the respondent’s usual responses can be treated as potentially meaningful disturbances of communication. PROSODY AND NONVERBAL BEHAVIOR

The way in which something is said often tells us as much or more about a person’s state of being than what he or she is saying more literally. Rhythms of speech, qualities of voice, breath, resonance, pitch, and so on may all indicate the relation of the speaker to his or her utterance as well as the kind of utterance it is meant or taken to be. These are ways that we convey (in alliance with contextual and visible clues) that a verbal statement, for example, “I went to town this morning” is (variously) a declaration, a question, full of some special importance, the introduction to a ghost story, uncertain, sincere, ironic, and much more—and also what kind of emotion is involved in the statement—rage, sadness, joy, fear, and the like. It is our experience that many of these prosodic and nonverbal markers are universals or else directly readable cross-culturally because they are signs or direct (or simulated) indexes of body states. These paralinguistic comments on the story text are relatively easy to grasp, and it is often possible in listening to recordings to locate the form (a quivering voice, a strong emphasis, a slowing of tempo, a quality of breathing, a sigh) that is conveying the information. (For more extensive commentary on gesture and nonverbal behavior, see Haviland 2000, 2004; Kendon 2004; Streeck et al. 2011; and Graham and Farnell, this volume.) Some of the information about the respondent’s momentary or characteristic organization (tension, stress, animation, excitement, relaxation, depression), the respondent’s emotional state, the meaning of some segment of speech, and the relations of cultural forms and private worlds is visual and communicated through facial expression, gesture, body posture, and physical movements. If video tape is not being used,

14_107-Bernard.indb 336

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

337

then the interviewer must make brief notes during the interview on what he or she observes—and expand on them as soon after the interview as possible—trying to discern what is being communicated by nonverbal phenomena, and ideally, exactly which aspects of nonverbal behavior are conveying a certain impression. SOME MISCELLANEOUS ASPECTS OF ANALYSIS

Categorization of Personal Forms We need a language to talk about—and think about—aspects of individuals and of social processes. When focusing on the person at the center of some world, we begin to tell our stories about their stories; we use our common sense and its largely unexamined terms or categories—this is how he or she feels, thinks, copes. But when it is necessary to theorize beyond the first tentative modeling of the respondent’s story, more practicespecific theoretical terms are required, terms derived from our professional intellectual history. The considerable danger of terms and concepts bearing on the personal pole of social experience is, as we have noted, that they have an intrapsychic, within-the-head bias, and much a priori and hidden theoretical baggage. Nevertheless, with due caution, many of the standard concepts of personality and general psychology have been productive as first approximations—with the caution that the anthropological task is to discover which items have to be added to, opened up, altered, refined, or, perhaps, rejected for the purposes of theorizing the differential effects of history and community on individuals, and individuals’ effects, in turn, on that history and community. Useful person-centered foci of theorization include such categories as self, identity, person, thinking, feeling and emotion, moral controls, defense mechanisms (to deal with problematic internal matters), coping mechanisms (how an individual tries to deal with problematic external problems), and conscious and unconscious aspects of mind, unconscious forms being deduced by the interviewer from various forms of the respondent’s discourse and behavior. (The reader is referred to the reference citations and to the vast literature on person-description for implications and use of these foci.) In relation to the problem of unconsciousness, it is useful to try to distinguish a “cognitive unconscious,” the inability to know things that are not named or cognitively discriminated by the local culture from a “dynamic unconscious,” the repression of painful knowledge that is disruptive or uncongenial to the self.3 Some of these categories can be approached directly as interview topics, but all of them can be discerned in the examination of various kinds of content and forms in the course of the interviews. The Extent and Locus of Personal Forms Suppose we perceive that the respondent is, say, anxious. It makes a significant difference if he or she is anxious because of the interview situation in general, because of some aspect of his or her shifting or sustained relationship to the interviewer, because of the specific topic being discussed, or because anxiety is a diffuse characterological aspect of the respondent. These are not necessarily independent of one another; the anxiety may be due to some combination of factors and include discussing a particular subject with this particular interviewer. A common source of error in interview analysis is to observe or identify a behavior accurately but to misplace, overgeneralize, or otherwise misunderstand its significance.

14_107-Bernard.indb 337

6/9/14 10:03 AM

338

Robert L. Levy and Douglas W. Hollan

Expression versus Communication It is useful to attempt to distinguish between expressive interview phenomena that indicate directly and unintentionally something about the state of the respondent (such as tearing, blushing, flushing, trembling and other involuntary meaningful acts) from acts intended as communication, intended to influence the observer or interviewer. Many behaviors are mixed, and it is possible to simulate purposely some expressive behavior for an intended communicative end. But it is also clear that some behaviors are relatively more shaped by personal organization, some more by the motivated use of some convention for the purposes of communicating an impression. Being able to distinguish between the two is essential to the analyses we wish to make. A NOTE ON OBSERVATIONS OUTSIDE OF THE INTERVIEW

Interviewers obviously encounter the interviewees to various extents outside of the interview. In a large and stratified community, they may encounter the respondent only rarely; in a small village community, respondents may be encountered every day and in a wide range of settings and situations. We cannot deal here with observation of individuals in public contexts. These public behaviors must be brought into relation to interview behaviors, and are, wherever available, essential data for observing the consonances, conflicts, and transformations of behavior and discourse in different local settings. Whenever possible, person-centered interviews should be conducted alongside more general ethnographic observation, since the two forms of data are mutually informative and telling. CONCLUDING NOTE

The main concern of person-centered interviewing and observation is to find ways to explore the relations—the spaces, condensations, unities, divisions, transformations, tensions, and coercions—that operate in the interplay of more public realms of experience, situated at different distances in space and time from individuals, and the more private ones constituting and impacting immediately on individuals. How do these realms affect and construct each other? To what degree are they relatively autonomous? The major difficulty in talking about how to think about and do person-centered studies is our heritage of folk, social science, and psychological conceptions. In its mainstream at least, this makes us separate person and context too neatly into different realms—the person and the community—and makes “person” too stable. This separation and reification of individual and context is based on only one aspect of our experience of living and has traditionally served to support many compelling moral, religious, and ideological purposes. Another, and more peripheral mood and ideology (shared by mystics and some structuralists and poststructuralists), is to make the individual and individual agency an illusion, a node in some field of history or context.4 The actual relations of individuals and individuality and agency to context vary according to different kinds of communities, positionality within each community, different individual temperaments, and different situations. This complex and shifting interplay is, however, more than a formless and ungraspable kaleidoscope of shifting shapes. These various forms and relations have explanations and are affected by the kinds of historical, communal, and personal forms that we know a good deal about.

14_107-Bernard.indb 338

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

339

Person-centered ethnography is a work in progress toward the slow empirical and conceptual elucidation of the interplay of social and personal forces. NOTES

Robert I. Levy died on August 29, 2003, in Asolo, Italy. For an evaluation and appreciation of his intellectual legacy, see the special issue of Ethos in his honor (Hollan 2005a, 2005b). 1. This example is from a respondent in a Nepalese Hindu city; see Levy (1990, 31, n. 11). 2. There are serious ethical issues involved here. If you have invited yourself into another person’s life, as most anthropologists do, then you have an obligation to listen seriously and nonjudgmentally to what this other person has to say. If you cannot do that, for whatever reason, then you have no business inviting yourself into the person’s life. This does not mean that anthropologists have to agree with everything their interviewees say or do, but it does mean they must listen and observe respectfully and conscientiously. 3. This distinction is related to our discussion of the difference between cognitive distortion and motivated distortion. 4. For a discussion of individual agency versus sociocultural determinism, see Levy (1998). REFERENCES

Anderson-Fye, E. 2004. A Coca-Cola shape: Cultural change, body image, and eating disorders in San Andres, Belize. Culture, Medicine, and Psychiatry 28: 561–95. Bauman, Z. 2001. The individualized society. Cambridge: Polity Press. Beck, U., and E. Beck-Gernsheim. 2001. Individualization: Institutionalized individualism and its social and political consequences. Thousand Oaks, CA: Sage. Bernard, H. R. 2011. Research methods in anthropology: Qualitative and quantitative approaches, 5th ed. Lanham, MD: AltaMira. Biehl, J., B. Good, and A. Kleinman, eds. 2007. Subjectivity: Ethnographic investigations. Berkeley: University of California Press. Borneman, J., and A. Hammoudi, eds. 2009. Being there: The fieldwork encounter and the making of truth. Berkeley: University of California Press. Bradburn, N. M., and S. Sudman. 1979. Improving interview method and questionnaire design. San Francisco: Jossey-Bass. Briggs, C. L. 1986. Learning how to ask: A sociolinguistic appraisal of the role of the interview in social science research. New York: Cambridge University Press. Classen, C. 1993. Worlds of sense: Exploring the senses in history and across cultures. New York: Routledge. Csordas, T. J. 1994. The sacred self: A cultural phenomenology of charismatic healing. Berkeley: University of California Press. Csordas, T. J. 2002. Body/meaning/healing: Contemporary anthropology of healing. New York: Palgrave. Davies, J., and D. Spencer, eds. Emotions in the field: The psychology and anthropology of fieldwork experience. Stanford, CA: Stanford University Press. DelVecchio Good, M. J., S. T. Hyde, S. Pinto, and B. J. Good, eds. 2008. Postcolonial disorders. Berkeley: University of California Press. Desjarlais, R. 2003. Sensory biographies: Lives and deaths among Nepal’s Yolmo Buddhists. Berkeley: University of California Press. Devereux, G. 1967. From anxiety to method. New York: Humanities Press. Douglas, J. D. 1985. Creative interviewing. Beverly Hills, CA: Sage. Erikson, E. 1950. Childhood and society. New York: W. W. Norton.

14_107-Bernard.indb 339

6/9/14 10:03 AM

340

Robert L. Levy and Douglas W. Hollan

Erikson, E. 1959. Identity and the life cycle: Selected papers. Psychological Issues 1: 1–171. New York: International Universities Press. Erikson, E. 1968. Identity: Youth and crisis. New York: W. W. Norton. Ewing, K. P. 2006. Revealing and concealing: Interpersonal dynamics and the negotiation of identity in the interview. Ethos 34: 89–131. Fassin, D., ed. 2012. A companion to moral anthropology. Malden, MA: Wiley-Blackwell. Faubion, J. D. 2011. An anthropology of ethics. Cambridge: University of Cambridge Press. Fink, B. 2007. Fundamentals of psychoanalytic technique: A Lacanian approach for practitioners. New York: W.W. Norton. Fontana, A., and J. H. Frey. 1994. Interviewing: The art of science. In Handbook of qualitative research, ed. N. K. Denzin and Y. S. Lincoln, 361–76. Thousand Oaks, CA: Sage. Garrett, A. M., and E. P. Zaki. 1982. Interviewing, Its principles and methods, 3rd ed., rev. by M. M. Mangold and E. P. Zaki. New York: Family Service Association of America. Geurts, K. L. 2002. Culture and the sense: Bodily ways of knowing in an African community. Berkeley: University of California Press. Gorden, R. L. 1975. Interviewing: Strategy, techniques, and tactics, rev. ed. Homewood, IL: Dorsey Press. Gorden, R. L. 1992. Basic interviewing skills. Itasca, IL: F. E. Peacock Publishers. Greenson, R. R. 1967. The technique and practice of psychoanalysis. New York: International Universities Press. Hanssen, K. 2002. Ingesting menstrual blood: Notions of health and bodily fluids in Bengal. Ethnology 41: 365–79. Haviland, J. 2000. Pointing, gesture spaces, and mental maps. In Language and gesture: Window into thought and action, ed. D. McNeill, 13–46. Cambridge: Cambridge University Press. Haviland, J. 2004. Gesture. In A companion to linguistic anthropology, ed. A. Duranti, 197–221. Malden, MA: Blackwell. Herdt, G., and R. J. Stoller. 1990. Intimate communications: Erotics and the study of culture. New York: Columbia University Press. Hollan, D. 1996. Cultural and experiential aspects of spirit beliefs among the Toraja. In Spirits in culture, history, and mind, ed. A. Howard and J. Mageo, 213–35. London: Routledge. Hollan, D. 2000a. Constructivist models of mind, contemporary psychoanalysis, and the development of culture theory. American Anthropologist 102: 538–50. Hollan, D. 2000b. Culture and dissociation in Toraja. Transcultural Psychiatry 37: 545–59. Hollan, D. 2001. Developments in person-centered ethnography. In The psychology of cultural experience, ed. H. Mathews and C. Moore, 195–214. New York: Cambridge University Press. Hollan, D. 2005a. Mind and experience in Tahiti, Nepal, and beyond. Ethos 33: 430–32. Hollan, D. 2005b. Setting a new standard: The person-centered interviewing and observation of Robert I. Levy. Ethos 33: 459–66. Hollan, D. 2005c. Dreaming in a global world. In A companion to psychological anthropology: Modernity and psychocultural change, ed. C. Casey and R. B. Edgerton, 90–102. Malden, MA: Blackwell. Hollan, D. 2008. Selfscapes of well-being in a rural Indonesian village. In Pursuits of happiness: Well-being in anthropological perspective, ed. G. Mathews and C. Izquierdo, 211–27. New York: Berghahn Books. Hollan, D. W. 2011. The vicissitudes of “empathy” in a rural Toraja village. In The lives of others: Empathy in the Pacific, ed. D. W. Hollan and C. Jason Throop, 195–214. New York: Berghahn Books. Hollan, D. 2012. Cultures and their discontents: On the cultural mediation of shame and guilt. Psychoanalytic Inquiry 32: 570–81.

14_107-Bernard.indb 340

6/9/14 10:03 AM

10: Person-Centered Interviewing and Observation

341

Hollan, D. 2013. Sleeping, dreaming, and health in rural Indonesia and the urban U.S.: A cultural and experiential approach. Social Science and Medicine 79: 23–30. Hollan, D. W., and C. Jason Throop, eds. 2011. The anthropology of empathy: Experiencing the lives of others in Pacific societies. New York: Berghahn Press. Hollan, D. W., and J. C. Wellenkamp. 1994. Contentment and suffering: Culture and experience in Toraja. New York: Columbia University Press. Hollan, D. W., and J. C. Wellenkamp. 1996. The thread of Life: Toraja reflections on the life cycle. Honolulu: University of Hawaii Press. Holstein, J. A., and J. F. Gubrium. 1995. The active interview. Thousand Oaks, CA: Sage. Howes, D. 2003. Sensual relations: Engaging the senses in culture and social theory. Ann Arbor: University of Michigan Press. Hyman, H. H., and W. J. Cobb. 1975. Interviewing in social research. Chicago: University of Chicago Press. Kahn, R. L., and C. F. Cannell. 1957. The dynamics of interviewing. New York: John Wiley. Kendall-Taylor, N. 2009. Treatment seeking for a chronic disorder: How families in coastal Kenya make epilepsy treatment decisions. Human Organization 68: 141–53. Kendon, A. 2004. Gesture: Visible action as utterance. Cambridge: Cambridge University Press. Kleinman, A. 2006. What really matters: Living a moral life amidst uncertainty and danger. New York: Oxford University Press. Kleinman, A., and J. Kleinman. 1991. Suffering and its professional transformation: Toward an ethnography of interpersonal experience. Culture, Medicine, and Psychiatry 15: 275–301. Kluckhohn, C. 1945. The personal document in anthropological science. In The use of personal documents in history, anthropology, and sociology, ed. L. Gottschalk, C. Kluckhohn, and R. Angell, 79–173. New York: New Science Research Council Bulletin 53. Kracke, W. 1978. Force and persuasion: Leadership in an Amazonian society. Chicago: University of Chicago Press. Lambek, M., ed. 2010. Ordinary ethics: Anthropology, language, and action. New York: Fordham University Press. Lester, R. J. 2011. How do I code for black fingernail polish? Finding the missing adolescent in managed mental health care. Ethos 39: 481–96. LeVine, R. A. 1982. Culture, behavior, and personality: An introduction to the comparative study of psycho-social adaptation, 2nd ed. New York: Aldine. LeVine, R. A. 2010. Commentary: From the mother’s point of view. Ethos 38: 458–60. Levy, R. I. 1973. Tahitians: Mind and experience in the Society Islands. Chicago: University of Chicago Press. Levy, R. I. 1977. Tahitian gentleness and redundant controls. In The socialization of aggression, ed. A. Montague, 222–35. Oxford: Oxford University Press. Levy, R. I. 1984. Emotion, knowing and culture. In Culture theory: Essays on mind, self, and emotion, ed. R. A. Shweder and R. A. LeVine, 214–37. Cambridge: Cambridge University Press. Levy, R. I. 1989. The quest for mind in different times and different places. In Social history and issues in human consciousness, ed. A. Barnes and P. Stearns, 3–40. New York: New York University Press. Levy, R. I. 1990. Mesocosm: The organization of a Hindu Newar city in Nepal. Berkeley: University of California Press. Levy, R. I. 1994. Person-centered anthropology. In Assessing cultural anthropology, ed. R. Borowsky, 180–87. New York: McGraw Hill. Levy, R. I. 1996a. Essential contrasts; Differences in parental ideas about learners and teaching in Tahiti and Nepal. In Parents’ cultural belief systems, ed. S. Harkness and C. Super, 123–42. New York: Guilford Publications.

14_107-Bernard.indb 341

6/9/14 10:03 AM

342

Robert L. Levy and Douglas W. Hollan

Levy, R. I. 1996b. Gods, spirits and history: A theoretical perspective. In Spirits in culture and mind, ed. J. Mageo and A. Howard, 11–28. London: Routledge. Levy, R. I. 1998. Selves in motion. In Selves in time and place: Identities, experience and history in Nepal, ed. D. Skinner, A. Pach, III, and D. Holland, 321–30. Lanham, MD: Rowman & Littlefield. Mattingly, C. 2012. Two virtue ethics and the anthropology of morality. Anthropological Theory 12: 161–84. McCracken, G. D. 1988. The long interview. Beverly Hills, CA: Sage. Merton, R. K., M. Fiske, and P. L. Kendall. 1956. The focused interview: A manual of problems and procedures. Glencoe, IL: Free Press. Obeyesekere, G. 1981. Medusa’s hair: An essay on personal symbols and religious experience. Chicago: University of Chicago Press. Obeyesekere, G. 1990. The work of culture: Symbolic transformation in psychoanalyis and anthropology. Chicago: University of Chicago Press. Ortner, S. 2005. Subjectivity and cultural critique. Anthropological Theory 5: 31–52. Parish, S. M. 1994. Moral knowing in a Hindu sacred city. New York: Columbia University Press. Parish, S. M. 1996. Hierarchy and its discontents. Philadelphia: University of Pennsylvania Press. Parish, S. M. 2008. Possible selves: Subjectivity and suffering in American culture. New York: Palgrave. Payne, S. L. 1951. The art of asking questions. Princeton, NJ: Princeton University Press. Pittinger, R. E., C. F. Hockett, and J. J. Danehy. 1960. The first five minutes: A sample of microscopic interview analysis. Ithaca, NY: Paul Martineau. Quinn, N., ed. 2005. Finding culture in talk: A collection of methods. New York: Palgrave. Richardson, S. A., B. S. Dohrenwend, and D. Klein. 1965. Interviewing: Its forms and functions. New York: Basic Books. Robbins, J. 2004. Becoming sinners: Christianity and moral torment in a Papua New Guinea society. Berkeley: University of California Press. Rogers, C. 1942. Counseling and psychotherapy. New York: Houghton Mifflin. Rubin, H. J., and I. S. Rubin. 1995. Qualitative interviewing: The art of hearing data. Thousand Oaks, CA: Sage. Simon, G. M. 2009. The soul freed of cares? Islamic prayer, subjectivity, and the contradictions of moral selfhood in Minangkabau, Indonesia. American Ethnologist 36: 258–75. Spradley, J. P. 1979. The ethnographic interview. New York: Holt, Rinehart, and Winston. Stevens, R., K. O’Connor, L. Garrison et al. 2008. Becoming an engineer: Toward a three dimensional view of engineering learning. Journal of Engineering Education 97: 355–68. Streeck, J., C. Goodwin, and C. LeBaron, eds. 2011. Embodied interaction: Language and body in the material world. New York: Cambridge University Press. Sullivan, H. S. 1954. The psychiatric interview. New York: W. W. Norton. Synnott, A. 1993. The body social: Symbolism, self, and society. London: Routledge. Throop, C. J., and D. W. Hollan, eds. 2008. Whatever happened to empathy? Special issue of Ethos 36: 385–489. Werner, O., G. M. Schoepfle, and J. Ahern. 1987. Systematic fieldwork. Newbury Park, CA: Sage. Whyte, W. F. 1960. Interviewing in field research. In Human organization research, ed. R. N. Adams and J. J. Preiss, 352–74. Homewood, IL: Dorsey Press. Whyte, W. F. 1984. Learning from the field: A guide from experience. Beverly Hills, CA: Sage. Wikan, U. 1990. Managing turbulent hearts: A Balinese formula for living. Chicago: University of Chicago Press. Yan, Y. 2009. The individualization of Chinese society. New York: Berg. Zigon, J. 2008. Morality: An anthropological perspective. New York: Berg.

14_107-Bernard.indb 342

6/9/14 10:03 AM

CHAPTER 11

Structured Interviewing and Questionnaire Construction Susan C. Weller

Most of what we know about what people think and do comes from interviews and questionnaires. This chapter focuses on the development of interview materials for collecting direct informant-based information with interviews and questionnaires. It is organized by interview purpose and presents different approaches to interviewing and question formats within the context of study goals. A mixed methods approach is recommended, beginning with open-ended questions in an exploratory or pilot phase and then integrating those results into a second phase using structured or systematic interviewing techniques and questionnaires. This two-step process is widely used across the social sciences in the development of interview materials. The initial stage of any study should include a descriptive exploration of the topic under study. A variety of strategies are available for conducting semistructured individual or group interviews. In general, the less that is known about an area, the more appropriate unstructured, open-ended interviewing methods are. For new areas of investigation, the goal is to develop questions and materials relevant to the area of inquiry and the people being studied. If an existing questionnaire or scale is to be used, especially if it will be used on a new population, then the initial interviews serve to verify that the questions and content are appropriate for the new population. The initial phase, then, focuses on eliciting relevant themes, questions, and responses for further study. A productive technique for doing this is the free-listing interview. The second stage incorporates those results into the development or modification of structured interview materials for a more systematic and detailed examination of the topic and responses across people. In anthropology, the first phase can be quite lengthy, as the purpose is often to explore topics in a new population, a new setting, and a new language. Descriptive information may then be used to frame a study on cultural beliefs or behaviors. In psychology, an initial phase of interviewing may be used to generate items for a new scale or to modify existing scale items (e.g., questionnaires) for use on a new population. In sociology, large surveys begin with a pilot or preliminary phase of interviewing to test clarity, comprehension, and question content. The combination of an initial descriptive exploratory phase followed by a systematic, structured phase produces a study much superior to one based on either method alone, but it also involves a greater commitment of time and energy. Projects relying solely on either responses to open-ended questions or on responses to a series of agreement rating scales can be biased and inaccurate. Responses to open-ended questions are limited by memory bias: People can recall fewer items (reasons, cases, etc.) than 343

14_107-Bernard.indb 343

6/9/14 10:03 AM

344

Susan C. Weller

they can recognize when presented with a complete listing of relevant items. This means that a spontaneous, unstructured request for information, while retrieving important information, may not retrieve complete information. When someone fails to mention a particular item, the item may not be important or it simply may have been forgotten. Also, there are different response styles that result in different amounts of information per person; some people provide long, detailed answers while others give short ones. Further, the use of different prompts and probes in response to individuals’ responses effectively changes the questions and makes it extremely difficult to compare responses to open-ended questions across individuals. Use of a structured format with the same set of questions and responses for all respondents produces comparable information across people and facilitates detailed comparisons across individuals and groups. If the questions or responses, however, are researcher generated and are not preceded by descriptive interviewing to verify relevance and wording, the interview may focus on items of interest to the researcher and may misrepresent or entirely miss topics of importance to the informants. A preferable approach is to combine both qualitative and systematic interviewing, taking advantage of the strengths of each: using open-ended questions to explore a topic and develop an understanding of relevant themes, questions, and responses and then using a structured interview to collect systematic data with those themes, questions, and responses. After a descriptive or qualitative phase elicits relevant themes, structured interview materials or questionnaires can be developed to examine knowledge, attitudes, beliefs, and behaviors about those themes. A variety of question formats are available. For example, most interviews contain general information questions requesting sociodemographic information from the respondent. These questions can be constructed in a variety of formats (close-ended, multiple choice, or open-ended) and are designed to collect specific information like age, gender, years of education, racial/ethnic identity, religious affiliation, number of children, and the like. Questions may also ask about behaviors (“In the last year, how many times did you visit a doctor?”) or relationships (“Name the people with whom you have discussed important personal matters during the past six months.”). These types of questions request information about the respondent or about people in his or her social network. Questions can assess knowledge. Knowledge tests evaluate the degree to which an individual or group possesses knowledge about a particular topic. Tests may be constructed with multiple choice, true/false, or open-ended questions. A specific assumption of a knowledge test is that the correct answer to each question is known, so respondents’ answers are scored as correct/incorrect in relation to that standard. Questions can also assess attitudes. Attitudinal scales attempt to measure the degree that an individual demonstrates or possesses a specific predefined construct that is usually psychological, such as authoritarianism, acculturation, or depression. The most common format for attitudinal questions is a series of statements, typically with a rating scale for each, where respondents are asked to rate their relative agreement with or the frequency of each statement. Similar to knowledge tests, responses are “scored” with reference to an a priori defined standard or criterion and then combined across statements to create a single score or scale of the construct.

14_107-Bernard.indb 344

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

345

Another type of study explores respondents’ classification of a set of items and describes the categories or dimensions used by people to discriminate among items in a set. Classification studies try to uncover respondents’ own dimensions of discrimination rather than access their adherence to a priori defined dimensions. Informants are asked to compare items in terms of their similarity, without reference to any specific dimensions or criteria. Formats appropriate for the collection of similarity data include: pile-sorting tasks, where respondents are asked to sort items into piles according to their similarity; and paired or triadic comparisons of items, where respondents judge the similarity of pairs of items. Finally, the purpose of a study may be to describe the beliefs of a group of respondents. While a classification study focuses on respondents’ beliefs (e.g., how they divide up the world into sets and subsets), beliefs may be studied in greater depth by administering a series of related questions on a single topic. For example, a series of questions might ask about specific attributes or assertions, like possible effects of global warming. Question formats differ from those appropriate for classification studies and include: open-ended, multiple-choice, ordering or ranking, and interval or frequency estimate type questions. Classification and belief studies depart meaningfully from knowledge and attitudinal studies in the handling of informants’ responses. In knowledge and attitudinal studies, responses are recoded or scored against a predetermined standard; in classification and belief studies, responses are not recoded, transformed, or scored. Thus, while many formats are applicable across a variety of study purposes, not all formats lend themselves to every purpose. PHASE I: EXPLORATORY INTERVIEWS: GETTING INFORMATION TO DEVELOP STRUCTURED INTERVIEW MATERIALS

The first phase of a project should be dedicated to gaining a broad understanding of the area of study. Without general background knowledge, it is impossible to know what questions are appropriate. So, depending on how familiar you are with the topic and the population you intend to study, a project begins with unstructured and semistructured interviews and progresses to more structured interviews. Initial interviews may explore a topic in general to gain a broad understanding of the topic and terminology. The first step in this initial phase of interviewing, however, focuses on learning whether or not the topic of study is relevant to the population and discovering the “right” questions to ask. The second step focuses on eliciting more substantive information. Then, elicited information may be used to develop new interview materials or to check the appropriateness of existing materials. The elicitation of items, statements, and themes relevant to the topic of study is the focus of the initial stage of interviewing, whether interviews are conducted individually or in groups. The set of items is sometimes called a semantic or cultural domain. A domain is a set of related themes, concepts, or statements on a single topic. For this purpose, items are elicited from informants in their own words. Without such elicitation of items directly from informants, items may reflect ideas of the researcher and not the informants. Convenience or purposive sampling is often used in this phase of a study, where a small group of people are selected for interviewing based on characteristics that are important to the study (Johnson 1990; see also Teddlie and Tashakkori 2009).

14_107-Bernard.indb 345

6/9/14 10:03 AM

346

Susan C. Weller

ASKING THE RIGHT QUESTIONS

The first step is to find out what questions to ask. What are meaningful and productive questions? If you are new to the topic, the people, and maybe even the language, one of the best sources on getting started is Spradley’s (1979) book, The Ethnographic Interview. Although more than 30 years old, this book is still one of the best sources on getting started with ethnographic or qualitative interviewing precisely because Spradley begins at the beginning by discussing how to find an informant and what to ask. Informants should be selected according to the purpose of the study and should come from the target population. If the study is about urban gardeners, then initial interviews should be conducted with urban gardeners. In this formative stage, often only a half-dozen people are necessary. Informants should be accessible and have time to sit and talk. An informant should be someone with expertise on the topic, someone with at least a year of full-time experience and who is currently involved in the activity or topic. Initially, the investigator’s role is as a “student,” to learn enough about a topic to ask reasonable questions about it. Grand tour, mini-tour, and taxonomic questions help you understand what is relevant to your informants and the terminology and organization of the domain. Grand tour questions (Spradley 1979, 86–88) are very productive in starting interviews and learning about a topic by providing an overview. Grand tour questions are general questions that ask for a description of a place, a process, or a typical day. “Could you describe the inside of the jail for me?” “Can you tell me all the things that happen when you get arrested for being drunk, from the first moment you encounter the police, to going to court and being sentenced, until you finally get out of jail?” (Spradley 1979, 86). What you are asking for is a review of something, allowing informants to talk about whatever they want; you will hear about things of importance to the informants as they tell you their impression of how things are organized. Mini-tour questions (Spradley 1979, 88) concentrate on unpacking meaning from smaller or more specific activities. Often embedded within a longer, more general description are smaller experiential units and processes. Similar to the grand tour overview questions, mini-tour questions ask for descriptions of these smaller events: “You said that a table of guys gave you a hard time last night, Can you give me an example of someone giving you a hard time?” (Spradley 1979, 88). Taxonomic questions (Spradley 1979, 132–54) may be used to elicit an entire taxonomy from one or more informants. A taxonomy is a structure of set and subset relations among domain items. General questioning like “What kinds of ____s are there?” with comparative and contrastive questions like “Is ___ a kind of ___?” can be used to construct a taxonomy of domain items. Taxonomic relations can distinguish relevant categories of kinds of things, attributes, functions, causes, and examples (Spradley 1979, 110). This type of interviewing is excellent for mapping out terminology (especially in a new language or with a new population) and gaining an understanding of the interrelations among items. It is a logical process of interviewing, developed from observing courtroom cross-examinations and can be seen in the early work in this area (Conklin 1969; Frake 1964; Meztger and Williams 1966; see also D’Andrade 1995). Ground-breaking work by Berlin et al. (1968, 1973) detailed indigenous knowledge of plants, and Berlin and Kay (1969) described color terms primarily with taxonomic

14_107-Bernard.indb 346

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

347

interviewing techniques. Taxonomic interviewing was popular in the 1960s and 1970s and was often relied on as a primary data collection technique to describe structure within a semantic or cultural domain. Few people today consider this a primary data collection technique, but it is a very valuable way to quickly get into and understand terms and relations from an informant’s perspective.

Item Generation: Domain Definition Given that you now have a topic that is relevant to the informants you intend to study and know enough about the topic to ask reasonable questions about it, you are ready to actually begin eliciting items. The goal is to elicit a set of related words, statements, or themes relevant to the domain you wish to study. The emphasis here is to obtain the items directly from informants. While an item pool can be created by simply sitting down and writing a series of statements, it is best to elicit items from informants and write statements from those items. Free listing is a productive technique to elicit terms as the goal is to get an exhaustive list of themes from each person, thereby reducing the number of necessary interviews. Themes may come from individual or group interviews or from other sources, such as from narratives and case reports. FREE LISTING

In free listing, an open-ended question is used to obtain a list or set of items from each informant. (What kinds of ______s are there? Name all the _____s you know.) For the study of social networks, a question might focus on listing all the network members (Name all your friends.) or listing all the type of relationships or exchanges that you can have with people in your network. The goal is to have a comprehensive sample of items by getting an exhaustive list from each person. Responses should be at the same level of contrast, without any set-subset relationships among items. While taxonomic interviews map out terminology and relationships among sets and subsets of items more broadly, free listing focuses on a single level of contrast. If terms indicate a class of items, then the class should be explored and specific items listed. Free listing can help define the set of domain items and its boundaries. It can be used to discover descriptive terms for Alzheimer’s disease (Karlawish et al. 2011) or genitalia (Cain et al. 2011). Some fields rely heavily on free listing to generate inventories of things (e.g., in ethnobotany, free listing is used to generate inventories of plants and to assess plant knowledge) (Canales et al. 2005; Mathez-Stiefel and Vandebroek 2012; Miranda et al. 2007; Schunko and Vogl 2010; Vogl et al. 2004). Listing can also be used to elicit perceptions of other environmental features (Mathevet et al. 2011). Some areas or topics are so clearly defined that a single question can elicit domain items. These domains are easy to identify because of the ease with which informants can produce a list of items. For example, Henley (1969, 177) asked a sample of 21 students to list (in 10 minutes) all the animals that they knew. Individual lists ranged from 21 to 110 animals, and the median number of animals listed was 55. Weller (1984) asked 20 women in the United States and Guatemala “to name all the illnesses or expressions for being sick that they could think of.” These are clear, unambiguous requests that generate many items. When responses were tabulated to see the number of people who named each item, many items were named by a majority of the sample and some were

14_107-Bernard.indb 347

6/9/14 10:03 AM

348

Susan C. Weller

named by only one or two people. A total of 423 animals were listed: Four were named by over 90% of the sample and 175 were named only once. Cancer, the most frequently mentioned illness in the U.S. sample, was mentioned by 75% (15/20) of the sample; 6 items were mentioned by 50% or more of the sample; and 30 items were mentioned by at least 15% (3/20) of the sample. Because salient themes and items tend to be mentioned by more people and mentioned earlier in individual lists (Bousfield and Barclay 1950; Friendly 1977), further study of the animal and illness terms focused on using items easily recognized and omitting items mentioned by only a few people. FREE LISTING WITH MULTIPLE, RELATED QUESTIONS

A series of related listing questions may be used to elicit exhaustive lists from individuals. The series of questions may be perceived by some informants as being all the same, but others respond differently to each question and provide detailed responses to some questions and not to others. In a study of women’s preferences for different infant feeding methods (Weller and Dungy 1986), a series of questions was used to try and tap the set of reasons that might influence a woman’s decision to breast- or bottle-feed their infant. Multiple questions were asked of each informant, to capture the positive and negative aspects of each feeding method. In all, women were asked 18 free-listing questions to elicit lists of reasons for choosing either breast- or bottle-feeding, but all 18 questions tapped into the single domain of characteristics of infant feeding methods: • • • • • • • • •

Please tell me the reasons why you want to breast-feed. Why do you think some people breast-feed? Why did you decide not to bottle-feed? What are the advantages of breast-feeding? What are the disadvantages of breast-feeding? What are all the things you like about breast-feeding? What are all the things you dislike about breast-feeding? When is breast-feeding appropriate? In what situations would you not want to breast-feed?

(Each question was repeated substituting bottle-feeding for breast-feeding.) FREE LISTING WITH CONTRASTING QUESTIONS

A related format, that also uses multiple questions, is the use of contrasting questions. Here, items are compared (in pairs) and informants are asked about the distinguishing features. Young (1980) used this format in studying choices for health care. To elicit reasons for choosing a particular health care source, he asked informants why they might go to a doctor and not a pharmacist, why/when they would consult a pharmacist and not a doctor. The anchored comparison helps elicit more detailed information than the general question of “why/when would you go to a doctor?” or “why/when would you go to a pharmacist?” Another study elicited descriptive attributes of social success, by free listing positive and negative attributes (Freeman et al. 1981; Romney, Smith et al. 1979). For one subsample, informants were asked to name people who they thought were successful and to describe each one; then they were asked to think of people who were failures and to describe them. For another subsample, they were

14_107-Bernard.indb 348

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

349

asked to think of five friends or acquaintances and to describe all the ways that each was successful and then all the ways that each was a failure. SUCCESSIVE FREE LISTING (LINKED LISTS)

Related lists of items can also be elicited by asking for an exhaustive list of items in one domain; and then using those responses to asking for a new list in a related domain. This linked-listing task has been called “successive free listing” (Ryan et al. 2000). In a study of adolescent behaviors and possible punishments, interviews with Anglo and Hispanic adolescents explored adolescent “misbehaviors” and “adult disciplinary responses” (Weller et al. 1987). Verbatim responses of 29 Anglo and 27 Hispanic adolescents (with approximately equal numbers of males and females) were recorded. Each interview took one to two hours to complete and consisted of open-ended and free-listing type questions, descriptive answers, and probes by interviewers to seek further explanations. The following issues were explored: 1. “What things do you (or other teenagers) do that make your parents/mother/father/ adults, etc., angry?” 2. For each response to the previous question: • “When you do _____, what do your parents, etc. do?” • “What other things might be likely to make adults upset or angry?” • “And if _____makes adults/etc. angry, what might they do in response?”

The purpose here was to elicit an exhaustive a list for each informant for each question, so the question was changed slightly and asked again as informants exhausted their list. First, questions focused on eliciting teen misbehaviors. Multiple questions were used as probes: “What things do you do that make your parents angry?” “What things do other teenagers do that make their parents angry?” Then a second domain of adult behaviors was also elicited, linked to the responses given to the first question(s): “When you do _____, what do your parents do?” “And if _____ makes adults angry, what might they do in response?” Information on the second domain was requested listing possible responses after listing all misbehaviors. Thus, two related lists were elicited: the set of things teenagers do and the set of things adults do in response. Responses were tabulated across all 56 adolescents for each of the two domains. OBTAINING ITEMS FROM OTHER SOURCES

Lists of items generated by informants also can be supplemented with items from other sources. In a study of possible cultural differences in the definition of punishment and child abuse, punishment items listed by Anglo and Hispanic adolescents were supplemented with physical abuse descriptions from a hospital emergency department (Weller et al. 1987). Because extreme forms of punishment and abuse are infrequent and would not be expected to be reported in a small sample, a list of the most frequently reported forms of physical abuse was obtained from hospital emergency room records and was incorporated into the final list of items. Similarly, items can come from published sources, from existing questionnaires and scales, or from participant observation. To create a taxonomy of factors relating to disclosure of medical errors,

14_107-Bernard.indb 349

6/9/14 10:03 AM

350

Susan C. Weller

Kaldjian et al. (2006) obtained examples primarily from the medical literature and supplemented that set with examples from interviews. LIST LENGTH

Informants should be able to generate lists of about a dozen items. List length may be affected by context (Miranda et al. 2007) and expertise (Hutchinson 1983). If lists are short, try probing more. Prompts and probes can elicit more items. Avoid asking questions that can be answered with a “yes” or “no.” Thus, rather than asking “Are there any more ___s?” Ask “What other kinds of ____s are there?” This nonspecific prompt can help elicit more items and also helps your informant understand that you want an exhaustive list. Another prompt is to repeat what has already been listed: “You said that ____ and ____ are kinds of ____s. What other kinds of ____s are there?” Here, you remind the informant what he or she was thinking and convey the message that there are more items. The main question can be repeated in a slightly different way, as with the multiple questions about infant feeding methods, adolescent behaviors, and health care sources. Brewer (2002) compared the effectiveness of (1) a nonspecific probe; (2) repeating listed items; and (3) reading back each item and asking the informant to think about the item and other items that are similar to it. All three methods produced longer lists, but the third technique increased the list length by almost 50%. If such probes fail to generate richer lists, you might try a different format for the focus of the question, by using multiple or contrasting questions, or by using an altogether different focus. It is possible that the set may exist in your mind (the researcher), but not necessarily in the minds of the informants. RECORDING RESPONSES

Responses should be recorded verbatim. The point of generating items from the informants is to discover their definition of items in their language (verbatim). All ambiguous phrases and thoughts, however, should be clarified. The interviewer should probe and seek to determine explicitly what is meant: “What do you mean by _____?” Or, “Can you tell me more about that?” The goal is to elicit statements or themes that are clear so that only one meaning is conveyed (e.g., if a statement is repeated to someone not present at the interview, they should understand the exact meaning intended by the informant). An example of this in the infant-feeding study was that some women stated that they had chosen breast-feeding because it was convenient. Others stated that they had chosen bottle-feeding because it was convenient. Further probing in each of these situations revealed that the breast-feeders meant that they could feed their infant without having to prepare or clean bottles and the bottle-feeders meant that they could feed their baby anywhere without the embarrassment of exposing their breasts. Thus, the latter full statements more clearly express the reasons for choosing a particular feeding method. It is not sufficient, then, to use a general theme, such as convenience, when that theme has more than one meaning. A goal in recording responses is to be sure that you have captured the essence or the underlying meaning in the informants’ own words, as much as possible, so that you may use specific statements, phrases, and idioms in subsequent interviews with the caveat that the exact meaning is understood.

14_107-Bernard.indb 350

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

351

SUMMARIZING RESPONSES

Unique, verbatim answers or themes are tabulated across respondents. Domains may be defined with the use of single questions, multiple questions, contrasting questions, linked listing questions, and sometimes with supplementary items from other sources. Answers are then tabulated by informant and not by question. This is especially important when using multiple questions to elicit items, so that when someone mentions something more than once, it is counted only once, for that informant. The final tabulated list should reflect the number of unique people who mentioned each item. Final statements should be in clear language with consistent syntax. Statements should convey the same meaning to each and every reader. In the infant-feeding study, the 18 most frequently mentioned themes from the English-speaking Anglo and the Spanish-speaking Hispanic lists were chosen for study and changed to a neutral form “A way to feed your baby that. . . .” The items were also balanced so that half of the items referred to breast-feeding and half to bottle-feeding and half contained positive attributes and half were negative. Although the list contained a culled and modified set of the multitude of statements collected, language and ideas remained concordant with those in the original interviews. Tabulation of responses helps provide a sense of the relative salience for the themes across people. SAMPLE SIZE

The necessary sample size for free-listing interviews is a function of variation. This is true for both qualitative and quantitative research. The less variation (e.g., the more consistent the responses are) across people, the smaller the necessary sample size. For some domains, a sample size of 10 may be sufficient and for other domains, or for increased accuracy, sample sizes of 50 or more may be needed. Typically, a sample of about 20 informants is adequate, especially with a good list length per person. As the number of interviewed informants increases, say in increments of five; from 5 to 10, 10 to 15, and so forth, there will reach a point where little new information is added to the content and order of tabulated items. This is sometimes referred to as the point of saturation. Thus, the sample size is adequate when the addition of new people or groups does not alter the frequency distribution of items and few new items are added. By getting a list of items from each informant, more information is obtained per informant and fewer informants are needed, and saturation is reached more quickly. With a meaningful question and probing, each informant should be able to generate a list of at least 6 things, usually around 10 to 12, and often many more. Agreement on items, statements, or themes is estimated simply by counting the number of informants that mentioned each. The set or domain is defined by the items mentioned by multiple informants. The most frequently mentioned items are the most salient items. Psychologists have shown that the most salient items will be named by more people and those items will appear higher up in individual lists (Bousfield and Barclay 1950; Friendly 1977). Salience of items is estimated most simply with the frequency distribution (e.g., the percent of the sample that named each item) and can be used to make comparisons between samples (Ross and Medin 2005; Thompson and Juan 2006). Sometimes salience is estimated with a consensus analysis to identify items mentioned by a majority of the sample (Mathevet et al. 2011). While the set of

14_107-Bernard.indb 351

6/9/14 10:03 AM

352

Susan C. Weller

items obtained with free-recall listing is not necessarily definitive, it should nevertheless capture most well-recognized items. GROUP INTERVIEWS

Free-listing interviews may be conducted with individuals or groups (focus groups). An important thing to remember, however, is that the sample size for group interviews is not the number of participants, but is closer to the number of groups. Lists generated from group interviews do not reflect the thoughts of each individual in the group. Rather, the interviews reflect the group’s thoughts and thus only one list is generated per group. Individual interviews are much more productive than group interviews in terms of generating ideas. Group interviews generate only about 60% as many topics as do individual interviews (Fern 1982; Morgan 1996). Larger groups are more productive than smaller groups, so one group of eight people is preferable to a group of four; but more groups are better, so two groups of four each are better than one group of eight (Fern 1982). Saturation for group interviews often occurs with four to six groups of eight people each (Morgan 1996). It is important to note the total amount of time invested in interviewing: 20 individual free-list interviews that average 45 minutes each results in 15 total contact hours of interviewing, four groups of eight people typically result in 4 to 8 total contact hours of interviewing, and eight groups of four would have 8 to 16 hours of interviewing. Production of ideas and differences between methods may also be a function of the time invested in interviewing. NARRATIVES, CASE HISTORIES, AND TEXTUAL MATERIAL

Another approach to gaining an understanding of a topic or domain is to collect descriptive accounts, like narratives or case histories. Themes can be identified in textual materials in phrases and ideas that are discussed, repeated, labeled as categories of things, or used as metaphors (Ryan and Bernard 2003). Quinn (1987) culled themes relevant to American beliefs about marriage based on informants’ descriptions of marriage. Chavez et al. (1995) recorded women’s descriptions of possible causes of cancer and used recurring themes for further study. Kempton et al. (1995) also began their systematic study of U.S. environmental beliefs by collecting descriptive narratives and identifying themes from the descriptions. Johnson and Griffith (1996) conducted detailed interviews about pollution, its causes, sea life that is affected, and seafood; and then selected themes from the transcripts for further study. Analysis of textual materials can only suggest possible interconnections and relationships among themes. Unstructured methods of interviewing and response narratives are excellent for suggesting hypotheses, but more systematic data are needed to test the validity of observations and to make comparisons across groups. Personal case histories sometimes yield more detail on a single case, but typically require a larger sample size (more people and more cases) to cover the breadth of cases. A detailed history of the last illness case that occurred in the household collects information on only one case of one illness, and it is difficult to get case information on rare events. In contrast, interviews with individuals about “all the illnesses they know” can uncover information on a variety of illnesses.

14_107-Bernard.indb 352

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

353

PHASE II: STRUCTURED INTERVIEWING TECHNIQUES AND QUESTIONNAIRE CONSTRUCTION

After establishing the items and content for study, a more structured interview format can be pursued. Open-ended, semi-structured formats facilitate the collection of new information with the flexibility to explore topics in-depth with informants. Meaningful comparisons across people may not be possible, however, because informants have been encouraged to discuss different items and so have not really been asked the “same” questions. Structured formats allow the investigator to make more detailed comparisons across people and groups and can verify impressions from lessstructured interview methods. This section describes a variety of question formats. The focus is on designing interview materials (questions, tests, and tasks) appropriate for the goal of the study. Thus, the section is organized by study purpose: general information questions, knowledge tests, attitude scales, classification studies, and assessment of cultural beliefs.

General Information Questions Most studies include questions to collect information on respondents’ sociodemographic characteristics. Questions are straightforward requests for information: age, gender, ethnicity, household composition, length of residence, and sometimes behaviors. These questions parallel those found in surveys. The term “survey,” however, is often used to refer to a combination of methodologies: the selection of respondents, method of interviewing, and questionnaire design (Fowler 2009). The selection of respondents usually focuses on procedures for selecting a random or representative sample. When a representative sample of respondents is used, results may be generalized from the sample to a larger population. Nonrandom or convenience samples can provide useful information, but generalization of findings should be done with great caution. The method of interviewing concerns whether interviews are conducted in person, on the phone, or by mail. In-person or face-to-face interviews may be administered by an interviewer or be self-administered and tend to have the highest participation rates. Phone interviews can only be administered by an interviewer, but may be computer assisted by having the questionnaire on a computer. With computer-assisted telephone interviews (known as CATI in the sociological literature), the interviewer enters responses directly into a computer. Mail, email, and web-based questionnaires must be self-administered. More complex responses can be obtained in face-to-face interviews with the use of visual aids, if necessary. Questions and responses must be simplified for oral/phone presentation. Self-administered, open-ended questions usually do not produce useful information due to the lack of probing for clarification. Participation rates for the three different approaches parallel their costs. In general, face-to-face interviews have the highest participation rates and are the most expensive. Phone and mail methods tend to be less expensive but also have lower rates of participation. As follow-up procedures (call backs and re-mailings) are intensified, phone and mail participation rates (and costs) increase. Self-administered questionnaires in mail, email, and web-based sources tend to have the lowest participation rates. Participation rates below 75% should be examined critically as the sample may

14_107-Bernard.indb 353

6/9/14 10:03 AM

354

Susan C. Weller

no longer be representative and may be biased. It may be preferable to interview a small representative sample on the phone or in face-to-face interviews than to send out a great number of self-administered questionnaires in regular mail or email and get a low participation rate. Here, the focus is on questionnaire construction and it is assumed that most interviews will be conducted in person. The biggest weakness in questionnaire design is often the result of an investigator that simply drafts a set of questions, assuming that anyone can write a questionnaire. The result is often a set of poorly worded questions with unclear response categories. Unclear questions lead to uninterpretable responses. Sociologists and psychologists have spent an enormous amount of time designing questionnaires, studying the effects of different wording and ordering of questions on responses as well as the interaction between interviewer and respondent. It is a waste of research effort not to take advantage of their experience and knowledge. Recommendations on wording and ordering of items can be found in the sociology literature. See, for example, Fowler’s (1995) Improving Survey Questions for a very good short, focused description; Bradburn et al.’s (2004) Asking Questions is a more complete overall reference; and Fink’s (2003) The Survey Kit is also a handy overview. It is certainly worth investing some time, even if only a few days, to review some of these materials. Question formats include: open-ended, close-ended multiple choice, and rating scales. Open-ended questions should be simple and seek clear, short answers. Questions should be written as complete questions, so they are asked in the same way for each person. For example, instead of just having “Age ___” on the questionnaire, it is preferable to have “How old are you?” or “What is your date of birth?” Close-ended questions should be concise with a complete listing of mutually exclusive response categories. Rating scales are usually appropriate only for literate informants with a moderate degree of education, although they may be simplified and asked in an oral interview (Weller and Romney 1988). In general, questions should proceed from broad, general requests for information to questions requesting specific or more detailed information. This is done so that questions requesting detailed information do not bias the responses for more general information. Similarly, less personal questions should precede those perceived to be more private or threatening. Questions requesting sociodemographic information may be asked initially, especially if they help establish whether or not the informant fits the study inclusion criteria. Other sociodemographic questions may be asked at the very end of the interview, especially those adding extra information and those thought to be more personal or threatening, as with questions in the United States regarding income. Inclusion and exclusion criteria for participants are established as part of the study design or protocol. They are the explicit conditions for including or excluding someone in the study. If you want to study “Latina” women, then before interviewing anyone, you should define who is and who is not a Latina woman. Thus, the initial questions may seek to establish the informant’s gender, ethnicity (by self-report and also possibly by birthplace and language preference), and age (in years or parental status). The advantage of having all inclusion and exclusion criteria-related questions first is that an interview may be terminated quickly for people who do not meet study criteria. It is

14_107-Bernard.indb 354

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

355

advantageous to collect some information on everyone, even the excluded individuals, to see if there are differences between those who do and do not choose to participate. Only questions relevant to the study should be included in the interview. Each question should link directly or indirectly to the purpose of the study. Questions should concern factors implicated by theory, factors mentioned in the literature, and factors that might potentially affect results. Too often, extraneous questions are included without considering how responses will be handled. For example, a question on marital status might be included, but if the real interest is whether a woman is living with the father of her child, then a direct question to that effect would provide more useful information. Still, it’s best to ask too many rather than too few questions: A question or answer can always be ignored after it is collected, but it’s usually difficult or impossible to go back and ask a question that was omitted inadvertently. If you want to know how your sample results compare with those from a larger population, use questions from large or national surveys. Not only can you compare responses with those of the larger survey, but you can take advantage of the time and effort that went into the development and wording of the questions. Even simple questions can be borrowed directly from such surveys. Also, you can compare different sets of questions purported to measure the same thing. For example, questions about ethnicity can come from multiple sources: You can ask about ethnicity using the questions and categories used in a national census and also from questions you have developed that you believe are more appropriate indicators. Using census categories allows you to discuss your results in terms of national categories and to compare your findings with other reports. Using a new series of questions in conjunction with census questions allows for a direct comparison of the two ways to define ethnicity. When beginning to design a questionnaire, take advantage of previous scholarly work and look around for published questions (and responses) and do not hesitate to use them. For example, in the United States, check the U.S. Census, the American Community Survey, the General Social Survey (also done in many countries around the world), the National Health Interview Survey, the National Crime Survey, and the Consumer Expenditure Survey. Also see the World Fertility Survey and the World Values Survey. When writing questions, keep the study’s purpose in mind. Translate the purpose into specific questions that will directly or indirectly provide information relevant to the purpose. Also, have a plan about how responses will be used to meet the study’s purpose. Good questions are ones that respondents understand, that all respondents interpret in the same way, and for which the respondents’ understanding is the same as the intended meaning. All questions should be administered in the same way to all respondents. The wording of questions should be clear and simple. Avoid ambiguity in meaning and define terms if necessary. Avoid compound questions with more than one concept embedded in a question. It is preferable to use multiple, related, and simple questions than ask complicated or long questions. It is important to ask things that informants know about and can answer meaningfully. Questions like “What kind of health insurance do you have?” may reveal that people know whether they have health insurance and maybe the company, but they simply may not know much about the actual coverage. And people cannot answer complex questions, like “What proportion of your time

14_107-Bernard.indb 355

6/9/14 10:03 AM

356

Susan C. Weller

is spent doing ___?” To answer involves an estimate of time spent on different activities and then divided by the total time. Instead ask: “Have you done ___ in the last month?” Minimize the difficulty in answering. Requests for information for a shorter, more recent time period rather than a longer period of time tend to get more accurate answers. The U.S. National Crime Survey asks about experiences over the past six months, and interviews about illness episodes typically ask about experiences in the past two weeks. Bias in responses tends to be toward what people usually do and not what they did on a specific occasion or time period. Respondents also tend to “telescope”: When asked about behaviors during a specific time period, they report actions from a longer period of time. If asked if someone went to the dentist in the past year, people tend to say yes if they visited the dentist within the past two years. To improve recall on behaviors during a specified time period, it is important to mark the time period with an important event and to ask several questions about the behavior. For example, rather than asking, “Has anyone in the household been ill in the past three months?” ask instead, “Since Easter/Holy week, has anyone been ill in the household?” or, “Since our last visit, has anyone been ill in the household?” (Weller et al. 1997). Alternatively, ask this as a series of questions. If you are interested in illnesses during the past week, begin by asking about illnesses over the past year, then in the past three months, and then in the past two weeks. Multiple questions signal that the question is important and can improve accuracy (Fowler 1995). To minimize problems in reporting accuracy, clarify the goal(s) of the study with respondents. Emphasize that there are no right or wrong answers; that responses are confidential and anonymous; and that providing accurate information is important. Inasmuch as possible, the interviewer should be matched to the respondent by gender (men interview men, women interview women) and background (similar ethnicity and SES). Where possible, borrow questions from national surveys. Avoid ambiguous words and complicated concepts, ask simple, straightforward questions. Make questions easy to answer. Give help with recall over a specific time period by marking the period with specific, memorable events and use multiple questions to improve accuracy. Responses to multiple, related questions can be combined to form an index or scale. COMBINING RESPONSES TO CREATE SCALES AND INDICES

As requested information becomes more abstract (i.e., as questions move from simple ideas like gender and age to more complex ideas such as social class), more questions are needed to get a reliable estimate of the concept. For concepts that cannot be measured simply or directly, use proxy questions to get information associated with or indicative of the underlying concept. Then combine the responses to obtain a more reliable and accurate estimate. For example, we believe that social class or socioeconomic status exists, even though there is no direct, single question or ruler by which we can assess or categorize an individual or household. In developed countries, we often use combinations of educational level, income, and occupation to estimate socioeconomic status (see Haug 1977). In less-developed countries and among populations with little variation in occupation, education, and

14_107-Bernard.indb 356

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

357

income, such variables may not be helpful in differentiating social strata. In less developed and rural areas, it’s more helpful to ask a series of questions related to socioeconomic status (e.g., questions about house construction materials, water source, ownership of material goods) and to combine responses to differentiate households. A summative score of responses to a series of questions creates an index or scale. A summative score should be valid; it should measure the idea or construct that it is intended to measure. Content validity focuses on the construction of a scale: Are the items reasonable and do they appear to measure the same thing? Criterion or predictive validity is the degree to which a scale predicts the idea or construct it is trying to measure. Construct validity goes a step further for scales with reasonable content and internal consistency; construct validity concerns the association between a scale and other measures theoretically related (but not necessarily the same construct) to the construct that the scale is attempting to measure. First, the choice of reasonable questions and proxy variables helps ensure that a combination of responses to those questions will also be reasonable. Second, items selected for combination in a scale should be “scalable” (i.e., they should be positively correlated) with internal consistency and good reliability. A principal components analysis can indicate how to optimally combine variables that are in different units of measurement. A principal components analysis clusters items into groups according to their inter-correlations; items with the same pattern of responses across people (those that have the same pattern of high values and low values across people) are grouped together. Finally, the scale should correlate positively with similar scales and should correlate with other measures in ways predictable by theory. In developing a scale of financial resources in rural Guatemala, Weller et al. (1997) asked over two dozen questions about household composition, characteristics of the head of household (gender, age, education, ability to read, ability to write), house construction (walls, roof, and floor), and assets (ownership of land, appliances, vehicles, and animals). Some questions requested yes/no type responses: “Do you own your house?” “Do you have a bicycle?” Others requested the number of people or animals; and responses to multiple choice responses (household construction materials) were coded as present or absent. Seeking to develop a scale concordant with community perceptions, Weller et al. (1997) also asked three informants in six villages to rank 10 families according to their economic resources and retained only those questionnaire items that correlated with the community judgments (10 of the original 28 questions). A principal components analysis of those questions for the larger sample showed that variables most indicative of financial resources (including monthly income) grouped together on the first factor, and variables representing other dimensions of socioeconomic status (educational level and household size) grouped on successive factors. Weller et al. (1997) wanted a relatively simple scale that could be used in other studies in the region, so they used the principal components solution to identify which variables should be combined (those on the first factor), but not for a weighted combination of variables. To overcome the problem of different units of measure, variables were dichotomized (so they would be in the same units) and summed. Each household received a cumulative score (+1) for the presence of each indicator: monthly

14_107-Bernard.indb 357

6/9/14 10:03 AM

358

Susan C. Weller

income greater than the median; ownership of any appliance; more than two rooms in the house; non-dirt floor; more than three chickens; adobe, brick, or block walls (as opposed to bamboo, wood, or plastic); land ownership; and ownership of a bicycle. Summing across the eight variables created a nine point (0–8) scale. The final scale was concordant with other scales previously constructed to assess socioeconomic status in rural Guatemala (Freeman et al. 1977; Johnston et al. 1987). In fact, such scales are surprisingly similar across rural regions of the world and use indicators such as floor construction (dirt vs. other), type of cooking fuel, and availability of animals for sale. Guttman scaling is another way to combine household indicators of socioeconomic status. Guttman scaling reveals whether there is a cumulative and sequential ordering of variables: If someone has an item on the list, they would also tend to have the objects that precede the item. Similarly, if a household lacked an item, it would tend to lack subsequent items. Dewalt (1979, 106–15) described a nine-point “material style of life” Guttman scale from the presence or absence of eight variables: iron, radio, bed, cooking facilities off the floor, sewing machine, wardrobe, stove, and television. This means that responses indicated that if a household has a bed, they also had a radio and an iron. Dewalt checked the validity of the scale by comparing the final scale to informant ratings of wealth and found them highly correlated. Guttman scaling has been used to describe the acquisition of consumer goods in Polynesia (Kay 1964; Weller and Romney 1990, 79–83) and in the United States (Dickson et al. 1983; Kasulis et al. 1979). Guest (2000) presents a detailed example for Ecuadorian fishermen using 12 material goods. A related, alternative model for representing the order of acquisition is the Rasch model (Soutar and Cornish-Ward 1997). Guttman scaling can be used to represent a variety of cumulative activities and skills (e.g., social participation activities among the elderly [Bukov et al. 2002] and men’s skill in building and creating objects [Johnson 1995]). Responses can be combined across related questions or variables to create a single scale or index. Such indices are more reliable and accurate than use of a single question, especially when the request is for information more abstract than someone’s age, height, or weight. While the responses to simple questions may be combined to estimate the household socioeconomic status, a variety of other variables may be similarly combined to obtain better estimates of behaviors and experiences. For example, Handwerker (1996) used a combination of questions to better estimate household activities and experiences of violence and affection. SOCIAL NETWORK QUESTIONS

Social network studies focus on interrelationships among people and organizations. Questions on social networks tend to focus on two different approaches (Bernard 2012; Scott 2000; Wasserman and Faust 2009). One approach looks at personal or egocentered networks, where a respondent is asked about his or her relationships with others and the others may or may not know one another. The second approach looks at complete, whole-group (sociocentric) networks, where each person is asked about his or her relationships or interactions with every person in a group. With ego-centered networks, questions may seek information on the number and types of friends or shared activities and interests. These questions can measure quali-

14_107-Bernard.indb 358

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

359

tative attributes of relationships between people and can be used to estimate social support or social capital. A first step is to use open-ended and free-list interviews to explore types of people, relationships, and functions that are important, to form a meaningful context for subsequent questions, and to be able to ask about those relationships in a meaningful way. Second, in subsequent questions with a new sample, questions, would ask systematically about who might offer help, advice, or support in different scenarios (e.g., Burt 1984, 1986; Freeman and Danching 1997). Questions can also drill down and collect detailed information on the type and quality of relationships in the respondent’s personal network: for example, “Name 10 people who ___,” and then for each person named, ask about their characteristics and the relationships between the people to estimate information on the connectedness and density of the personal network (McCarty 2001). Studies of whole group networks focus on a defined group of people and ask about the frequency and/or quality of their interactions. Here, initial interviews first must list and define all people in the group. Then, each person can be asked about their relationship (advice seeking, exchanges, social interaction, etc.) with every other person in the group. Each person is presented with a list of all group members and asked to check the names of those they interact with the most, or rate them on rating scales on the frequency with which they interact or rank the entire list in terms of how often they interact. For example, Johnson et al. (2003) studied a work group network at the South Pole Research Station and had them rate one another on an 11-point rating scale indicating the frequency of social interaction. Although much more intensive, each person can also be asked about the relationships between everyone else in the group (Krackhardt 1987, 1990). Johnson and Orbach (2002) studied the network of people judged to be important in passing a particular piece of legislation (North Carolina state senators, cabinet-level secretaries, legislative committee chairs and co-chairs, staff, resources managers, lobbyists, and private citizens) and provide an example of how to collect data on a large group of people and minimize the response burden (length of the task) while doing so. CHALLENGES TO VALIDITY

Accuracy of responses can be compromised by questions that are interpreted differently by different respondents. Questions should be in complete, grammatically correct language, and read the same way to each person. One way to understand how informants interpret a question is to interview a small sample of individuals and ask them to think out loud; ask them to describe their interpretation of the question and to list possible answers (cognitive interviews, Fowler 1995). Another source of inaccurate responses is the informants’ own memory. Informants may report an event that actually happened 12 months ago as occurring 6 months ago. Marking a period with an important or widely recognized event (since ____ occurred) reduces this telescoping effect (Loftus and Marburger 1983). Informants also may “misremember” an event, reporting instead what they think happened or what usually happens. Informants are much better at telling you what they typically do, than what happened at a specific time. Freeman et al. (1987) asked a group of individuals about attendance at a group presentation the previous week. Errors consistently counted those

14_107-Bernard.indb 359

6/9/14 10:03 AM

360

Susan C. Weller

who usually were in attendance but were not there as being there, and counted those who usually were absent, but were there, as absent. Reporting errors tend to be biased in the direction of typical behaviors. Bernard et al. (1980; see also Bernard et al. 1984) found poor informant accuracy when people reported with whom they interacted for a specified period of time, but reports may have been biased toward more typical behaviors rather than what actually occurred in the specified time interval. When informants’ reports of social interactions were compared to observed interactions for a specific time period, studies with longer observation periods (a better sample of typical behaviors) tended to have better informant accuracy. D’Andrade (1974) found that coding of behaviors immediately after they occurred corresponded more to the similarity among adjectives rating the behaviors than to the behaviors that actually occurred. So, if someone was remembered as having smiled, they were more likely to be attributed with actions associated with smiling like having been facilitative, friendly, and so on, whether they were or not. One explanation for this bias is that people who smile are usually helpful and friendly. Accuracy of responses also may be affected by the interview itself. Contextual effects have long been documented and studied by sociologists and, generally, better responses are obtained when the interviewer and the informant share characteristics such as gender and ethnicity (Schuman and Presser 1996). An informant’s lack of experience with the interview process may decrease accuracy, and informants may offer socially desirable responses or may deliberately mislead you. Accuracy may be increased as participants understand the purpose of the interview and the degree of confidentiality in responding.

Knowledge Tests A knowledge test consists of a series of questions designed to test someone’s ability or knowledge. The answers—the correct answers—to the questions are known, and responses are scored or recoded as correct/incorrect. First, the content domain is established that covers the subject matter or ability to be tested. Then, test questions are drafted. Question format may be multiple choice (with two or more choices) or openended (requesting single word or short phrase answers). Performance of respondents is usually described as the percentage of correct responses (of the total number of questions) or as a percentile, comparing an individual’s performance with the distribution of scores across respondents. Just as sociologists have much expertise in writing general information questions, psychologists have extensive expertise in developing knowledge tests. Nunnally’s (1978) book, Psychometric Theory, presents a thorough review of issues involved in developing a test. It is important after drafting, administering, and scoring a test to also assess its reliability. An assessment of a test’s reliability and the resultant modification of the test can greatly improve a test’s ability to discriminate between knowledgeable and less knowledgeable informants. Reliability is the degree to which a variable or test obtains the same result when administered to the same people, under the same circumstances. A test with low reliability is analogous to a very sloppy measuring device; it may be valid, but it has a lot of measurement error. For example, if you measured the height of a sample of college undergraduates with a weight–height measuring device typically

14_107-Bernard.indb 360

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

361

found in a physician’s office and also with a 6-inch pocket ruler, you might find that the pocket-ruler estimates could conceivably contain measurement error large enough to mask the difference in average height between men and women. The more accurate the measuring device, the greater the ability to detect smaller differences. The same is true for tests. If a test can be streamlined and limited to questions that best differentiate degree of knowledge of the subject matter (thus, increasing the reliability), it can be a shorter, more accurate, and hence more powerful test. RELIABILITY

Reliability of a test can be assessed in a variety of ways. One way to assess reliability is to give the same test twice, after an interval of time, to the same individuals. Reliability, then, is estimated by the correlation between the two sets of scores. Because the Pearson correlation coefficient is used, reliability ranges from zero to one. This type of reliability, test-retest reliability, is limited in that scores may improve due to practice or learning effects and change can occur in the time interval. Alternatively, equivalent, but non-identical tests can be administered, but it’s difficult to develop “equivalent but non-identical” tests. A third approach is to create two tests by arbitrarily dividing one test in half and calculating separate scores for the odd-numbered and even-numbered items and administer the test once. This type of reliability, split-half reliability, is estimated by the correlation between the two sets of scores. The best overall estimate of reliability, because it subsumes the previous estimates, is provided by the reliability coefficient (Nunnally 1978). The reliability coefficient, sometimes called coefficient alpha or Chronbach’s alpha, is mathematically equivalent to calculating all possible split-half reliabilities and, while it may sound complex, it is widely available as an easily accessible option in most statistical software packages. For a test to have high reliability, all of the test questions must be on a single topic and be at the same general level of difficulty. This means that items should be positively intercorrelated, and performance on individual items should be concordant with the overall score. A test question would not be a good estimate of ability if the “best” or high scorers got it wrong and those with lower total scores tended to get it right. Such questions reduce the accuracy of the total score. An item analysis helps identify items that do and do not correlate positively with the total score. The item-to-total correlation for each question tells how well responses for each question parallel the total score. If the correlation is not positive, or is small (less than +.20), the question should be dropped (Nunnally 1978). Items considered for omission can be modified in future applications. Writing good questions with multiple choice answers is very difficult! The overall reliability of a test, the reliability coefficient alpha, is a function of the intercorrelation among the questions (the degree to which they measure the same concept) and the number of items (the more items on a single topic the more accurate the estimate): Reliability = k / (1 + (k − 1) ) where k is the number of questions and is the average Pearson correlation coefficient between all pairs of questions. Thus, a reliable test can be created with a few highly

14_107-Bernard.indb 361

6/9/14 10:03 AM

362

Susan C. Weller

correlated items or with a lengthy test of weakly related items. When dichotomous responses are analyzed, this formula is called Kuder-Richardson 20 (KR-20). The reliability coefficient and the performance of each item (in the item analysis) can readily be obtained in most major statistical packages. EXAMPLE

In a study in rural Guatemala, Ruebush et al. (1992) developed a test to assess local knowledge about the causes, symptoms, and treatment of malaria. Experience both with residents of the region and the National Malaria Service led to a draft questionnaire or test with 65 true/false items. Since the correct answers to the questions comprised the scientific or biomedical model of malaria transmission and treatment, an initial pilot test was a very simple one to see if National Malaria Service workers (those with more biomedical experience) scored higher than the rural residents. This involved a day’s worth of interviewing, in a single rural village, interviewing a half a dozen respondents and a few National Malaria Service workers. A quick tabulation of responses and scores, in the field, helped identify ambiguous questions with unclear answers. A revised version with 65 true/false questions was administered to a larger sample of residents and National Malaria Service workers. Responses, where 0 = no/false and 1 = yes/true, were compared to the correct answers and recoded to 1 if the answer was correct and to 0 if the answer was incorrect. A reliability analysis, in particular the item analysis, helped identify questions that did not perform well because they did not contribute to the total score. The 65-item test had a reliability coefficient of .82. The reliability analysis indicated that reliability could be improved by omitting items with low item-to-total score correlations. The omission of 25 items created a 40-item test with a reliability coefficient of .91. Thus, the shorter version of the test had better discriminatory ability, and comparisons between groups could be made with greater precision. This procedure is also used in identifying poor test questions on multiple choice exams for large college classes. Scores from knowledge tests indicate how well people know the correct answers. In the above example, the answers constituted the scientific or biomedical model of malaria, but the scores did not indicate whether wrong answers were due to a lack of knowledge or to different beliefs. In the malaria study, Ruebush et al. (1992) also analyzed responses in their original form without coding them as correct/incorrect and used the modal response for each question as an estimate for local beliefs regarding the answers. Cultural beliefs can then be compared to the scientific answers used to score the knowledge test to identify areas where errors might be due to differences in beliefs. Trotter et al. (1999) conducted a detailed study of Latino AIDS beliefs and compared the results to performance on the national AIDS knowledge test. They found that Latino errors on knowledge tests were most likely not due to different cultural beliefs (see section below on Exploration of Specific Beliefs).

Attitude Scales and Tests Similar to knowledge tests, attitudinal scales or tests measure the degree to which individuals and groups possess specific constructs. (A construct is an a priori defined concept.) Development of attitudinal scales begins by defining the domain of items

14_107-Bernard.indb 362

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

363

relevant to the particular attitude being studied. Statements are generated that describe or are indicative of the attitude. The statements are then administered to respondents, usually with a checklist or rating scales. Informants indicate whether the statement describes their feelings and thoughts. Responses are scored by summing together responses after reversing or reflecting some responses (e.g., reversing scale values by subtracting them from the largest anchor point value plus one), so that the meaning of the values is consistent and small (or large) scores all indicate the absence (or presence) of the attitude. This recoding of responses parallels the handling of responses with knowledge tests in that responses are scored in accordance with a previously determined standard. Attitude scales have been developed for a variety of topics, like depression, acculturation, and quality of life. Question formats can be dichotomous or checklist questions, but are usually rating scales indicating degree of agreement or frequency. ADAPTING EXISTING MATERIALS AND SCALES

There are many advantages to using existing questionnaires and standardized scales. Most important, it allows you to take advantage of the considerable amount of work that went into the development of the scale and facilitates communication with a larger group of scholars. The main disadvantage in using existing materials, especially standardized attitudinal scales, is the questionable validity of the results when applied to a new population. A scale developed on one population may not be directly transferable to another population as scale meaning and performance may not readily generalize to the new population. The application of a scale in a new setting can miss concepts that are important to the new group; ideas or elaborations of ideas in the new population may not be tapped or fully articulated in the original scale. Nevertheless, the advantages of adopting existing interview materials, when and where they exist, usually outweigh the disadvantages. One approach is to borrow and adapt materials as necessary. A thorough discussion of how to translate and modify materials (especially, tests) is presented by Brislin (1986; see also Jowell et al. 2007; Schrauf and Navarro 2005). Cross-cultural psychologists have extensive expertise in developing tests and scales that are comparable across cultural boundaries. The first step in adapting a test for another culture or another setting is to translate statements and rating scales. Materials should be translated from the source language to the target language by one person and then translated back into the source language by another person. Brislin recommends two full translation loops (four people). This is especially important for psychological concepts. Taking statements through such translation loops allows the investigator to see which concepts translate. Statements that retain their meaning through translation and retranslation are easily and directly usable. Statements that change meaning or are not consistent across translations need to be modified. The next step involves assuring that test questions are appropriate. One way to validate items on a test or statements in an attitude scale is to generate the item pool de novo. When applying the scale to a new group, even within the same language, it’s advisable to generate new items. Open-ended, free-listing questions with a small sample can sometimes reveal quickly and directly the content validity of the items. If newly generated items match or overlap with statements and concepts already included in the

14_107-Bernard.indb 363

6/9/14 10:03 AM

364

Susan C. Weller

scale or test, the scale probably needs little or no modification. If, on the other hand, open-ended interviews elicit many ideas and themes not well developed or measured on the test, then the test probably needs revision. One solution is to add new questions at the end of the set of standard questions. Adding new questions at the end allows you to score the scale in the accepted way and build on the body of literature relevant to the scales as well as to base an analysis on a new set of items. In a study of pre-term deliveries among inner-city African American women, a standardized measure of stressful life events was modified for that population. Stress, a severe strain or reaction that can be brought on by events and experiences, was measured with a checklist of 43 stressful life events (Holmes and Rahe 1967) that may have occurred in the past year, such as death of a spouse or change in residence; a greater number of positive answers indicates a greater number of life-changing events and possible higher stress. Before using the scale in a study of inner-city women, the investigators conducted open-ended, descriptive interviews with pregnant African American women about the stressful experiences in their lives. Interviews began with a discussion of stress to discover how it was defined and understood. Then discussions covered the kinds of things that caused stress. The results showed that although the women shared a general definition of stress and had experienced similar stress-causing situations, their stressful life events didn’t correspond completely with those in the Holmes and Rahe scale. For example, the women experienced stressful events that were not captured in the scale, such as loss of heat or electricity, being beaten or hit by a husband or boyfriend, and being evicted from their homes (being homeless). To be able to communicate with a larger group of researchers who might use the same scale, the investigators added new items to the end of the scale, rather than modify the scale itself. This gave them the flexibility to analyze stress in terms of either the standardized approach or as a modified test. Stress scales also have been adapted for use in other cultures (Ice and Yogo 2005). A limitation with attitudinal scales is their questionable validity when used on populations different from that on which the scale was developed. In general, this does not indicate a problem with the test, but instead is a problem with the application and conclusions. Validity, in its most general sense, is the degree to which something does what it is supposed to do. A valid question, scale, or test is one that measures what it is intended to measure. Content validity refers to the appropriateness of the items: Does the content of the items appear to be relevant to the topic that is being assessed? If responses from open-ended interviews with members of the target population overlap with the ideas contained in the existing set of questions, the questionnaire is appropriate for the new application. If the two sets of items overlap on many ideas but not all, the existing materials can be modified by adding new questions. If there is little overlap between the ideas and themes captured in new interviews and the existing materials, an alternative or new test is needed. CREATING A NEW INDEX OR SCALE

Nunnally (1978, 604–9) describes the process of creating an attitude scale. His discussion is summarized here as five steps:

14_107-Bernard.indb 364

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

365

1. An item pool is created by writing about 40 items on a single topic. (Themes may be taken from free-listing interviews or other sources.) Half of the statements should be moderately positive and half moderately negative. Statements where all or many respondents answer similarly do not help to differentiate people. Thus, neutral statements are not helpful nor are strong statements. 2. Statements are composed into a draft questionnaire and administered to individuals similar to whom the scale will eventually be administered (the target population). Questions may have dichotomous or rating scale responses. The number of respondents should be approximately ten times the number of items. (The sample size recommendation is because principal components analysis is used to ensure that statements are inter-correlated and cluster together as a single conceptual group.) 3. Responses are scored so that high scores all indicate the presence of the concept or trait and low scores indicate an absence of the trait. This means that some responses must be reversed or reflected prior to summation. If items were rated on 7-point rating scales where 1 = agree and 7 = disagree for positive statements, then responses for negative items need to be subtracted from 8 so that 1 = disagree and 7 = agree. Similarly, when responses are dichotomous and 0 = no and 1 = yes, then coding for negative statements should be reversed prior to summation and analysis. 4. Fourth, an individual’s score is the sum of his or her responses across items (after appropriate reversal of some items). Reliability of the total score is calculated from the average correlation among items and the number of items (alpha or KR-20). Reliability of individual items is determined by each item’s correlation to the total score (itemto-total correlation). All items should have a positive item-to-total correlation. (Items with a negative item-to-total correlation need to be reflected or omitted; see step 3.) 5. The final items are selected with high item-to-total correlations, say 10 positive and 10 negative statements from the original 40. A 20 item summative scale should have a reliability coefficient greater than .80.

Development of reliable and valid attitudinal scales is usually an iterative process involving data collection from several samples. For example, Lewis et al. (1984) were interested in measuring stress in pre-adolescent children. Previous studies of stress contained items relevant to adults or items thought to be relevant for children. The investigators began with individual and small group interviews with 50–60 fifth and sixth graders and asked, “What happens that makes you feel bad, nervous, or worry?” From the responses to this question (three questions), the researchers compiled a list of 22 items agreed on by the group. The degree to which the themes were well captured and expressed in existing scales of stress for children provides evidence for the validity of those scales. The degree to which the themes were mutually exclusive with existing scales challenges the valid use of such scales with children. The researchers determined that the themes were sufficiently unique to this population that they proceeded to create a new scale. Their next step was to pretest the 22 items as a questionnaire, rated on 5-point scales as to “How bad each would make you feel” and “How often each occurs.” The results of the pretest indicated that two items were almost always rated as “not bad,” and so were eliminated. The final 20-item test was then administered to 2,400 fifth graders.

14_107-Bernard.indb 365

6/9/14 10:03 AM

366

Susan C. Weller

RATING SCALES

Modifying existing materials or developing new materials involves making sure the content of the questions is appropriate for the population you are studying. Equally important is the format for responses to those questions. Cliff (1959) studied the effect of different descriptors that can be used to anchor rating scales and to help respondents interpret the points on a rating scale. He found that certain adverbs increase or decrease the value of an adjective by a predictable and measurable amount: “slightly” decreases an adjective by about half and “extremely” increases an adjective by about 50%. Thus, a rating scale constructed with “slightly pleasant,” “pleasant,” and “extremely pleasant” would have three ordinal categories with fairly equal intervals. Rating scales can be collected orally, if the task is simplified. For example, a 4-point rating scale can be presented orally (for phone administration or for someone who cannot read) by using two questions: First, “Is your health good or poor?” Then, “Is your health poor [#2] or very poor [#1]?” Or, “Is your health good [#3] or very good [#4]?”) (See Weller and Romney 1988, chapter on Rating Scales.)

Classification Studies In a departure from knowledge tests and attitudinal scales where the answers are known, classification studies attempt to understand and describe the ways in which individuals classify items into categories. This technique helps us understand categories of things according to informants. For a set of items, similarity data are collected from respondents without directing them to the criteria for making comparisons; judgments are made only in terms of the similarity or difference between items. Similarity distinctions are very basic distinctions in all cultures. Formats appropriate for similarity data collection are: pile sorting of items and paired or triadic comparisons of items. Typically, responses are aggregated across informants and the similarity information is represented with a spatial plot or tree structure to summarize the relationships among items. Results reveal relevant categories and sub-groupings of items that are salient to informants. A classification study has at least three parts. First, the set of items for study must be defined. Second, similarity between each pair of items is estimated. Third, the similarity data are represented with a spatial or tree model. Similarity information can be collected directly with judged similarity or indirectly with a measure of similarity between pairs of items across a series of questions (their similarity in profiles). Direct, judged similarity may be collected with the names of items written on cards and sorted into piles according to their similarity (pile sorting); with items presented in pairs and each pair is rated on the degree of similarity (paired comparisons), or items can be presented in sets of three and the most different item is selected (triadic comparisons). PILE SORTING

After the set of items for study has been defined, the name of each item can be written on a card or visual stimuli (pictures or objects) can be used. Informants are asked to read or review all of the items and to put them into piles, so that similar items are together in the same pile. Instructions are deliberately kept at a general level: Group the items according to their similarity without providing any specific criteria or examples.

14_107-Bernard.indb 366

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

367

Individuals may make as many or as few piles as they wish. Pile sorting was originally described by Miller (1969) and is reviewed in detail in Weller and Romney (1988) (also see Bernard and Ryan 2010). Judged-similarity data help us understand informants’ categories or perceptions of items. Sometimes in research you may be faced with a list of informant-generated items and want to know if some are redundant and whether there are categories of items that can be used to summarize main themes, or how to reduce the number of items but retain some items from each important category, or simply to describe perceptions and subcategories of things. Pile sorting is a way to find the categories as perceived by informants and not coded by the investigator, although pile sorting by the investigators can be used to develop coding categories (Hsaio et al. 2006; Sayles et al. 2007). Instructions are given to ask informants to sort the cards (or photos or things) into piles so that things that are similar are together in a pile. Things that belong together or are alike go together in a pile. Informants can make as many piles as they wish: “These are things that people have said. Please read all the cards and then sort them into piles, so that similar things are together in a pile, and different things are in different piles.” Additionally, instructions can be added about the number of piles: “Please make two or more piles” or “Please make seven to nine piles.” This is generally an easy task, and the fewer words on the cards, the easier it is. Pile sorting has been used to describe illnesses (Breiger 1994; Ross et al. 2002; Weller 1983, 1984), HIV risky behaviors (Macauda et al. 2011; Stanton et al. 1993), drugs (Carlson et al. 2004), addictions (Penka et al. 2008), problems among homeless youth (Ensign and Gittelsohn 1998), and types of dental pain (Moore et al. 1986). The method can be also used to study perception of plants (Benz et al. 2007; Berlin 1992; Berlin et al. 1974; Calvet-Mir et al. 2008; Nolan 2002) and animals (Boster and Johnson 1989; Lopez et al. 1997). Some applications include the study of social networks (Freeman et al. 1988, 1989; Johnson and Miller 1983; Miller and Johnson 1981), recreational activities (Miller and Hutchins 1989; Parr 1996; Roberts and Chick 1979; Roberts and Natrass 1980; Roberts et al. 1981); concepts of success and failure (Freeman et al. 1981; Romney et al. 1979), pilot errors (Roberts et al. 1980), activities of the elderly (Harman 2001), and emotions (Alvarado 1998; Lutz 1982). Pile sorting has also been used to develop salient and reliable categories for coding of qualitative data (Hsaio et al. 2006; Sayles et al. 2007). Kirk and Miller (1978) were interested in the perception of coca in South America and used pile sorting to discover if it was considered as a food product, a beverage, or a drug. They collected pile sort similarity data on 16 words, including foods, condiments, beverages, cigarettes, and drugs. They selected samples of 12 informants from each of 12 different sites: 2 cities in Colombia, 1 in Ecuador, and 6 locales in Peru (with 4 separate samples in Lima). Because Kirk and Miller used small, convenience samples, they used multiple samples to check the consistency or reliability of their results. Although a single, large representative sample would provide information on perceptions of coca; multiple, diverse, convenience samples can sometimes provide similar information—if the results are consistent across the diverse groups. If the results differ, then further work is necessary to discover what factors are associated with the difference. In this case, results were similar across samples, so the samples were combined.

14_107-Bernard.indb 367

6/9/14 10:03 AM

368

Susan C. Weller

Figure 11.1. Perceived similarity among foods and drugs (adapted from Figure 2 in Kirk and Miller 1978, 144; reprinted with permission).

Kirk and Miller’s classification or grouping of items appears here as Figure 11.1 in a dendrogram or tree-like representation (an adaptation of their “Figure 2 Diameter Method,” p. 144); it is a taxonomy of the similarity between items from a hierarchical clustering analysis. Here, “meat” and “food” are the most similar pair and are linked together at the lowest level of the tree, indicating the highest level of similarity. A cluster of edible things is then formed with other foods and condiments: meat, food, and vegetables joins with salt, and hot pepper. The beverages, coffee and chocolate, also belong to this cluster. Another cluster contains the drugs: herb, COCA, and marijuana are in one subgroup; and liquor, cigarettes, poison, and pills are in another. Thus coca, although chewed and often drunk as tea, is perceived to be a drug, similar to marijuana. The pile sort is a widely used and quick way to estimate similarity among items for a group of people. Informants are asked to sort the items into piles so that things that are similar are in the same pile together. The task is easily understood and facilitates conversation. After an individual has finished sorting items, she or he can describe the groupings. The data are best used to describe a group of individuals, rather than a single individual, because the data are sparse. Information from each individual only indicates if an item is paired with another or not (without information on the degree of similarity). Thus, only dichotomous (yes/no or one/zero) data are collected for each pair of items from each individual. Because of the sparseness of information at the individual level, the method is recommended for larger samples of people (at least 30 people) and for larger sets of items (two dozen or more items, where other methods of

14_107-Bernard.indb 368

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

369

data collection become prohibitive). Note also that informants must know how to read to sort words, although pictures or things can be sorted. To collect pile sort data, write or type the names of items on cards (and number the backs of each card). Then shuffle (and randomize) the cards and present them to an informant. Ask the informant to sort the cards according to their similarity so that similar things are in a pile together. Responses can be recorded immediately or later, if the piles are preserved by putting colored cards between the piles and putting a rubber band around the total set. Responses are recorded by piles. For example, if someone sorts seven things into four piles: Pile 1: 1, 2, 3 Pile 2: 4, 5 Pile 3: 6 Pile 4: 7

Here, seven items have been sorted into four piles: items #1, 2, and 3 are together; and items # 4 and 5 are together. Items # 6 and 7 were not put into piles with any other items. Similarity between each pair of the seven items can then be recorded into a square, symmetric table, or matrix. Since items 1, 2, and 3 are together, each pair in the group (1 and 2, 2 and 3, 1 and 3) are tabulated as similar. Items 4 and 5 also occur together and are tabulated as similar. All other pairs are not perceived to be similar and are coded with zeros. (See Weller and Romney 1988; for more detail, also see Bernard and Ryan 2010.) Responses are tabulated into a matrix for each individual and then summed together into an aggregate matrix for the entire sample of informants. The tabulation of responses can be done by hand or with the aid of computer software. ANTHROPAC (Borgatti 1996) translates the pile sort information for each respondent into individual and group matrices. The matrices can then be analyzed in ANTHROPAC or in other statistical software to represent and see the clustering of items into groups and subgroups. Variations on pile sorting include: allowing informants to “split” items, so that an item may go into more than one pile; constraining the number of piles an informant may make; or collecting successive pile sorts from each individual. Stefflre (1972) asked informants, when they were finished sorting items, if any items should go into more than one pile. Items or cards were then split and put into multiple piles. In the unconstrained version of the pile sort, informants may make as many or as few piles as they wish. In the constrained version; informants are instructed to make a specific number of piles, say between seven and nine piles (Romney et al. 1979). The constrained version of the pile sort attempts to control for individual differences in style; some individuals make finer discriminations between items (splitters) than others (lumpers). Burton (1975) proposed a method for assigning greater weight to the responses of splitters in an unconstrained sorting task. Because of the strong effect of such style differences, sorting tasks are usually not appropriate for comparisons between informants (Arabie and Boorman 1973; Boorman and Arabie 1972; Boorman and Oliver 1973). Comparisons between informants, rather than items, can be made only with an equal number of piles per informant or with successive pile sorts (Boster 1986a; Truex 1977; and see Weller

14_107-Bernard.indb 369

6/9/14 10:03 AM

370

Susan C. Weller

and Romney [1988] and Boster [1994] for more information on successive sorts). Successive pile sorting allows for more detailed information to be collected on each person (Lynch and Holmes 2011; Ross et al. 2011). PAIRED COMPARISON AND TRIADS SIMILARITY DATA

Since similarity data technically concerns pairs of items, sets of items can be created and informants can be asked directly about each pair. The advantage of such a design is that more detailed information is collected per informant and these designs can be used orally with people who cannot read. With k items there are k(k−1)/2 pairs or relationships to be estimated. Pile sort similarity data provide only dichotomous information (two values; co-occur = 1, do not co-occur = 0) on the k(k−1)/2 pairs for each informant. A direct rating of pairs, say on a 9-point rating scale, provides a 9-point range of information for each pair for each informant. A triad design offers a measurement range equal to the number of times each pair occurs. Thus, a paired comparison (two at a time) or a triadic (three at a time) design collects the same type of information as the pile sort, but collects more detail (finer discriminations of similarity) from each informant. The trade-off is that more information is collected per person, allowing for a smaller sample size and more reliable representation, but the tasks may be less interesting to informants than doing a pile sort. In triad designs, items are systematically arranged into sets of three (see Weller and Romney 1988; also Coombs 1954). Usually informants are instructed to pick the “most different” item in each set, which, in turn, identifies the most similar pair (the two remaining items). Pairwise similarity is thus estimated from responses. Picking the most different item is a simple task and can be done orally. Triads are really the only practical way to collect similarity data orally. Because of that, it is the method preferred for interviewing people with low literacy levels. Psychologists, working in more controlled conditions like classroom data collection, sometimes collect much more detailed information. For example, because a triad of items actually contains three pairs, some ask informants to identify the most similar pair in each triad and the least similar pair (Coombs 1954). In that way, all three pairs within each triad are ranked (1 = least, 2, and 3 = most similar). This latter method is much more intensive than the simple “pick the most different one,” and provides much more information per informant but is not practical for most field applications. Tasks collecting judged similarity data with systematic comparisons of items can collect more detailed information per informant, but the task can be lengthy. With k items there are: k (k−1) / 2 k! / [3! (k−3)!]

pairs in any set of items and triads.

Thus, with 10 items there are 45 pairs and 120 triads; with 21 items there are 210 pairs and 1330 triads. Because the paired comparison and triad designs quickly become cumbersome, there are special designs to limit the number of necessary subsets and still collect similarity judgments on pairs of items. These designs provide a systematic subset of possible

14_107-Bernard.indb 370

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

371

comparisons. An incomplete cyclic design for paired comparisons may include only 30 to 40% of the possible pairs and still accurately represent all pairs (Burton 2003). For triads, balanced-incomplete-block designs can be found in Burton and Nerlove (1976) or in Weller and Romney (1988). Balanced-incomplete-block designs are identified by the number of items to be compared (k), the size of the subsets (2 = pairs, 3 = triads, etc.), and the number of times each pair appears (lambda). A complete triads design for 21 items contains 1,330 unique sets of three items, but only 70 triads are necessary if a design is created where each pair occurs only once. A lambda-one design for 21 items has a large enough number of items to provide interesting results and yet is simple enough to be administered orally in the field. The first step in creating a triad design for a set of items is to select a balancedincomplete-block design for the number of items that you have (see Weller and Romney 1988). Second, enumerate the sets as specified in the design. Third, randomize the order of the sets and the order of items within each set (see Weller and Romney 1988, 33–34). Failure to randomize items can lead to biased selections by informants and might confound results (Romney et al. 1979). ANTHROPAC (Borgatti 1996) has an option to develop and print the data collection forms for many of the triad designs as well as tabulate the responses into a similarity matrix. Informants are asked to pick the most different item in each set. The task may be preceded by an example or two, but the examples should have obvious answers, they should come from a different domain, and the correct answer in each example should be in a different position within the set (second, first, third item). The similarity matrix containing the aggregate responses across all informants (whether from pile sorting, triads, or paired comparisons) can be analyzed to determine the perception or categorizations for the group. If pairs are rated, the first step is to list all possible pairs of items or to use a systematic subset of pairs. Remember that although there are k(k−1)/2 pairs in k items, there are systematic designs that can cut the number of necessary pairs in half (see Burton 2003). Second, the ordering of the pairs and the order of items within each pair is randomized. Third, a rating scale is created, where the smallest number indicates the least similar and the largest number indicates the most similar. Informants then judge the similarity of pairs of items on the rating scales. The rating scale value selected for each pair is tallied into a matrix. Applications using triads to collect similarity data include the study of kinship terms (Romney and D’Andrade 1964), animals (Henley 1969), occupations (Burton and Romney 1975; Magaña, et al. 1995), illnesses (Lieberman and Dressler 1977; Nyamongo 2002; Weller 1983; Young and Garro 1982), personality descriptors (Burton and Kirk 1979; Kirk and Burton 1977), and emotions (Alvarado and Jameson 2011; Romney et al. 1997). Triadic comparisons have also been used to study ethnobotanical classifications (Reyes-Garcia et al. 2004; Ross et al. 2005). In a study of emotion terms, Romney et al. (1997) compared monolingual English speakers’ and monolingual and bilingual Japanese-speakers’ similarity judgments of 15 emotion terms using triads (lambda 3) and paired comparisons (5-point rating scale) to judge the similarity of pairs of items. Figure 11.2 displays the similarity between terms and across the two monolingual samples in a spatial representation (adapted from Romney et al., 1997, Figure 2, p. 5491). Correspondence analysis was used to

14_107-Bernard.indb 371

6/9/14 10:03 AM

372

Susan C. Weller

Figure 11.2. Spatial representation of similarity among emotion terms.

represent the similarity data in two dimensions. The figure may be interpreted as a “map,” where closeness in the picture indicates similarity. Thus, “disgust,” “anger,” and “hate” are perceived as similar to one another and different from “sad” and “happy.” Differences between the two samples are negligible for four terms, small for eight terms (e.g., “disgust/mukatsuku,” “hate/kirai,” and “anger/haragatatsu”), and large for three terms (“shame/hazukashii,” “anxious/fuan,” and “bored/tsumaranai”). Romney et al. conclude that there is a substantial amount of shared meaning in emotions between the English and Japanese samples. In a study of societal problems, Wish and Carrol (presented in Kruskal and Wish 1990, 36–41) asked 14 individuals to rate 22 societal problems in terms of their similarity. Rating scales were used to collect judged similarity on all 231 pairs. Additional rating scales were used to rate the 22 problems on other, specific dimensions to aid in interpretation of the similarity dimensions (e.g., the degree to which each problem affects most people). The similarity between the 22 items (aggregated across informants) was represented spatially in three dimensions using multidimensional scaling (MDS). MDS is another multivariate analysis appropriate for the analysis of inter-item similarity data. Similarity relations are translated into distances creating a spatial representation like a map. Thus, closeness in the representation indicates similarity. The three dimensions that best explained informants’ perception of the societal problems were the degree to which the problem affected most people, the degree to which the problem was the responsibility of local government, and the degree to which the problem was technological. Figure 11.3 shows the latter two dimensions (adapted from Kruskal and Wish 1990:40, Figure 12b). In the lower-left quadrant of the figure are problems (“Failures in welfare”) thought to be the responsibility of local government; in the upper-right quadrant are those that are not the responsibility of local government (“Inflation”). Technological problems are in the lower-right quadrant and nontechnological problems are in the upper-left.

14_107_11_Weller.indd 372

6/9/14 10:59 AM

11: Structured Interviewing and Questionnaire Construction

373

Figure 11.3. Spatial representation of similarity among societal problems. SENTENCE-SUBSTITUTION OR PROFILE DATA

Similarity between items can be collected directly with judgments of similarity (pile sorting, triads, or paired comparisons) or similarity can be estimated indirectly, between the “profiles” of pairs of items across a series of questions. For example, D’Andrade et al. (1972) asked about the attributes of 50 illness terms by repeating the set of 50 attribute questions for each illness (2,500 questions); then they estimated the similarity between the illnesses from their proportion of shared attributes. This interviewing procedure—the systematic comparison of a set of items with a set of attributes or features—is sometimes called sentence-substitution data collection because the items are systematically substituted into sentence frames containing the attributes for the interview. Similarly, information can be collected in this way for social relationships within a group of people, where people are asked to rate the social relationship, the frequency of interaction, or the similarity between themselves and each person in the group (Wasserman and Faust 2009, 45–55). Sentence-substitution interviews begin with two related lists. The first list is the set of domain items and the second is a set of statements about the domain items. The latter list may include descriptive statements, attributes, features, or uses (behaviors) relevant to the domain items. In the interview, each item is paired with every attribute and informants are asked to judge the acceptability or veracity of the newly formed statement. The task is easy to understand and may be administered orally. For oral administration, a matrix can be used to indicate the intersection of the two lists (rows as attributes and columns as domain items) and each question regarding each

14_107_11_Weller.indd 373

6/9/14 10:59 AM

374

Susan C. Weller

attribute can be read by systematically substituting each of the domain items. For written administration, all statements should be completely written out with correct syntax. Responses can be dichotomous (yes/true or no/false), or a rating scale can be used for each question. Usually, the responses of informants are aggregated into a single item-by-attribute table, where responses are represented with the modal response (for categorical data), averaged (for rating scales), or cultural consensus is used to estimate the best answers (see the section below on Exploring Beliefs). Similarity between items may be calculated from their shared attributes (or similarity between attributes can be calculated from their co-occurrence in items). From either, a square symmetric matrix of similarities is obtained. In D’Andrade et al.’s (1972) study of illnesses and illness attributes, the similarity between each pair of illnesses (across attributes) was calculated with a Pearson correlation coefficient. The itemby-item correlation matrix was represented with MDS and hierarchical clustering. Clustering results can be used to interpret the similarity between items and to reorder the rows (attributes) and columns (items) in the aggregate item-by-attribute response table so that the joint item-attribute clusters can be seen. This multi-step process (correlation matrices, multivariate analysis, and reordering of the rows and columns to see patterns) can now be accomplished in a single step with correspondence analysis (Weller and Romney 1990). Several applications have used sentence-substitution questions to explore illnesses and their symptoms and treatments (D’Andrade et al. 1972; Garro 1986; Maupin et al. 2011, Ross et al. 2012; Stefflre 1972; Young 1978). Other examples include looking at how Peace Corps workers are perceived (types of people by behaviors, in Stefflre 1972) and beliefs about adolescent punishments (adolescent behaviors by punishments, in Weller et al. 1987). D’Andrade (1976) and Young (1978) also attempted to identify attributes that best differentiated illness categories. A strength of this method is the linking of two related sets of items. Sentence-substitution data provide rich and valuable information, but the interview can be lengthy. Interviews like Stefflre’s (1972) and D’Andrade et al.’s (1972) comparison of 50 items and 50 attributes (2,500 questions) were carried out over a few days and informants were reimbursed for their time. A more general form of this type of interviewing is the systematic collection of information on any two related lists of items to create a profile of information for one set of items based on the second set of items. For example, interviews with members of a small face-to-face social group (whole network) may ask that each group member “Name the individuals with whom you interact the most,” “Name the three people with whom you interact the most,” or “Rate each person in terms of how much you interacted with them in the past month.” Although these three questions vary from unconstrained and constrained dichotomous responses (named = 1, not named = 0) to rated (or ranked) responses for each person in the group, the information collected refers to the set of all group members. The two related lists each contain the names of all members: The first list indicates the informant or person interviewed and the second list indicates the informant’s responses or choices for everyone else in the group. Similarity is then calculated between informants, based on their profile of responses/choices. Similarity in their pattern of choices may be calculated with a Pearson correlation coefficient (phi) or

14_107-Bernard.indb 374

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

375

other measure and represented spatially with MDS, correspondence analysis, or graph theoretic techniques (Wasserman and Faust 2009). RELIABILITY AND VALIDITY OF SIMILARITY DATA REPRESENTATIONS

Data collection and analysis for the study of classifications include three steps: (1) getting the similarity data; (2) tabulating the data into a single table or matrix for each group; and (3) getting a descriptive model or representation of the similarity relationships. Similarity data may be collected directly with pile sorting, triads, or paired comparisons or measured indirectly from the shared attributes across items. With direct judged similarity, a similarity matrix is created for each individual and then the matrices are summed together into a single matrix. Tabulation of similarity can be done by hand or by computer (Borgatti 1996). With indirect measures of similarity, a matrix of similarity coefficients (e.g., Pearson correlation coefficients) is generated by a computer program. Finally, the aggregate similarity between items in the form of a square, symmetric matrix of similarities is usually represented with a descriptive, visual multivariate technique. Descriptive statistical analyses used for the representation of similarity data include clustering (Mezzich and Solomon 1980), nonmetric MDS (Kruskal and Wish 1990; Mezzich and Solomon 1980), factor analysis or principal components analysis (Weller and Romney 1990), and correspondence analysis (Weller and Romney 1990). These analyses are available in most major statistical packages. Hierarchical clustering represents the relationships between items in a tree-like structure or dendrogram, like a taxonomy. Both MDS and correspondence analysis provide spatial representations of data so that similar items are closer together on a map or plot of items, as can factor analysis or principal components. Correspondence analysis is a sister of principal components, appropriate for scaling qualitative/categorical data. Correspondence analysis allows for the simultaneous scaling of items and attributes in the same spatial configuration, facilitating a sentence-substitution data analysis. A variety of studies have been undertaken to test the validity and reliability of using one of these multivariate models to represent similarity data. Validity concerns the degree to which these multivariate models actually represent how people perceive and think about the items. Simple exercises include submitting a set of interpoint distances (where similarity is the degree of propinquity) for analysis and checking to see if the same information can be retrieved. As mentioned, although there are many types of clustering methods, the average-link method (Sokal and Sneath 1963) tends to outperform others in being able to retrieve known structures (Milligan 1980). Green and Carmone (1970) illustrate MDS’s ability to translate such information into an accurate “map” with a configuration of points representing the letters “A” and “M”; Kruskal and Wish do so with a map of the United States. Weller and Romney (1990) repeat Kruskal and Wish’s example and show that correspondence analysis also can translate inter-city mileages into a map. Magaña et al. (1981) studied the perception of a college campus and compared estimates of distances, triad judgments, and distances from hand-drawn maps and found the MDS representations to accurately reflect true distances. A more complicated form of validation concerns the degree to which such models are accurate representations of what people think. Judged similarity data, when rep-

14_107-Bernard.indb 375

6/9/14 10:03 AM

376

Susan C. Weller

resented with multivariate techniques, predict memory performance, judgments, and reasoning task performance. Friendly (1977) used hierarchical clustering and MDS models of free-recall listing and similarity data to successfully predict memory performance tasks. Similarly, Romney et al. (1993) used an MDS model of similarity data to predict list length in a free-recall listing task. Hutchinson and Lockhead (1977) used MDS model inter-item distances to predict reaction time judgments concerning similarity. Rumelhart and Abramson (1973) used an MDS model to predict informants’ responses on analogical reasoning tasks concerning animals. Most studies have found similarity judgments to be highly reliable. This means that there often tends to be little intracultural variation in these judgments. Romney, Smith et al. (1979) in a study of concepts of success and failure, compared results across several samples and found them to be highly concordant. A check on the internal consistency (agreement) in similarity judgments is an important step in justifying an aggregate representation. Similarity between items, using different methods of estimating similarity, is usually concordant (compare D’Andrade et al. 1972 and Weller 1983; and see Young and Garro 1982 and Romney et al. 1997).

Exploration of Specific Beliefs A series of questions on a single topic may be used to evaluate knowledge, attitudes, or beliefs. In studies of beliefs, however, the purpose is to discover the answers and not to measure deviance from a standard. Thus, only the original responses are used and they are not scored, transformed, or recoded as for attitudinal scales. Studies focusing on beliefs are similar to classification studies, except that classification studies rely on similarity data without reference to specific criteria and studies of beliefs often focus on specific criteria. Questions for studies of beliefs are written following the same process as for studies concerning knowledge tests and attitude scales. Question formats include: open-ended questions requesting short answers or phrases; questions with predetermined multiple choice response categories (including dichotomous yes/no or true/false); requests to rank order items on a specific topic; and open-ended questions requesting numeric estimates (like frequencies, distances, or ages). Typically, beliefs are estimated by aggregating responses across informants. To assess beliefs, interviews are conducted with a series of statements or questions all on the same topic, all in the same format, and all at the same level of difficulty. As with all interview materials, the items should be relevant to the informants and should be developed from content obtained in open-ended interviews. Clear and simple wording should be used, so that each question is understood in the same way by each person. The actual format of questions is guided by the purpose of the study. If the purpose is to discover detailed beliefs (e.g., a cultural model of the causes, symptoms, and treatments for an illness), then an appropriate format may be a series of yes/no or true/false questions covering attributes of the illness (e.g., Weller et al. 2012). With yes/no or true/ false questions, care must be taken to balance the list with approximately half positive (true or yes) and half negative (false or no) statements. Alternatively, a project might focus on a single question, “What causes breast cancer?” (Chavez et al. 1995), and a set of possible causes can be rank ordered from most to least likely causes. Or possible sources of support can be judged as appropriate for

14_107-Bernard.indb 376

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

377

scenarios where help might be needed, “To whom would you go for advice or support?” (Berges et al. 2006; Dressler et al. 1997). Or a researcher may ask simple open-ended questions such as asking informants to identify plants (Boster 1986b). Questions may look like those for a knowledge test or an attitudinal scale; the difference is how responses are handled and analyzed. A description of beliefs involves a summarization or aggregation of responses for each question. Intuitively, the best estimate of an answer is provided by the majority response or an average of responses across informants (D’Andrade 1987). Such measures, called central tendency measures in statistics, are the best single description of responses to a question. Thus, open-ended or categorical responses are best described by the majority or modal response, and ranked or interval data are best described by the median (midpoint) or mean (average) response. Aggregate measures, however, are accurate only to the degree that there is little to moderate variability in responses. The notion of homogeneity in responses for a single question can be generalized to a set of questions. Homogeneity across informants’ responses for a series of questions can be assessed with a measure of agreement. Field data indicate that agreement is related to accuracy (Young and Young 1962); if you ask three people where the post office is and they all give identical answers, then it is more likely that the information is true and correct, than when their answers conflict. The relation between consistency and validity can be expressed as a general principle of aggregation. The accuracy of aggregated responses is a function of the agreement among informants and the number of informants (the Spearman-Brown Prophesy Formula, described in Weller and Romney 1988). In other words, the agreement between each pair of informants is measured with a Pearson correlation coefficient and averaged across all pairs of informants; the higher the average agreement among informants, the fewer informants are necessary to achieve an accurate estimate of the “true” answers from an aggregation of their responses (see also Weller 1987). Thus, shared beliefs can be estimated by combining the responses of informants, if there is sufficient agreement among informants. The cultural consensus model estimates culturally appropriate answers to a set of questions and the degree to which each informant shares those answers (for an overview, see Weller 2007). It assumes that the ethnographer does not know the answers to the questions or how much each informant knows about the domain under consideration. The analysis determines if there are highly shared beliefs and, if so, estimates the answer for each question and how much each informant knows those answers. The model also includes a method for estimating the number of informants needed to provide given levels of confidence in the answers for different levels of shared cultural knowledge. With highly shared beliefs, accurate results can be obtained with few informants. Within cultural consensus theory, there are two models or approaches: formal and informal. The formal model is a psychological process model of how questions are answered with varying degrees of knowledge and bias that estimates the knowledge levels of respondents and likelihood that specific answers are correct (Romney et al. 1986). The model can only accommodate categorical responses, such as multiple choice data, including dichotomous data (yes/no or true/false) or open-ended responses (a word or

14_107-Bernard.indb 377

6/9/14 10:03 AM

378

Susan C. Weller

short phrase). The analysis is similar to a factor analysis of people, but requires special software, as Bayesian methods are used to solve for estimates. Categorical responses can be accommodated by the formal cultural consensus model. The formal model is appropriate for open-ended responses (a series of questions requesting a single word or short phrase), lists of statements requiring dichotomous choices (true/false or yes/no), dichotomous judgments of statements formed by linking two related lists (sentence-frame substitutions), and multiple choice responses. Open-ended questions were used by Boster (1986b), who walked Jivaroan women through a garden and asked them to name plants. Extensive work has been done with dichotomous responses (true/false and yes/no), especially on illnesses: AIDS (Baer et al. 1999b; Baer et al. 2004; Trotter et al. 1999), asthma (Pachter et al. 2002), the common cold (Baer et al. 1999a), diabetes (Smith 2012; Weller et al. 2012), folk illnesses (nervios, Baer et al. 2003; empacho, Weller et al. 1993; susto, Weller et al. 2002), and maternal health knowledge and infant health (Miller 2011). A true/false format also was used to explore beliefs about pollution and safety of seafood (Johnson and Griffith 1996). Sentence substitutions can be used to find normative answers to the joint assertions formed by combining two related lists (Garro 1988; Maupin et al. 2011; Ross et al. 2012). Multiple choice responses have been used to study gender concepts (de Munck et al. 2002) and shared knowledge about fish habitats and behaviors (Garcia-Quijano 2009). Garcia-Quijano (2009) asked five types of questions for 16 different species of fish. For example, fishermen were asked about each fish’s habitat (bays, mangrove channels, sand bottoms, seagrasses, reefs, open water, mud bottoms, and deep water); and their seasons (early winter, late winter, spring, early summer, late summer, or fall). The informal cultural consensus model is the most widely accessible model with the least assumptions about the data (Romney et al. 1987). The informal model is a collection of analytical procedures that approximate the results of the formal model, estimating answers or the ordering of answers on a specific construct and estimating the degree to which each person’s responses correspond with that ordering. For this model, items are typically ordered from most to least on a specific concept. Conceptually, this model averages responses across people to estimate answers and then correlates each person’s answers with the aggregated answers of the group to estimate their correspondence to the group consensus or their “cultural knowledge.” This approach includes a reliability or factor analysis of people rather than items and can be run in most major statistical software packages (see Weller 2007). For example, Caulkins (2001; Trosset and Caulkins 2001) studied ethnic identity by having people in different regions of the United Kingdom rate scenarios (e.g., a “child performing a song for family guests while standing on the kitchen stool”) on how “Welsh” each was. Consensus can also be used to identify shared values and norms within organizations (Caulkins and Hyatt 1999, Jaskyte and Dressler 2004, 2005; Smith et al. 2010). Ranked responses are accommodated in the informal cultural consensus model. Applications using the informal model include studies of illnesses, social support, and occupational prestige to examine shared beliefs within and across subgroups. Chavez et al. (1995) compared the beliefs of four different groups of Latinas and one group of physicians by having each group of informants rank order 30 potential causes of breast cancer. Magaña et al. (1995) compared U.S., Mexican, and Guatemalan perceptions

14_107-Bernard.indb 378

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

379

of socioeconomic status and prestige by comparing informants’ rank-orderings of occupations. Koster and Tankersley (2012) examined perceived hunting ability in dogs by having hunters rank the dogs and Koster et al. (2010) examined hunters’ desire for particular meat flavors by ranking meats in terms of their desirability and taste. In a high AIDS mortality area of Africa, Kiš (2007) had residents rank order reasons for attending a funeral to understand changing values about why people would and would not attend a funeral. Fully ranked data may be collected with paper and pencil, interactively with cards, and orally (for detail on ranking methods, see Weller and Romney 1988). Respondents may be asked to rank k items by putting a number next to each item using paper-and-pencil data collection. Or names of items can be written on cards and spread out on a table and the informant is asked to pick up the most important item, then the next important item, and so on until all the cards are ordered. For both cases, data recording would have the list of items and the researcher would record the rank assigned to each. Smith et al. (2004; Smith et al. 2010) used cards to collect information on patient and physician priorities in primary care. Balanced incomplete designs also allow for the collection of information orally by presentation of subsets (triads or pairs) of items, and combining responses to create a full rank order of items for each person. Cain et al. (2011) studied cultural norms on the appropriateness of genitalia terms by having respondents rank terms in sets of three. Reyes-Garcia et al. (2004) had informants rank plants on their usefulness by orally presenting the plants in pairs and then combining responses. Rating scales can sometimes be used, but care must be taken to ensure that positive and negative items are used and that the range of scale values is used by each person. One means for doing this is to use a constrained rating scale task called a Q-sort (Weller and Romney 1988). The rating scale is typically arrayed on a table and informants are instructed to put each item on a rating scale value, with the caveat that the researcher constrains the task by limiting how many items can go on each value. For example, a researcher can request that 16 items be placed so that one item is rated as “1,” two as “2,” three as “3,” four as “4,” three as “5,” two as “6,” and one as “7.” This can be accomplished by having the desired number of pockets under each rating scale value so it is clear how many items “go with” each value. Rocha (2005, 363) used a Q-sort to collect ordered data. For example, photographs of 34 crops were ordered on their difficulty to be tended and were put into five piles from most to least difficult, and soil types and fertilizers were rated on a 3-point scale with three piles, ensuring that all respondents used all three ordinal categories. Similarity data may also be used, if similarity is collected with a systematic comparison method (triads or paired-comparisons) or with successive pile sorts. Reyes-Garcia et al. (2004) examined intra-cultural variation in shared knowledge of plants by collecting judged similarity data with triadic comparisons, calculating the similarity between pairs of items, then representing shared knowledge with the cultural consensus model. Romney et al. (1997) and Alvarado and Jameson (2011) used triad similarity data to study normative meaning of emotion terms cross-culturally. Lynch and Holmes (2011) used successive pile sorts to study lay perceptions of food group categories and Ross et al. (2011) used successive sorting to study illnesses.

14_107-Bernard.indb 379

6/9/14 10:03 AM

380

Susan C. Weller

An important application has been the comparison of what people have (social support, material goods, etc.) with what local norms indicate they should have. Dressler et al. (1997) studied cultural preferences for different sources of social support by having informants rank order possible sources of support in terms of their appropriateness in different scenarios. The agreement between individual circumstances and group norms has been called cultural consonance (Dressler 1996; Dressler et al. 1997; Dressler et al. 2005; Dressler et al. 2012; Reyes-Garcia et al. 2010). The validity and accuracy of estimates provided by the informal consensus model are illustrated by Romney et al.’s (1987) study on causes of death, where rankings of the frequency of perceived causes of death in the population were compared with actual mortality rates. Dawes’s (1977) study on estimated heights as compared to actual heights illustrates validity, although the study preceded the formalization of the cultural consensus model. Webster et al. (2002) correlated answers obtained from consensus rankings to other performance measures and personality characteristics. Also, Romney and Weller (1984) used individual correspondence to group answers to predict individual accuracy in reporting social interaction patterns. While simple aggregation of responses with moderate to high agreement is a sound procedure, there are some limitations in the application of consensus theory to response data. Clarity of questions is always an issue. Questions must be clear and understandable to all participants and interpreted in the same way. Missing data can be an issue, and care must be taken to get answers to as many questions as possible. The formal model assumes that there is no response bias, although newer formulations of the model can estimate the amount of bias as well as item difficulty (Karabatsos and Batchelder 2003; Oravecz et al. 2014). Response bias can have many forms; with field data, it may be the simple pattern of respondents to tend to say “yes” to all questions about which they have doubt or conversely to say “no” when in doubt. It is also important to note that “I don’t know” currently cannot be handled as a response choice, but is instead accounted for with guessing. Another issue is to be sure that positive and negative statements are both represented; a very skewed distribution (very few positive answers or very few negative answers) can affect the model’s estimates. SUMMARY

Sociological and psychological literature offer many lessons about writing questions. The authors in those areas have extensive experience in writing questions for surveys and tests, important for all types of questionnaires and interview materials. Questions should be simple, clear, and interpreted in the same way by everyone (the respondent and the researcher). First, preliminary or ethnographic interviewing and free listing provide valuable information for the development of content for questions. Second, when questionnaires are drafted, pilot testing with responses and/or interpretation of questions provides valuable feedback on the clarity of questions. Great insight can be gained with as few as three interviews. Finally, development of questions for surveys, knowledge tests, attitude scales, or belief studies all involve the same processes. Time invested during development can save grief in the interpretation stage. There are no shortcuts.

14_107-Bernard.indb 380

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

381

REFERENCES

Alvarado, N. 1998. A reconsideration of the structure of the emotion lexicon. Motivation and Emotion 22: 329–44.  Alvarado, N., and K. A. Jameson. 2011. Shared knowledge about emotion among Vietnamese and English bilingual and monolingual speakers. Journal of Cross-Cultural Psychology 42: 963–82. Arabie, P., and S. A. Boorman. 1973. Multidimensional scaling of measures of distances between partitions. Journal of Mathematical Psychology 10: 148–203. Baer, R. D., S. C. Weller, J. E. Garcia de Alba Garcia et al. 2003. A cross-cultural approach to the study of the folk illness “nervios.” Culture, Medicine, and Psychiatry 27: 315–37. Baer, R. D., S. C. Weller, J. E. Garcia de Alba et al. 2004. A comparison of community and physician explanatory models of AIDS in Mexico and the U.S. Medical Anthropology Quarterly 18: 3–22. Baer, R. D., S. C. Weller, L. M. Pachter et al. 1999a. Cross-cultural perspectives on the common cold: Data from five populations. Human Organization 58: 251–60. Baer, R. D., S. C. Weller, L. M. Pachter et al. 1999b. Beliefs about AIDS in five Latin and AngloAmerican populations: The role of the biomedical model. Anthropology and Medicine 6: 13–29. Benz, B., H. Perales, and S. Brush. 2007. Tzeltal and Tzoltil farmer knowledge and maize diversity in Chiapas, Mexico. Current Anthropology 48: 289–300. Berges, I. M., F. Dallo, A. DiNuzzo et al. 2006. Social support: A cultural model. Human Organization 65: 420–29. Berlin, B. 1992. Ethnobiological classification: Principles of categorization of plants and animals in traditional societies. Princeton, NJ: Princeton University Press. Berlin, B. O., D. Breedlove, and P. Raven. 1968. Covert categories and folk taxonomies. American Anthropologist 70: 290–99. Berlin, B. O., D. Breedlove, and P. Raven. 1973. General principles of classification and nomenclature in folk biology. American Anthropologist 75: 214–44. Berlin, B. O., D. Breedlove, and P. Raven. 1974. Principles of Tzeltal plant classification: An introduction to the botanical ethnography of a Mayan-speaking people of Highland Chiapas. New York: Academic Press Berlin, B. O., and P. D. Kay. 1969. Basic color terms: Their universality and evolution. Berkeley: University of California Press. Bernard, H. R. 2013. Social research methods: Qualitative and quantitative approaches, 2nd ed. Thousand Oaks, CA: Sage. Bernard, H. R., P. D. Killworth, D. Kronenfeld, and L. Sailer. 1984. On the validity of retrospective data: The problem of informant accuracy. Annual Review in Anthropology 13: 495–517. Bernard, H. R., P. D. Killworth, and L. Sailer. 1980. Informant accuracy in social network data IV: A comparison of clique-level structure in behavioral and cognitive network data. Social Networks 2: 191–218. Bernard, H. R., and G. W. Ryan. 2010. Analyzing qualitative data. Los Angeles: Sage. Boorman, S. A., and P. Arabie. 1972. Structural measures and the method of sorting. In Multidimensional scaling: Theory and applications, Vol. 1, ed. R. Shepard, A. K. Romney, and S. B. Nerlove, 225–49. New York: Seminar Press. Boorman, S. A., and D. C. Olivier. 1973. Metrics on spaces of finite trees. Journal of Mathematical Psychology 10: 26–59. Borgatti, S. P. 1996. ANTHROPAC. Version 4.98. Natick, MA: Analytic Technologies. Boster, J. S. 1986a. Can individuals recapitulate the evolutionary development of color lexicons? Ethnology 25: 61–74.

14_107-Bernard.indb 381

6/9/14 10:03 AM

382

Susan C. Weller

Boster, J. S. 1986b. Exchange of varieties and information between Aquaruna manioc cultivators. American Anthropologist 88: 428–36. Boster, J. S. 1994. The successive pile sort. Field Methods 6: 11–12. Boster, J. S., and J. C. Johnson. 1989. Form or function: A comparison of expert and novice judgments of similarity among fish. American Anthropologist 91: 866–89. Bousfield, W. A., and W. D. Barclay. 1950. The relationship between order and frequency of occurrence of restricted associative responses. Journal of Experimental Psychology 40: 643–47. Bradburn, N. M., S. Sudman, and B. Wansink. 2004. Asking questions: The definitive guide to questionnaire design for market research, political polls, and social and health questionnaires. New York: John Wiley & Sons. Breiger, W. R. 1994. Pile sorts as a means of improving the quality of survey data: Malaria illness symptoms. Health Education Research 9: 257–60. Brewer, D. 2002. Supplementary interviewing techniques to maximize output in free-listing tasks. Field Methods 14: 108–18. Brislin, R. W. 1986. The wording and translation of research instruments. In Field methods in cross-cultural research, ed. W. J. Lonner and J. W. Berry, 137–64. Thousand Oaks, CA: Sage. Bukov, A., I. Maas, and T. Lampert. 2002. Social participation in very old age: Cross-sectional and longitudinal findings from BASE. Journal of Gerontology: Psychological Sciences 57B: 510–17. Burt, R. S. 1984. Network items and the General Social Survey. Social Networks 6: 293–340. Burt, R. S. 1986. A note on sociometric order in the General Social Survey. Social Networks 8: 149–89. Burton, M. L. 1975. Dissimilarity measures for unconstrained sorting data. Multivariate Behavioral Research 10: 409–24. Burton, M. L. 2003. Too many questions? The uses of incomplete cyclic designs for paired comparisons. Field Methods 15: 115–30. Burton, M. L., and L. Kirk. 1979. Sex differences in Maasai cognition of personality and social identity. American Anthropologist 81: 841–73. Burton, M. L., and S. B. Nerlove. 1976. Balanced designs for triad tests: Two examples from English. Social Science Research 5: 247–67. Burton, M. L., and A. K. Romney. 1975. A multidimensional representation of role terms. American Ethnologist 2: 397–407. Cain, D., S. Shensul, and R. Mlobeli. 2011. Language choice and sexual communication among Xhosa speakers in Cape Town, South Africa: Implications for HIV prevention message development. Health Education Research 26: 476–88. Calvet-Mir, L., V. Reyes-Garcia, and S. Tanner. 2008. Is there a divide between local medicinal knowledge and Western medicine? A case study among Native Amazonians in Bolivia. Journal of Ethnobiology and Ethnomedicine 4: 18–28. Canales, M., T. Hernandez, J. Caballero et al. 2005. Informant consensus factor and antibacterial activity of the medicinal plants used by the people of San Rafael Coxcatlán, Puebla, Mexico. Journal of Ethnopharmacology 97: 429–39. Carlson, R. G., J. A. McCaughan, R. S. Falck et al. 2004. Perceived adverse consequences associated with MDMA/Ecstasy use among young polydrug users in Ohio: Implications for intervention. International Journal of Drug Policy 15: 265–74. Caulkins, D. D. 2001. Consensus, clines, and edges in Celtic cultures. Journal of Cross-Cultural Research 35: 109–26. Caulkins, D. D., and S. B. Hyatt. 1999. Using consensus analysis to measure cultural diversity in organizations and social movements. Field Methods 11: 5–26.

14_107-Bernard.indb 382

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

383

Chavez, L. R., F. A. Hubbell, J. M. McMullin et al. 1995. Structure and meaning in models of breast and cervical cancer risk factors: A comparison of perceptions among Latinas, Anglo women, and physicians. Medical Anthropology Quarterly 9: 40–74. Cliff, N. 1959. Adverbs as multipliers. Psychology Review 66: 27–44. Conklin, H. 1969. Lexicographical treatment of folk taxonomics. In Cognitive anthropology, ed. S. Tyler, 41–59. New York: Holt, Rinehart and Winston. Coombs, C. H. 1954. A method for the study of interstimulus similarity. Psychometrika 19: 183–94. D’Andrade, R. G. 1974. Memory and the assessment of behavior. In Measurement in the social sciences, ed. H. M. Blalock, 139–86. Chicago: Aldine. D’Andrade, R. G. 1976. A Propositional Analysis of U.S. American beliefs about illness. In Meaning in anthropology, ed. K. Basso and H. Selby, 155–80. Albuquerque: University of New Mexico Press. D’Andrade, R .G. 1987. Modal responses and cultural expertise. American Behavioral Sciences 31: 194–202. D’Andrade, R. G. 1995. The development of cognitive anthropology. Cambridge: Cambridge University Press. D’Andrade, R. G., N. Quinn, S. B. Nerlove, and A. K. Romney. 1972. Categories of disease in American-English and Mexican-Spanish. In Multidimensional scaling: Theory and applications in the behavioral sciences, Vol. 2, ed. A. K. Romney, R. Shepard, and S. B. Nerlove, 9–54. New York: Seminar Press. Dawes, R. M 1977. Suppose we measured height with rating scales instead of rulers. Applied Psychological Measurement 1: 267–73. de Munck, V. C., N. Dudley, and J. Cardinale. 2002. Cultural models of gender in Sri Lanka and the United States. Ethnology 41: 225–61. DeWalt, B. R. 1979. Modernization in a Mexican ejido: A study in economic adaptation. Cambridge: Cambridge University Press. Dickson, P. R., R. F. Lusch, and W. L. Wilkie. 1983. Consumer acquisition priorities for home appliances: A replication and re-evaluation. Journal of Consumer Research 9: 432–35. Dressler, W. W. 1996. Culture and blood pressure: Using consensus analysis to create a measurement. Field Methods 8: 6–8. Dressler, W. W., M. C. Balieiro, and J. E. dos Santos. 1997. The cultural construction of social support in Brazil: Associations with health outcomes. Culture, Medicine, and Psychiatry 21: 303–35. Dressler, W. W., M. C. Balieiro, R. P. Ribeiro, and J. E. dos Santos. 2005. Cultural consonance and arterial blood pressure in urban Brazil. Social Science & Medicine 61: 527–40. Dressler, W. W., K. S. Oths, M. C. Balieiro et al. 2012. How culture shapes the body: Cultural consonance and body mass in urban Brazil. American Journal of Human Biology 24: 325–31. Ensign, J., and J. Gittelsohn. 1998. Health and access to care: Perspectives of homeless youth in Baltimore City, USA. Social Science and Medicine 47: 2087–99. Fern, E. F. 1982. The use of focus groups for idea generation: The effects of group size, acquaintanceship, and moderator on response quantity and quality. Journal of Marketing Research 19: 1–13. Fink, A. 2003. The survey kit, 2nd ed. Thousand Oaks, CA: Sage. Fowler, F. J. 1995. Improving survey questions design and evaluation. Applied Social Research Methods Series, Vol. 38. Newbury Park, CA: Sage. Fowler, F. J. 2009. Survey research methods, 4th ed. Applied Social Research Methods Series. Newbury Park, CA: Sage.

14_107-Bernard.indb 383

6/9/14 10:03 AM

384

Susan C. Weller

Frake, C. O. 1964. Notes on queries in ethnography. Transcultural Studies in Cognition 66: 132–45. Freeman, H. E., R. E. Klein, J. Kagan, and C. Yarbrough. 1977. Relations between nutrition and cognition in rural Guatemala. American Journal of Public Health 67: 233–39. Freeman, H. E., A. K. Romney, J. Ferreira-Pinto et al. 1981. Guatemalan and U.S. concepts of success and failure. Human Organization 40: 140–45. Freeman, L. C., and R. Danching. 1997. An international comparative study of interpersonal behavior and role relationships. L’Année Sociologique 47: 89–115. Freeman, L. C., S. C. Freeman, and A. G. Michaelson. 1988. On human social intelligence. Journal of Social and Biological Structures 11: 415–25. Freeman, L. C., S. C. Freeman, and A. G. Michaelson. 1989. How humans see social groups: A test of the Sailer-Gaulin models. Journal of Quantitative Anthropology 1: 229–38. Freeman, L. C., A. K. Romney, and S. C. Freeman. 1987. Cognitive structure and informant accuracy. American Anthropologist 89: 310–25. Friendly, M. L. 1977. In search of the M-gram: The structure of organization in free recall. Cognitive Psychology 9: 188–249. García-Quijano, C. G. 2009. Managing complexity: Ecological knowledge and success in Puerto Rican small-scale fisheries. Human Organization 68: 1–17. Garro, L. C. 1986. Intracultural variation in folk medical knowledge: A comparison between curers and noncurers. American Anthropologist 88: 351–70. Garro, L. C. 1988. Explaining high blood pressure: Variation in knowledge about illness. American Ethnologist 15: 98–119. Green, P. E., and F. J. Carmone. 1970. Multidimensional scaling and related techniques in marketing analysis. Boston: Allyn and Bacon. Guest, G. 2000. Using Guttman scaling to rank wealth: Integrating quantitative and qualitative data. Field Methods 12: 346–57. Handwerker, W. P. 1996. Constructing Likert scales: Testing the validity and reliability of single measures of multidimensional variables. Cultural Anthropology Methods 8(1): 1–6. Harman, R. C. 2001. Activities of contemporary Mayan elders. Journal of Cross-Cultural Gerontology 16: 57–77.  Haug, M. R. 1977. Measurement in social stratification. Annual Review in Sociology 3: 51–77. Henley, N. M. 1969. A psychological study of the semantics of animal terms. Journal of Verbal Learning and Verbal Behavior 8: 176–84. Holmes, T. H., and R. H. Rahe. 1967. The social readjustment rating scale. Journal of Psychosomatic Research 11: 213–18. Hsiao, A. F., G. W. Ryan, R. D. Hays et al. 2006. Variations in provider conceptions of integrative medicine. Social Science and Medicine 62: 2973–87. Hutchinson, J. W. 1983. Expertise and the structure of free recall. In Advances in consumer research, Vol. 10, ed. R. P. Bagozzi and A. M. Tybout, 585–89. Ann Arbor, MI: Association for Consumer Research. Hutchinson, J. W., and G. R. Lockhead. 1977. Similarity as distance: A structural principle for semantic memory. Aluman Learning and Memory 6: 660–78. Ice, G. H., and J. Yogo. 2005. Measuring stress among Luo elders. Field Methods 17: 394–411. Jaskyte, K., and W. W. Dressler. 2004. Studying culture as an integral aggregate variable: Organizational culture and innovation in a group of nonprofit organizations. Field Methods 16: 265–84. Jaskyte, K., and W. W. Dressler. 2005. Organizational culture and innovation in non-profit human service organizations. Administration in Social Work 29: 23–41.

14_107-Bernard.indb 384

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

385

Johnson, A. 1995. A Guttman scale analysis of Matsigenka men’s manufacturing skill. Field Methods 7: 1–12. Johnson, J. C. 1990. Selecting ethnographic informants. Qualitative Research Methods Series, Vol. 22. Thousand Oaks, CA: Sage. Johnson, J. C., J. S. Boster, and L. Palinkas. 2003. Social roles and the evolution of networks in isolated and extreme environments. Journal of Mathematical Sociology 27: 89–122. Johnson, J. C., and D. C. Griffith. 1996. Pollution, food safety, and the distribution of knowledge. Human Ecology 24: 87–108. Johnson, J. C., and M. L. Miller. 1983. Deviant social positions in small groups: The relations between role and individual. Social Networks 5: 51–69. Johnson, J. C., and M. K. Orbach. 2002. Perceiving the political landscape: Ego biases in cognitive political networks. Social Networks 24: 291–310. Johnston, F. E., S. M. Low, Y. deBessa, and R. B. MacVean. 1987. Interaction of nutrition and socioeconomic status as determinants of cognitive development in disadvantaged urban Guatemalan children. American Journal of Physical Anthropology 73: 501–6. Jowell, R., C. Roberts, R. Fitzgerald, and G. Eva. 2007. Measuring attitudes cross-nationally. Lessons from the European Social Survey. London: Sage. Kaldjian, L. C., E. W. Jones, G. E. Rosenthal et al. 2006. An empirically derived taxonomy of factors affecting physicians’ willingness to disclose medical errors. Journal of General Internal Medicine 21: 942–48. Karabatsos, G., and W. H. Batchelder. 2003. Markov chain estimation for test theory without an answer key. Psychometrika 68: 373–89. Karlawish, J., F. K. Barg, D. Augsburger et al. 2011. What Latino Puerto Ricans and non-Latinos say when they talk about Alzheimer’s disease. Alzheimer’s & Dementia 7: 161–70. Kasulis, J. J., R. F. Lusch, and E. F. Stafford, Jr. 1979. Consumer acquisition patterns for durable goods. Journal of Consumer Research 6: 47–57. Kay, P. 1964. A Guttman scaling model of Tahitian consumer behavior. Southwestern Journal of Anthropology 20: 160–67. Kempton, W., J. S. Boster, and J. A. Hartley. 1995. Environmental values in American culture. Cambridge, MA: MIT Press. Kirk, J., and M. L. Miller. 1978. Cognitions of coca in Columbia, Ecuador, and Peru. In A multicultural view of drug abuse, ed. D. E. Smith, S. M. Anderson, M. Buxton, N. Gottlieb, W. Harvey, and T. Chung, 132–46. Cambridge, MA: Schenkman Publishing.  Kirk, L., and M. Burton. 1977. Meaning and context: A study in contextual shifts in meaning of Maasai personality descriptors. American Ethnologist 4: 734–61. Kiš, A. D. 2007. An analyisis of the impact of AIDS on funeral culture in Malawi. NAPA Bulletin 27: 129–40. Koster, J. M., J. J. Hodgen, M. D. Venegas, and T. J. Copeland. 2010. Is meat flavor a factor in hunters’ prey choice decisions? Human Nature—An Interdisciplinary Perspective 21: 219–42. Koster, J. M., and K. B. Tankersley. 2012. Heterogeneity of hunting ability and nutritional status among domestic dogs in lowland Nicaragua. PNAS 109: E463–E470. Krackhardt, D. 1987. Cognitive social structures. Social Networks 9: 109–34. Krackhardt, D. 1990. Assessing the political landscape: Structure, cognition, and power in organizations. Administrative Science Quarterly 35: 342–69. Kruskal, J. B., and M. Wish. 1990. Multidimensional scaling. Quantitative Applications in the Social Sciences Series, Vol 11. Thousand Oaks, CA: Sage. Lewis, C. E., J. M. Siegel, and M. A. Lewis. 1984. Feeling bad: Exploring sources of distress among pre-adolescent children. American Journal of Public Health 74: 117–22.

14_107-Bernard.indb 385

6/9/14 10:03 AM

386

Susan C. Weller

Lieberman, D., and W. M. Dressler. 1977. Bilingualism and cognition of St. Lucian disease terms. Medical Anthropology 1: 81–110. Loftus, E., and W. Marburger. 1983. Since the eruption of Mt. St. Helens did anyone beat you up? Improving the accuracy of retrospective reports with landmark events. Memory and Cognition 11: 114–20. Lopez, A., S. Atran, J. D. Coley et al. 1997. The tree of life: Universal and cultural features of folk biological taxonomies and inductions. Cognitive Psychology 32: 251–95. Lutz, C. 1982. The domain of emotion words on Ifaluk. American Ethnologist 9: 113–28. Lynch, E. B., and S. Holmes. 2011. Food group categories of low-income African-American women. Journal of Nutrition Education and Behavior 43: 157–64. Macauda, M. M., P. I. Erickson, M. C. Singer, and C. C. Santelices. 2011. A cultural model of infidelity among African American and Puerto Rican young adults. Anthropology & Medicine 18: 351–64. Magaña, J. R., M. Burton, and J. Ferreira-Pinto. 1995. Occupational names in three nations. Journal of Quantitative Anthropology 5: 1149–68. Magaña, J. R., G. W. Evans, and A. K. Romney. 1981. Scaling techniques in the analysis of environmental cognition data. Professional Geographer 33: 294–310. Mathevet, R. M. Etienne, T. Lynam, and C. Calvet. 2011. Water management in the Carmague Biosphere Reserve: Insights from comparative mental models analysis. Ecology and Society 16: 43. www.ecologyandsociety.org/vol16/iss1/art43/ES-2011-4007.pdf (accessed March 9, 2013). Mathez-Stiefel, S., and I. Vandebroek. 2012. Distribution and transmission of medicinal plant knowledge in the Andean highlands: A case study from Peru and Bolivia. EvidenceBased Complementary and Alternative Medicine, P1-18. http://www.hindawi.com/journals/ ecam/2012/959285/ (accessed March 9, 2013). Maupin, J. N., N. Ross, and C. A. Timura. 2011. Gendered experiences of migration and conceptual knowledge of illness. Journal of Immigrant and Minority Health 13: 600–608. McCarty, C. 2001. Structure in personal networks. Journal of Social Structure 3(1). http://www .cmu.edu/joss/content/articles/volume3/McCarty.html (accessed March 9, 2013). Meztger, D., and G. Williams. 1966. Some procedures and results in the study of Native categories: Tzeltal firewood. American Anthropologist 68: 389–407. Mezzich, J. E., and H. Solomon. 1980. Taxonomy and behavioral science. London: Academic Press. Miller, E. M. 2011. Maternal health and knowledge and infant health outcomes in the Ariaal people of northern Kenya. Social Science & Medicine 73: 1266–74. Miller, G. A. 1969. A psychological method to investigate verbal concepts. Journal of Mathematical Psychology 6: 169–91. Miller, M. L., and E. Hutchins. 1989. On the acquisition of boardsailing skill. In The content of culture: Constants and variants, studies in honor of John M. Roberts, ed. R. Bolton, 153–70. New Haven, CT: HRAF Press. Miller, M. L., and J. C. Johnson. 1981. Hard work and competition in an Alaskan fishery. Human Organization 40: 131–39. Milligan, G. W. 1980. An examination of the effect of six types of error perturbation of fifteen clustering algorithms. Psychometrika 45: 325–42. Miranda, T. M., M. C. de Mello Amorozo, J. S. Govone, and D. M. Miranda. 2007. The influence of visual stimuli in ethnobotanical data collection using the listing task method. Field Methods 19: 76–86. Moore, R., M. L. Miller, P. Weinstein et al. 1986. Cultural perceptions of pain and pain coping among patients and dentists. Community Dental Oral Epidemiology 14: 327–33. Morgan, D. L. 1996. Focus groups. Annual Review of Sociology 22: 129–52.

14_107-Bernard.indb 386

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

387

Nolan, J. M. 2002. Wild plant classification in Little Dixie: Variation in a regional culture. Journal of Ecological Anthropology 6: 69–81. Nunnally, J. C. 1978. Psychometric theory. New York: McGraw-Hill. Nyamongo, I. K. 2002. Assessing intracultural variability statistically using data on malaria perceptions in Gusii, Kenya. Field Methods 14: 148–60. Oravecz, Z., J. Vandekerckhove, and W. H. Batchelder. 2014. Bayesian cultural consensus theory. Field Methods. Pachter, L. M., S. C. Weller, R. D. Baer et al. 2002. Asthma beliefs and practices in mainland Puerto Ricans, Mexican-Americans, Mexicans, and Guatemalans: Consistency and variability in health beliefs and practices. Journal of Asthma 39: 119–34. Parr, M. G. 1996. The relationship between leisure theory and recreation practice. Leisure Sciences 18: 315–32.  Penka, S., H. Heiman, A. Heinz, and M. Schouler-Ocak. 2008. Explanatory models of addictive behavior among Native German, Russian-German, and Turkish youth. European Psychiatry 23: S36–S42. Quinn, N. 1987. Convergent evidence for a cultural model of American marriage. In Cultural models in language and thought, ed. D. Holland and N. Quinn, 173–92. Cambridge: Cambridge University Press. Reyes-Garcia, V., E. Byron, V. Vadez et al. 2004. Measuring culture as shared knowledge: Do data collection formats matter? Cultural knowledge of plant uses among Tsimane’ Amerindians, Bolivia. Field Methods 16: 135–56. Reyes-Garcia, V., C. C. Gravlee, T. W. McDade et al. 2010. Cultural consonance and body morphology: Estimates with longitudinal data from an Amazonian society. American Journal of Physical Anthropology 143: 167–74. Roberts, J. M., and G. E. Chick. 1979. Butler County eight ball: A behavioral space analysis. In Sports, games, and play: Social and psychological viewpoints, ed. J. H. Goldstein, 65–99. Hillsdale, NJ: Lawrence Erlbaum. Roberts, J. M., G. E. Chick, M. Stephanson, and L. L. Hyde. 1981. Inferred categories for tennis play: A limited semantic analysis. In Play as context, ed. A. B. Cheska, 181–95. West Point, NY: Leisure Press. Roberts, J. M., T. V. Golder, and G. E. Chick. 1980. Judgment, oversight and skill: A cultural analysis of P-3 pilot error. Human Organization 39: 5–21. Roberts, J. M., and S. H. Nattrass. 1980. Women and trapshooting: Competence and expression in a game of skill with chance. In Play and culture, ed. H. B. Schwartzman, 262–91. West Point, NY: Leisure Press. Rocha, J. 2005. Measuring traditional agro-ecological knowledge: An example from peasants in the Peruvian Andes. Field Methods 17: 356–72. Romney, A. K., W. H. Batchelder, and S. C. Weller. 1987. Recent applications of consensus theory. American Behavioral Scientist 31: 163–77. Romney, A. K., D. D. Brewer, and W. H. Batchelder. 1993. Predicting clustering from semantic structure. Psychological Science 4: 28–34. Romney, A. K., and R. G. D’Andrade. 1964. Cognitive aspects of English kin terms. American Anthropologist 66: 146–70. Romney, A. K., M. Keiffer, and R. E. Klein. 1979. A normalization procedure for correcting biased response data. Social Science Research 2: 307–20. Romney, A. K., C. C. Moore, and C. D. Rusch. 1997. Cultural universals: Measuring the semantic structure of emotion terms in English and Japanese. PNAS 94: 5489–94. Romney, A. K., T. Smith, H. E. Freeman et al. 1979. Concepts of success and failure. Social Science Research 8: 302–26.

14_107-Bernard.indb 387

6/9/14 10:03 AM

388

Susan C. Weller

Romney, A. K., and S. C. Weller. 1984. Predicting informant accuracy from patterns of recall among individuals. Social Networks 4: 59–77. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: A theory of cultural and informant accuracy. American Anthropologist 88: 313–38. Ross, J. L., S. L. Laston, P. J. Pelto, and L. Muna. 2002. Exploring explanatory models of women’s reproductive health in rural Bangladesh. Culture, Health, & Sexuality 4: 173–90. Ross, N., T. Barrientos, and A. Esquit-Choy. 2005. Triad tasks, a multipurpose tool to elicit similarity judgments: The case of Tzotzil Maya plant taxonomy. Field Methods 17: 269–82. Ross, N., and D. L. Medin. 2005. Ethnography and experiments: Cultural models and expertise effects elicited with experimental research techniques. Field Methods 17: 131–49. Ross, N., J. Maupin, and C. A. Timura. 2011. Knowledge organization, categories, and ad hoc groups: Folk medical models among Mexican migrant in Nashville. Ethos 39: 165–88. Ross, N., C. Timura, and J. Maupin. 2012. The case of curers, noncureres, and biomedical experts in Pichataro, Mexico resiliency in folk-medical beliefs. Medical Anthropology Quarterly 26: 159–81. Ruebush, T. K., II, S. C. Weller, and R. E. Klein. 1992. Knowledge and beliefs about malaria on the Pacific coastal plain of Guatemala. American Journal of Tropical Medicine and Hygiene 46: 451–59. Rumelhart, D. E., and A. A. Abrahamson. 1973. A model for analogical reasoning. Cognitive Psychology 5: 1–28. Ryan, G. W., and H. R. Bernard. 2003. Techniques to identify themes. Field Methods 15: 85–109. Ryan, G. W., J. M. Nolan, and P. S. Yoder. 2000. Successive free listing: Using multiple free lists to generate explanatory models. Field Methods 12: 83–107. Sayles, J. N., G. W. Ryan, J. S. Silver et al. 2007. Experiences of social stigma and implications for health care among a diverse population of HIV positive adults. Journal of Urban Health 84: 814–28. Schrauf, R. W., and E. Navarro. 2005. Using existing tests and scales in the field. Field Methods 17: 373–93. Schuman, H., and S. Presser. 1996. Questions & answers in attitude surveys. Experiments on question form, wording, and context. Thousand Oaks, CA: Sage. Schunko, C., and C. R. Vogl. 2010. Organic farmers use of wild food plants and fungi in a hilly area in Styria (Austria). Journal of Ethnobiology and Ethnomedicine 6: 17. Scott, J. 2000. Social network analysis: A handbook, 2nd ed. Newbury Park, CA: Sage. Smith, C. A. S. 2012. Living with sugar: Influence of cultural beliefs on type 2 diabetes selfmanagement of English-speaking Afro-Caribbean women. Journal of Immigrant Minority Health 14: 640–47. Smith, C. S., M. Morris, W. Hill et al. 2004. Cultural consensus analysis as a tool for clinic improvements. Journal of General Internal Medicine 19: 514–18. Smith, C. S., M. Morris, F. Langois-Winkle et al. 2010. A pilot study using cultural consensus analysis to measure systems-based practice performance. International Journal of Medical Education 1: 15–18. Sokal, R., and P. Sneath. 1963. Principles of numerical taxonomy. San Francisco: W. H. Freeman. Soutar, G. N., and S. P. Cornish-Ward. 1997. Ownership patterns for durable goods and financial assets: A Rasch analysis. Applied Economics 29: 903–11. Spradley, J. P. 1979. The ethnographic interview. New York: Holt, Rinehart and Winston. Stanton, B. F., R. Aronson, S. Borgatti, and J. Galbraith. 1993. Urban adolescent high-risk sexual behavior: Corroboration of focus group discussions through pile-sorting. AIDS Education and Prevention 5: 162–74.

14_107-Bernard.indb 388

6/9/14 10:03 AM

11: Structured Interviewing and Questionnaire Construction

389

Stefflre, V. J. 1972. Some applications of multidimensional scaling to social science problems. In Multidimensional scaling: Theory and applications in the behavioral sciences, Vol. 2, ed. A. K. Romney, R. Shepard, and S. B. Nerlove, 211–43. New York: Academic Press. Teddlie, C., and A. Tashakkori. 2009. Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioral sciences. Thousand Oaks, CA: Sage. Thompson, E. C., and Z. Juan. 2006. Comparative cultural salience: Measures using free-list data. Field Methods 18: 398–412. Trosset, C., and D. D. Caulkins. 2001. Triangulation and confirmation in the study of Welsh concepts of personhood. Journal of Anthropological Research 57: 61–81. Trotter, R. T., II, S. C. Weller, R. D. Baer et al. 1999. Consensus theory model of AIDS/SIDA beliefs in four Latino populations. AIDS Education and Prevention 11: 414–26. Truex, G. F. 1977. Measurement of intersubject variations in categorizations. Journal of CrossCultural Psychology 8: 71–82. Vogl, C. R., B. Vogl-Lukasser, and R. K. Puri. 2004. Tools and methods for data collection in ethnobotanical studies of homegardens. Field Methods 16: 285–306. Wasserman, S., and K. Faust. 2009. Social network analysis. Cambridge: Cambridge University Press. Webster, C. M., A. L. Iannucci, and A. K. Romney. 2002. Consensus analysis for the measurement and validation of personality traits. Field Methods 14: 46–64. Weller, S. C. 1983. New data on intra-cultural variation: The hot-cold concept. Human Organization 42: 249–57. Weller, S. C. 1984. Cross-cultural concepts of illness: Variation and validation. American Anthropologist 86: 341–51. Weller, S. C. 1987. Shared knowledge, intercultural variation, and knowledge aggregation. American Behavioral Scientist 31: 178–93. Weller, S. C. 2007. Cultural consensus theory: Applications and frequently asked questions. Field Methods 19: 339–68. Weller, S. C., R. D. Baer, J. E. Garcia de Alba et al. 2002. Regional variation in Latino beliefs about susto. Culture, Medicine, and Psychiatry 26: 449–72. Weller, S. C., R. D. Baer, J. E. Garcia de Alba, and A. L. Salcedo Rocha. 2012. Mexican and Mexican-American explanatory models of diabetes: A comparison of community members, patients, and physicians. Social Science & Medicine 75: 1088–96. Weller, S. C., and C. I. Dungy. 1986. Personal preferences and ethnic variations among Anglo and Hispanic breast and bottle feeders. Social Science and Medicine 23: 539–48. Weller, S. C., L. M. Pachter, R. T. Trotter, II, and R. D. Baer. 1993. Empacho in four Latino groups: A study of intra- and inter-cultural variation in beliefs. Medical Anthropology 15: 109–36. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Qualitative Research Methods Series, Vol. 10. Thousand Oaks, CA: Sage. Weller, S. C., and A. K. Romney. 1990. Metric scaling: Correspondence analysis. Quantitative Applications in the Social Sciences Series, Vol. 75. Thousand Oaks, CA: Sage. Weller, S. C., A. K. Romney, and D. P. Orr. 1987. The myth of a sub-culture of corporal punishment. Human Organization 46: 39–47. Weller, S. C., T. R. Ruebush, II, and R. E. Klein. 1997. Predicting treatment-seeking behavior in Guatemala: A comparison of the health services research and decision-theoretic approaches. Medical Anthropology Quarterly 11: 224–45. Young, F. W., and R. C. Young. 1962. Key Informant reliability in rural Mexican villages. Human Organization 20: 141–48.

14_107-Bernard.indb 389

6/9/14 10:03 AM

390

Susan C. Weller

Young, J. C. 1978. Illness categories and action strategies in a Tarascan town. American Ethnologist 5: 81–97. Young, J. C. 1980. A model of illness treatment decisions in a Tarascan town. American Ethnologist 7: 106–31. Young, J. C., and L. Y. Garro. 1982. Variation in the choice of treatment in two Mexican communities. Social Science and Medicine 16: 1453–65.

14_107-Bernard.indb 390

6/9/14 10:03 AM

CHAPTER 12

Discourse-Centered Methods Brenda Farnell and Laura R. Graham

Discourse analysis comprises methods used by researchers across the social sciences as well as in literary studies. It is also used in clinical fields (see references in Wilce 2009). The term “discourse” is like a Rorschach—it might seem to provide a common referent, but it often brings out discrepant understandings depending on the disciplinary and theoretical locations in which it is employed (see Briggs 2001). For example, scholars influenced by postmodern and poststructuralist theories often follow Michel Foucault’s treatment of discourse as “practices that systematically form the objects of which they speak” (1972, 49) and thus focus on “political technologies” and the legitimization of institutional power rather than on detailed analyses of language use in social contexts. Likewise, critical discourse analysis distinguishes an approach that combines linguistic theory with social theory to examine discourse and power in social and political processes (Fairclough 1989, 1995a, 1995b, 2003; Gee 2011a [1999], 2011b; Jaworski and Coupland 2002; van Dijk 1991, 1993, 2008). In linguistic anthropology, the focus on discourse has fostered new engagements with broader anthropological issues, bringing evidence and deeper insight to social practices that constitute and perpetuate ideologies, social relations, hierarchies, political economies, and power relations. This development bridges an earlier separation between Foucauldian-inspired approaches and micro-analyses of discourse as socially situated practice (Briggs 2001; see also Briggs and Mantini-Briggs 2003). Discourse-centered approaches at the heart of linguistic anthropology increasingly influence sociocultural anthropology, particularly as the theoretical focus has shifted away from structure and toward practice. The term “discourse centered” (Sherzer 1987; Urban 1991) is an expedient label with which to describe movements in linguistic anthropology since the late 1980s. It does not, however, constitute a single theoretical school. Discourse-centered approaches to culture draw on theoretical resources from several intellectual ancestries and build on earlier work in sociolinguistics, ethnomethodology, the ethnography of speaking/ communication, literary theory, and performance approaches to language. These approaches distinguish themselves by focusing on the dialogical processes through which persons, social institutions, objects, and cultural knowledge are socially constructed through spoken discourse, textualizing practices, and other related signifying acts or forms of expressive performance. In current anthropological practice, discourse-centered methods are used in participant observation (or observant participation; see Moeran 2007; Whitehead 2009) 391

14_107-Bernard.indb 391

6/9/14 10:03 AM

392

Brenda Farnell and Laura R. Graham

in the context of ethnographic fieldwork. Within this general orientation, researchers pay close attention to how language is used in and across social situations, focusing particularly on utterances that occur in the context of social interaction, including utterances specifically elicited by the linguist or ethnographer, and interviews. This also includes forms of virtual discourse such as those found in social networking (chat rooms, blogging), texting, and other Internet/digitally mediated communication (see, e.g., Gershon 2010; Hill 2008; Keating and Mirus 2003; Keating and Sunakawa 2010) as well as the processes by which speech becomes text (Fengisen 1999; Graham 2011; Haviland 1996; Urban 1996a). A central proposition of discourse-centered research in linguistic anthropology is that “culture is localized in concrete publicly accessible signs, the most important of which are actually occurring instances of discourse” (Urban 1991, 1). That is, culture and self are viewed as actually constituted by speech and other signifying acts. This contrasts with some recent work in cognitive anthropology that uses discourse (talk) as data in the service of reconstructing cultural schema (e.g., Quinn 2005). Edward Sapir prefigured discourse-centered approaches when he wrote: “The true locus of culture is in the interactions of specific individuals and, on the subjective side, in the world of meaning which each one of these individuals may unconsciously abstract for himself from his participation in these interactions” (1949, 515). Discourse-centered work thus emphasizes the heterogeneous, multifunctional, and dynamic character of language use and the central place it occupies in the social construction of reality. According to a discourse-centered framework, culture is an emergent process, historically transmitted but continuously produced and revised through dialogues among its members. It is constantly receptive to new associations and interpretive moves. Such perspectives open up new discursive spaces within the discipline of anthropology that embody Ferdinand de Saussure’s original vision of a scientific study of “the functioning of signs within social life” (Saussure 1916, 33; see also Mertz and Parmentier 1985; Thibault 1997). Since new theoretical insights and questions give rise to methodological challenges, in this chapter we examine methods of research in light of the theoretical developments that gave rise to them. As Harré notes, “a theory determines where, in the multiplicity of natural phenomena, we should seek for its evidence” (1986a, 83). In our particular case, increasing theoretical attention to language in social contexts and an emerging understanding of discursive practices as constitutive of culture, have required, and are reinforced by, new ways of collecting and analyzing data. For example, discourse-centered approaches contrast with the formalist logicism and mentalism that define the notion of grammar characteristic of Chomskian generative linguistics. In that paradigm, linguistic utterances are analyzed as if they are generated by an idealized, individual, speaker–hearer independent of sociocultural context. Linguistic anthropologists have long been critical of this view of language (see Hymes [1964, 1971], and, for a cogent critique, Ellis [1994]). When discourse is considered from the point of view of language use in social contexts, the assumption that there is a single underlying abstract grammar or that languages can be considered as discrete and monolithic codes become questionable. Linguistic anthropologists find multiple linguistic varieties and speech styles existing simultaneously within a culture and deem

14_107-Bernard.indb 392

6/9/14 10:03 AM

12: Discourse-Centered Methods

393

traditional notions of descriptive grammar and linguistic structure inadequate (see Hymes 1971; Jakobson 1960; Silverstein 1976; Urciuoli 1996; Woodbury 1987). Many researchers using discourse-centered approaches consider the rules of grammar to be immanent in conversational and other expressive practices, not located in “mental structures” (Harré and Gillet 1994). A widely held position in discourse-centered research is succinctly captured in the statement, “People use rules to assess the correctness of their actions; rules do not use people as the vehicles of their causal efficacy to generate actions” (paraphrase of Wittgenstein in Mühläusler and Harré 1990, 7). Discourse-centered approaches don’t focus on language forms alone, but also on forms and processes. “Meaning does not inhere to words or grammar alone, and language function—the social meaning of what people say—goes well beyond the semantic-referential function or dictionary sense of meaning” (Urciuoli 1996, 5). In discourse-centered approaches, the referential or denotative (word and grammar meaning, independent of context) is but one aspect of linguistic function, such that attention to language as code (i.e., English, Spanish, Kwakiutl, etc.) is not the primary or sole focus of analysis. The principal objectives are to discover the social meanings inhering in language forms and their relationship to social formations, identity, relations of power, beliefs, and ideologies. Below, we provide a brief overview of discourse-centered approaches and point out their utility for addressing issues that concern anthropologists and social theorists. Later, we discuss methods and methodological issues related to discourse-centered work. WHY STUDY DISCOURSE?

Discourse-centered approaches in linguistic anthropology seek to dissolve the longstanding dilemma in Western social theory of how to connect social structure and/ or culture with individual human agency. The problem has been labeled variously as a tension between structure/event (Rosaldo 1980, 109), individual action/collective representations (Sahlins 1985, 108), structure/practice (Bourdieu 1977a), and structure/ agency (Giddens 1984). The enduring problem has been how to avoid the twin errors of individualism and the reification of internal unconscious mental structures, on the one hand (e.g., Freud, Chomsky, Lévi-Strauss, and Searle), and collectivism and the reification of external social structures, on the other (e.g., Durkheim and variations). These positions have been exposed as violating the logic of causal powers and, as a result, mislocating human agency (see Farnell 2012; Farnell and Varela 2008; Varela 1995; Varela and Harré 1996). When such psychological and sociological determinisms are abandoned and persons are viewed instead as causally empowered embodied agents with unique powers and capacities for making meaning, discursive practices emerge as the very means by which social action, cultural knowledge, and social institutions are achieved and enacted (Farnell 1994, 1995b, 2012; Graham 1995; Harré and Gillett 1994; Silverstein and Urban 1996; Urban 1996b; Urciuoli 1995; Williams 1982). Discourse-centered approaches thus offer social theorists a number of principles that should prove central to their goal of achieving more dynamic views of culture and the relationships between the individual and society (e.g., Bourdieu 1977a, 1977b, 1991; Giddens 1979, 1984; Sahlins 1976, 1985). Discourse analysis also provides a means to understand the precise

14_107-Bernard.indb 393

6/9/14 10:03 AM

394

Brenda Farnell and Laura R. Graham

mechanisms of power and authority “that perpetuate social injustices and ways that language can be pressed into the service of hegemonic agendas” (Graham 2011, 178; also Fenigsen 1996; Hill 2008). Earlier generations of ethnographers necessarily acquired much of their knowledge through the discursive practices of their informants. Yet many ethnographers tended to overlook the fact that the discursive forms in which this knowledge is packaged are themselves unique cultural forms. In traditional ethnography, Joel Sherzer and Greg Urban note: Discourse is invisible, a glass through which the ethnographer comes to perceive the reality of social relations, of ecological practice, and of belief. Little attention is given to the glass itself. We are rarely informed about the structure of the discourse through which knowledge is produced, conceived, transmitted and acquired by members of societies and by researchers. (1986a, 1–2)

The theoretical presuppositions that inform discourse-centered methods are a shift away from approaches to ethnography that define language and culture as transcendent realms or abstract systems that interact with or determine social practices or as preexistent patterns imperfectly expressed by each individual participant. Instead, discourse-centered perspectives present a view of language use as social action and cultural events as the sites where culture emerges from interaction among participants (Mannheim and Tedlock 1995, 2). For example, the dialectic between linguistic performances and wider sociocultural and political–economic contexts has been explored (see McElhinny 2007). Investigators are also interested in how individuals gain rights to particular modes of transforming discourse, or are denied such rights, and how the signifying acts of those with status and power achieve performativity (i.e., are effective) while those of others fail. Investigating how these kinds of processes operate situates languagein-culture in new ways. Positioning discursive practices as a central locus of cultural creativity opens up new possibilities for studying how processes of cultural continuity and cultural change take place. Many discourse-centered researchers also take the position that persons constitute themselves as a “self,” an embodied moral unit in the world, through discursive practices of reflexive talk. Keith Basso (1979) offers an exemplary ethnographic study of ways in which discourse is central to the cultural construction of self (see also AbuLughod 1993; Harré 1986b; Harré and Gillett 1994; Lutz 1988; Mühlhäusler and Harré 1990; Rosaldo 1982; White and Kirkpatrick 1985). The works of Donald Brenneis (1984, 1986), John Du Bois (1993 [1987]), Alessandro Duranti (1986), Laura Graham (1993), and Webb Keane (1991) also show how the “I” of discourse (Urban 1989) can project different selves and how the organization of talk can be linked to notions of the construction of truth and individual accountability. This reorients theories of person, self, and agency away from an ethnocentric, individualistic psychologism toward sociocultural dimensions of communication, cross-cultural variability (Hill and Irvine 1993), and enactment of indexical dynamics (Urciuoli 1996). The assigned locus of meaning shifts correspondingly from internal (private) mental structures (e.g., the Freudian unconscious or Searle’s psychological

14_107-Bernard.indb 394

6/9/14 10:03 AM

12: Discourse-Centered Methods

395

states) and the individual speaker/signer toward the dialogic, interactional processes within which meanings and identities are constructed. PERFORMATIVITY, INDEXICALITY, AND METAPRAGMATICS

In addition to resources provided by sociolinguistics and ethnnomethodology (e.g., Fishman 1970; Garfinkel 1967; Gumperz 1982; Labov 1972), the ethnography of speaking (Bauman and Sherzer 1989 [1974]; Gumperz and Hymes 1986 [1972]; Hymes 1962, 1964; and performance approaches to language (Bauman 1977; Hymes 1975), a second set of resources used in discourse-centered approaches draws on the linguistic turn in philosophy (see Rorty 1992 [1967]; Taylor 1985). This move contributed Ludwig Wittgenstein’s post-Cartesian emphasis on the social construction of mind, language, and emotion to linguistic anthropology (see Ahearn 2001; Dureau 2012; Harré 1986b; Lutz 1988; Lutz and Abu-Lughod 1990; Rosaldo 1982). Wittgenstein’s insights have since been complemented by the social constructionism of Vygotsky and G. H. Mead in a relocation of agency toward a discursive view of mind and person (see Harré 1984, 1986a, 1987; Harré and Gillett 1994). Also from philosophy, Austin’s speech act theory (1962) introduced speech acts and the notion of “performativity” into the study of language practices. Ethnographically grounded critical discussions of speech act theory stimulated important challenges to Western philosophical assumptions (see Bauman and Briggs 1990; Pratt 1987; Rosaldo 1982). The central insight that language is performative (i.e., utterances do not simply refer to real world events but create them, and saying is a form of doing—it achieves social action) has remained important. This has been developed further through Michael Silverstein’s (1976, 1977) employment of Charles S. Peirce’s ideas about indexical signs and Jakobson’s discussion of “shifters”—words that shift their reference depending on the context in which they are used (Jakobson 1960, 1971 [1957]; Silverstein 1977, see also Verschueren 1985, 2004). Emile Benveniste (1971b [1956]) also recognized an important connection between indexicality in personal pronouns and performativity in certain verbs (Benveniste 1971a [1956]), thereby also drawing attention to the construction of person through discourse. Silverstein argues convincingly that the preoccupation with the “semanticoreferential” or denotative function of language (i.e., where the meaning of words derives from their naming things and their relationships to other words, as in dictionary definitions), provided the basis for a uniquely biased Western linguistic ideology, a “referentialist ideology” (Hill 2008). In this ideology, other functions of language, especially “nonreferential” indexical functions (ways in which linguistic utterances also carry social/pragmatic meanings through spacio-temporal contiguities to the social context) are accorded secondary importance. Prague School linguists and the Russian formalists both advanced theories of the multifunctionality of language. Until Jakobson’s work became influential in Western linguistic circles, however, these ideas were little known outside of the former Soviet Union (see Hymes 1975). Bakhtin (1981) and Voloshinov (1973) contributed complementary concepts from Soviet literary studies. Utilizing these resources, Silverstein and Bonnie Urciuoli emphasize that the defining properties of meaningful action are precisely those not visible in a grammaticalsemantic model of language, the units and rules of which are essentially timeless.

14_107-Bernard.indb 395

6/9/14 10:03 AM

396

Brenda Farnell and Laura R. Graham

Urciuoli (1996) synthesizes insights from Jakobson, Hymes, and Silverstein to describe exactly how indexical language is constitutive of self and social action: The property that language shares with all sign systems is its indexical nature: its maintenance and creation of social connections, anchored in experience and the sense of the real. Linguistic indexes may be grammaticalized or lexicalized as “shifters”—devices that locate actions in time and space: personal pronouns, verb tenses, demonstratives, and time and space adverbs. These are deictic in that they point outward from the actor’s location. The structure of action fans out from the center, the locus of I and you, to delineate where and when everything happens relative to the central actors: he and she versus I and You; there versus here; then versus now; present versus non-present (past or future). . . . The indexes that embody discourse extend beyond pronouns, adverbs and verbal categories both to the sounds and shapes of speech that identify the actor with a particular group and to the speech acts marking the actor’s intent as others recognize it. In short indexes make the social person. (p. 190)

This interest has prompted ethnographic studies of spatial and temporal deixis as well as the social functions of pronouns. (For ethnographic studies of deixis, see Farnell 1995a; Hanks 1990; Haviland 1993). Fundamental connections between indexicality and performativity are at work because: Indexicality, a sine qua non of language, may be more or less creative: the more an index creates the who, what, when and where of the action, the more performative it is. The more an index maintains the status quo of action, the more presupposed (and less performative it is). Performativity may be thought of as a process that sometimes surfaces as an explicit formula (commands, promises, etc.) but it is more often implicit. Any index can be performative, depending on the dynamics of the context. (Urciuoli 1995, 190; see also Silverstein 1977)

Formal discourse features also frequently carry indexical value. For example, speakers may use reported speech as a means to distance themselves from the content of their utterances or it may be used to add the moral weight of other voices to the speaker’s own (see, e.g., Bakhtin 1986; Hill and Irvine 1993, 6–7). Speakers can use evidential particles to index their orientation to a message (see papers in Hill and Irvine 1993; see also Basso 1995; Chernela 2011). The analysis of person markers and other deictics has been particularly fruitful in the studies of issues of the social construction and representation of identity and personhood. Formal pragmatic devices such as honorifics frequently convey significant information about social relationships between individuals (see Agha 1996; Brown and Gillman 1960; Irvine 1992; Pressman 1997). Ethnographic studies that use these resources have focused attention on such topics as political economies of language (see Irvine 1989), language ideologies (Silverstein 1976; Woolard 1989, 1992; also Kroskrity et al. 1992), and conflict resolution (Briggs 1996; Murphy 1990; Urciuoli 1996; WatsonGegeo and White 1990; see also Bauman and Briggs 1990 for further discussion). Silverstein (1993) brings attention to the importance of metapragmatic elements and functions in language use. Metapragmatic refers to linguistic signs that point to extra linguistic meanings that may be indirectly encoded in speech and examines how

14_107-Bernard.indb 396

6/9/14 10:03 AM

12: Discourse-Centered Methods

397

such elements come to be interpreted by social actors in specific contexts. This concept also provides analytic tools to describe and analyze the frequently implict metadiscursive aspects of socially situated language (i.e., how talk should be interpreted), bringing greater dynamic precision to concepts such as Goffman’s “frames” and Bateson’s “metacommunication” (Urban 2006). Anthony Webster (2010), for example, productively uses metadiscourse to examine how a Navajo poet both dramatizes language use and communicates implicit messages about the importance and affective power of continued use of the Navajo language within the Navaho nation. A third important source of theoretical insights has been conversation analysis grounded in the symbolic interactionism of G. H. Mead, Goffman’s microsociology (1972 [1967], 1974, 1981), and the sociological study of conversation (sociolinguistics, ethnomethodology), with its emphasis on the negotiated and emergent quality of meanings and the interactional construction of social structure and institutions. Researchers who specialize in conversational analysis, a tradition that developed largely within sociology with the work of Sacks and Schegloff, focus especially on cultural strategies that make conversations function effectively or not (see, e.g., Sacks et al. 1974; Schegloff 1972; Schegloff and Sacks 1973; see also Fishman 1968, 1970; Garfinkel 1967, 1972; C. Goodwin and Heritage 1990; C. Goodwin 1981, 2006; Grimshaw 1990, 1994; Gumperz 1982; Tannen 1982, 1984, 1994). These studies note strategies for turn-taking, conversational overlap, utterance length (holding the floor), patterns of interruption, conversational “repair,” and participation frameworks (see, e.g., Goodwin and Goodwin 2006). This line of inquiry has been fruitful in analyzing ways in which gender roles are created and enacted through talk (see, e.g., M. Goodwin 1990, 2006; see also papers in Tannen 1994) as well as highlighting the social conventions that structure such interactions. Although a great deal of work in conversational analysis focuses on patterns of communication among English speakers, this approach has also been fruitful in cross-cultural contexts (e.g., Moerman 1988; Mori 1999; Testa 1988). Although conversation analysis in the sociological tradition of Sacks et al. (1974) has been largely context free, anthropologists have added context, thereby enabling this work to speak to broader social issues and anthropological concerns (e.g., Sidnell 2009, 2010; Sidnell and Stivers 2012; Zeitlyn 2004). For example, the co-construction of meaning is especially important in analyses of discourses of power such as medical encounters, “denying, for instance that doctors unilaterally impose diagnoses or therapies” (Wilce 2009, 206). LANGUAGE IDEOLOGIES AND REFLEXIVE DIMENSIONS

Discourse-centered approaches also seek to understand the ideologies (beliefs and attitudes) speakers maintain toward language in general, specific languages, and varieties of languages used within their social sphere. Preceded by Fishman’s work on language loyalty and attitudes in the United States (1966) and Silverstein [1979; 2001 [1981], as well as Bauman (1983), in the late 1990s language ideologies and their social consequences became a distinct field of inquiry (see Briggs 2005; Inoue 2004; Makihara and Schieffelin 2007; Schieffelin et al. 1998; Woolard 1992, 1998; Woolard and Schiefflin 1994).

14_107-Bernard.indb 397

6/9/14 10:03 AM

398

Brenda Farnell and Laura R. Graham

Studies show that language ideologies have significant effects on speakers and social relations. For example, Hill (1993, 2008) shows that prevailing language ideologies mask discriminating effects when Euro-American speakers use Hispanicized words and phrases in English speech (“mock Spanish”) and elevates the social standing of its speakers. This, like the Hollywood “Indian English” Meek (2006) describes, perpetuates negative social stereotypes and leads to social discrimination without sanction. Irvine and Gal’s (2000) concept of iconization illuminates the processes by which this occurs: In iconization, a dialectic process occurs in which social attitudes toward speakers transfer onto their language or linguistic features and vice versa. For example, in contemporary U.S. society, since Deaf people are often considered to be defective by the general population, the process of iconization effectively exposes how their “defects” are transferred to American Sign Language. The language itself is then understood by many as being a defective and inferior code (Calton 2012). See also Senghas and Monaghan (2002) for review of literature on Deaf signing communities. Language ideologies also play a significant role in education and language policy (see Lippi-Green [1997] on the English Only movement and legislation in the United States; Heller [2011] on language politics in Quebec, Canada; and Jaffe [1999] on language ideology and politics in Corsica). Language ideologies are also operative in language revitalization efforts. Meek (2007) demonstrates how traditional ideologies that prescribe “correct” speech in the presence of elders inhibit American Indian children in immersion programs from interacting with elders who are fluent speakers of the native language that the programs are designed to maximize. Hill (2002) reminds us that the notion of endangered languages as well as those engaged in language revitalization efforts are themselves steeped in language ideologies (see also Perley 2011). Discoursecentered approaches recognize that language study and representation of scholarship are themselves political and ideologically saturated (Briggs 2001; see also French 2010; Irvine and Gal 1998). Recognizing that language ideologies are as much at work in our own disciplinary discourse as in the discourse practices of the peoples we study, many discoursecentered researchers explore reflexive dimensions of their research and seek to deconstruct and challenge dominant Western conceptions of language and social life (Bauman and Briggs 2003; Lucy 1990). For example, several ethnographic studies challenge Western “personalist” (individualist) notions of discourse (Hill 2008; Holquist 1983), in which the individual is understood to be solely responsible for utterance production (e.g., Searle 1969). These studies reveal language ideologies in which discourse is thought to be the product of multiple speakers (Brenneis 1984, 1986; Chafe 1993; Du Bois 1993 [1987]; Duranti 1986, 1993; Graham 1993; Hill and Zepeda 1993; Irvine 1996; Keane 1991; Rosaldo 1982). Reflexive studies thus seek to expand traditional definitions of what counts as discourse, discourse production, and interpretive processes in ways that allow a better understanding of indigenous classifications. Studies that include consideration of complementary vocal acts such as wailing, chanting, laughter, the use of sound symbolism, the blurring of boundaries between music and speech, and the coproduction of utterances expand the traditional boundary within the modality of socially produced and meaningful sound (Brenneis 1986; Duranti 1986; Feld 2012 [1982]; Fox 2004; Graham 1993; Hill 2009; Lederman 1963;

14_107-Bernard.indb 398

6/9/14 10:03 AM

12: Discourse-Centered Methods

399

Nuckolls 1992, 2010; Seeger 2004b [1987]; Urban 1988). The concept of soundscape (Schafer 1977, 1994), as humanly produced sound in relation to the natural environment (e.g., Feld 2012 [1982]; Hirschkind 2006) as well as socially organized production of human sound (e.g., Graham 1995), makes important connections between linguistic anthropology and related fields such as ethnomusicology and the anthropology of the senses. We return to soundscape below. Likewise, ethnographic studies of sign languages and the coproduction of speech and gesture seek to make problematic the reductionism inherent in the conflation of discourse with speech. Including the modality of bodily action into what counts as discourse challenges the traditional Western division of communicative practices into verbal and nonverbal. Language ideologies exist in which discourse is conceived as the simultaneous production of vocal signs and action signs (Farnell 1995a, 2001; see also Kendon 1989; McNeill 1985, 1992, 2000; Streeck et al. 2001; Wilce 2009; and the section Embodiment of Discourse below). In this critically reflexive arena, metalevel discourses on language and social life provided by native speakers are no longer simply sources of data for the presentation of a problem-free indigenous perspective. Speakers become intellectual collaborators who can make substantial theoretical contributions. Greater reflexive awareness of the very processes of interviewing, recording, and analyzing have been called for and demonstrated (Ardener 1989 [1978]; Bauman and Briggs 1990; Briggs 1986, 2007; Farnell 1994; Quinn 2005; Urban 1996a). CONTEXT AND CIRCULATION: ENTEXTUALIZATION, RECONTEXTUALIZATION, AND ENREGISTERMENT

Expanding what counts as discourse beyond “socially situated speech” to “socially situated speech and bodily action” (Farnell 2012; Kendon 1980, 1983; Streeck et al. 2011) prompts reflection about what an adequate theory of context might look like (see Goodwin and Duranti 1992). Clearly, a theory of context must take into account the co-presence of other semiotic modalities such as the spatial (see Keating 2000). Conceptions of context must also be elaborated to account for those events taking place elsewhere in space and/or time that give meaning to novel events and actions. Emergent performances are necessarily situated in cultural histories and long-term patterns of action. Performances are situated within the context of histories of performance and cultural knowledge. Actors also bring individual perspectives to the situations in which they operate, and a context analysis must account for the strategies participants use “to constitute the culturally and historically organized social worlds that they inhabit” (Duranti and Goodwin 1992, 5). This means defining context “from the inside” rather than, as in traditional approaches to folklore, considering context “from the outside in”; context is not some kind of “surround” that exerts influence on expressive forms and texts (Bauman 1992, 142). It is agent-centered, emergent, and may vary across participants in a given event. For example, Davies and Harré (1990) introduce the dynamic concept of “positioning” to replace Goffman’s (1981) concept of “footing,” “alignment,” or “stance” to understand better how social actors position and reposition themselves in conversations according to their personal histories and goals, noting that actors are

14_107-Bernard.indb 399

6/9/14 10:03 AM

400

Brenda Farnell and Laura R. Graham

simultaneously positioned by others according to factors such as class, ethnicity, and gender (Davies and Harré 1990). Studies considering contextual dimensions of socially situated discourse have provided a foundation for further analyses of processes first identified by Bahktin and Voloshinov regarding the ways that discourse moves between contexts or circulates within and between social domains (Bakhtin 1981, 1986; Voloshinov 1973). Urban (2001), for example, considers the ways that discursive elements accelerate or impede the circulation of cultural ideas and forms. New concepts such as intertextuality, entextualization, de- and re-contextualization, and enregisterment have emerged to describe more precisely the processes and effects of the circulation and reinscription of discourse in and between various modalities (Agha 2007; Bauman 2004; Bauman and Briggs 1990; Silverstein and Urban 1996). Discourse is highly intertextual or dialogic (Bahktin 1981); that is, any present utterance is “replete with echoes, allusions, paraphrases, and outright quotations of prior discourse” (Mannheim and Tedlock 1995, 7). Utterances and narratives typically reference other texts, and pieces of previous discourse are frequently inserted into newly created texts (see Bauman 1992, 2004). Often references to prior discourse take the form of direct or indirect quotations (see, e.g., Bakhtin 1981; Briggs 1992; Lucy 1990; Silverstein 1985; Tannen 1995; Urban 1984). Entextualization, decontextualization, and recontextualization are processes that connect discourse with wider sociocultural concerns (see Bauman 2004; Bauman and Briggs 1990; Goodman 2005; Silverstein and Urban 1996). Entextualization refers to the processes by which utterances (words, phrases, or larger stretches) are rendered detachable and repeatable in new contexts. They can become recontexualized, carrying social meanings from one situation to another, via processes of iconicity and indexicality (Irvine and Gal 2000). For example, Graham’s (2011) analysis of ways in which news editors recontextualize quoted speech in print underscores ways that these practices are imbued with power and authority. These entextualized pieces of discourse contribute to the creation of emergent meanings in novel contexts (see Bauman and Briggs 1990; Ochs 1979a; also Bauman 2004; Graham 2000; Kuipers 1990; Silverstein and Urban 1996; Spitulnik 1996; Urban 1996a). Indeed, previous discourse is a further dimension of context because discourse itself is the source of much presupposed knowledge. Registers, repertoires of linguistic forms understood locally, emerge and become recognizable through processes of “enregisterment” (Agha 2007). Shared registers allow people to see themselves as “talking about the same thing,” because they are using the same referring expressions, but the indexical ordering (Silverstein 2003) of those expressions is subject to considerable variation across registers. For example, as Urciuoli (2009) points out, the term “culture” is enregistered quite differently in semiotic anthropological discourses than in higher education or corporate or government diversity discourse. Wilce (2009, 208) notes that psychological discourses have rapidly disseminated through popular genres such as the advice column. For example, a Bangladeshi advice column is contributing to the enregisterment of a “psychiatric Bangla,” that is, the sedimenting of a psychiatric register.

14_107-Bernard.indb 400

6/9/14 10:03 AM

12: Discourse-Centered Methods

401

METHODS: DATA COLLECTION

The ethnographic focus on socially situated discourse contrasts with earlier methods for language study in American anthropology as well as those adhered to by contemporary descriptive linguists. Two methods for studying language that became central to the Boasian tradition of the study of North American Indian languages are text collection and elicitation (see, e.g., Boas 1940; Radin 1949). Historically, anthropological linguists took written texts as the starting point for analyses. A fieldworker who analyzed grammatical structure based on the patterns that emerged in the texts typically elicited these texts from native speakers. Before the use of tape recorders, this process necessarily involved a stop–start process of dictation, which destroyed the natural flow of a storytelling performance or other speech event, and texts were usually collected outside of normal cultural contexts of narration. Such limited technological means necessarily precluded any detailed attention to the role of socially situated discourse within social life. Analytical goals centered on grammatical structure, the mythological, and other ethnographic content of narratives. Throughout the late 1970s and early 1980s, the emphasis on narrative performance of verbal art directed attention away from the study of formal patterning and symbolic content of texts typical of the Boasian tradition and an earlier poetics. The focus instead was on the emergence of verbal art in social interaction and fine-grained analyses of speech events. This interest connected with a long tradition of thinking about language and society that emerges clearly in the writings of Vico, Herder, and von Humboldt, who argue that verbal art is a central dynamic force in shaping linguistic structure and study. Attention to poetics from Sapir, the Russian formalists, and members of the Prague School contributed to the development of a focus on performance and poetics in the 1970s and ’80s (Bauman and Briggs 1990, 59). A second method of early data collection involved eliciting words and sentences from a native speaker and transcribing the sounds into a written form. Usually, the anthropological linguist worked with one or a small number of informants. Each utterance produced by the native speaker was seen as an instantiation, a token, of the abstract code or language. Through elicitation of words and sentences, the investigator tried to determine the structure of the grammar underlying the spoken utterances. In this way, it was possible to determine and describe such things as the sounds of a language (its phonetic inventory), the way these sounds combined (the phonology), and syntactical patterns (word order, grammatical particles). Descriptive linguists, who aim to describe language structure, still use this methodology to produce dictionaries and grammars. Although discourse-centered research incorporates both text-focused collection and methods of analysis, and elicitation, it differs from these earlier approaches by taking socially situated discourse and textualizing practices as the starting point for analysis. Consultants’ comments on recorded texts provide rich contextualizing information that can be used to generate further questions and elicitations. While discourse-centered research incorporates elicitation as a fieldwork technique, such elicitation usually supplements socially situated utterances: Elicitation and interviewing generally follow and support the former. Elicitation is limited because, although it

14_107-Bernard.indb 401

6/9/14 10:03 AM

402

Brenda Farnell and Laura R. Graham

frequently enables researchers to determine basic grammatical patterns, when and why social actors use certain forms in specific situations for what kind of pragmatic ends remain to be investigated. Interviews, too, can supplement direct observation by providing important information. (For an excellent discussion of problems in interview-based research, see Briggs 1986; Quinn 2005; Urban 1996). Briggs (2007) reminds us that interviewing is informed by ideological constructions of discourse production, circulation, and reception. He suggests an “anthropology of interviewing” that could illuminate how anthropologists’ modes of knowledge production intersect with practices that play crucial roles in the media, corporation, electoral politics, state bureaucracies, and a wide range of professions. He reminds us that increased visibility via the media has naturalized interviewing in ways that reshape how social science research is positioned. Communities and individuals within communities vary in the extent to which they are able to provide metalinguistic commentary (i.e., talk about) any formal features of discourse or criteria for judging performance (see Hymes 1981; Silverstein 2001 [1981]). In cultures that don’t engage in elaborate metalinguistic glossing, native speakers may not identify distinctive styles or genres or may not be able to describe characteristics of speech or criteria for evaluating performance. Observing how participants in an event react to performances can shed light on such features. For example, noting side commentary and what causes laughter, smiling, frowning, silence, changes in bodily postures, or responsive gestures can be very instructive. Such visible bodily languages also require careful translation from one cultural context to another, however, and potential meanings associated with such action signs must also be ascertained ethnographically, rather than assumed (Chapman 1982; Farnell 1999; Williams 2004 [1991], 149–73). There is no single best method of collecting information on language use and ideologies within a community. The starting point, though, is to seek, or discover, through participant observation, those forms of discourse that will form the data for analysis. (For a discussion of ethnographic methods and transcription, see Duranti 1997, 84–161.) Appropriate procedures will depend on the theoretical focus of the research, the relationship of the ethnographer to the community, and the specific situation. Negrón (2011), for example, in a tour de force of data collection, wired up 11 people. She shadowed them for six days and then let them loose for another week or so. This technique enabled her to collect data that previously would not have been possible; similar to ways that the invention of the tape recorder was transformative to the study of language in social life, this example underscores the extent to which technological advances, method, and theory go hand in hand. One method advocated as a way to make “ethnographies of communication” (e.g., Bauman and Sherzer 1989 [1974]; Gumperz and Hymes 1964; Hymes 1967, 1974; Saville-Troike 1989; Sherzer and Darnell 1972) involves making an inventory of distinct spoken forms, both marked (not typical) and unmarked (typical), and the contexts in which they occur. Recording a broad range of spoken discourse provides data from which to compare and contrast distinct expressive forms and styles along formal and functional lines. A researcher may want to learn if speakers change how they talk when addressing certain individuals (such as elders, relatives, or those of certain social

14_107-Bernard.indb 402

6/9/14 10:03 AM

12: Discourse-Centered Methods

403

roles or genders) or in distinct social settings (e.g., ceremonial contexts), or if there are special oratorical styles or styles of interacting when telling stories or myths. If special forms or styles of speech do exist, then the researcher would attempt to map their social distribution (i.e., find out who has the capacity and authority to use these forms) and try to discover how the forms are learned, in what situations they can occur, and to what extent speakers employ metadiscursive practices. Depending on the researcher’s theoretical focus, it might also be important to note whether some expressive modes of speaking stimulate other forms of expression (see Feld 2012 [1982]) and to determine the relationships between speech, gesture, facial expression, eye gaze, body position, and spatial orientation and other sensory modalities (see, e.g., Duranti 1992; Farnell 1995a, 1995c; Goodwin and Goodwin 2006; Hanks 1990; Haviland 1993, 1996, 2000; Kaeppler 1995; Kendon 1989; Wiget 1987). The notion of a soundscape (Schafer 1977; 1994) has directed researchers’ attempts to record and describe the full range of socially meaningful sounds used within a community (see Feld 2002, 2012 [1982]; Graham 1995). A soundscape describes the sounds that people make and use over, say, a 24-hour period, together with explanations of their meanings. It includes talk, musical forms, silences (Basso 1972), and attention to groups that are silenced (Gal 1991; Lederman 1984). Description of a soundscape might include variations in the social use of sound that occur due to seasonal changes, in response to (or to provoke) natural events, and in conjunction with ceremonial cycles or ways human sounds pattern in relation to and interact with sounds of the natural environment. The concept of soundscape draws attention to the fact that speech is just one modality within an entire arena of socially constructed and meaningful sound production. Distinct genres, such as lament, song, and oratory, may have important interconnections yet contrast in significant ways (see Graham 1984, 1986). Moreover, communities vary in the way that they divide the continuum between speech and song (see Feld and Fox 1994; List 1963). In a soundscape, expressive forms that sound like speech (or song) to Western ears may be classified in ways that challenge ethnographers’ assumptions. The concept of “acoustic culture” (the social organization of sound) has produced ethnographic studies of commercial sound studios and new media (see, e.g., Fox 2004; Meintjes 2003; Porcello 2005; Samuels 2004). The notion of soundscape has proven useful in calling researchers’ attention to unmarked forms of speech, which tended to be overlooked in earlier ethnographies of communication. Everyday or unmarked forms of discourse constitute important data for analysis because they contain information about language ideology, socialization, and ways in which gender, age, ethnic identity, and power relationships are linguistically constructed and conceived. These unmarked forms are also a base against which more salient forms (oratory, storytelling, poetry, etc.) can be compared and contrasted. Moreover, marked forms derive some of their meaning through their contrasts with everyday forms of expression (see, e.g., Urban 1985). The focus of much contemporary research is more specifically defined at the outset than earlier approaches. Discourse-centered research seeks to understand, for example, the relationships between language and gender (e.g., Bucholtz et al. 1999; Gaudio 2009; Hall and Bucholtz 1995; Inoue 2006; Mendoza-Denton 2008; Philips et al. 1987),

14_107-Bernard.indb 403

6/9/14 10:03 AM

404

Brenda Farnell and Laura R. Graham

whiteness and race (e.g., Bucholtz and Trechter 2008; Chun 2004; Hill 1998, 2008; Lo and Reyes 2009; Urciuoli 2009) identity formation and power (e.g., Alim 2006; Morgan 2007; Roth-Gordon 2009). It also seeks to understand the ways in which code switching, bilingualism (e.g., Kulick 1992; Lo 1999; Myers-Scotton 1993, 2006; Urciuoli 1996; Woolard 2006; Zentella 1997, 2005), and language socialization (e.g., Lo and Fung 2011; Ochs 1988; Schieffelin 1990; Schieffelin and Ochs 1986a, 1986b) are constitutive and expressive of selves, ideologies, and social relations. Depending on the research questions, researchers select those social contexts that facilitate observation of relevant linguistic features. For example, in her study of language socialization among Kaluli children, Schieffelin (1990) used ethnographic observation to determine the activities and times of day when caregivers and children interact most intensively. She found that opportunities for lively interaction occurred in the early morning when families awaken and begin to prepare the first meal. Interactions during the late afternoon when mothers return from the garden and begin preparations for the evening meal and during trips to the stream for bathing were other opportunities for lively interaction and were particularly fruitful occasions for studying socialization practices (Schieffelin 1990, 25). Like time, space may be another dimension that permits access to distinct types of social interaction. Thus, once Ochs (1988) broke through the “honored guest” status in the Samoan household, she was able to enter the domestic space at the back of the house where previously inaccessible forms of interaction between caregivers and children take place. Garvey (1984) and Ochs (1979a) show how the study of children’s talk is an area with its own methodological problems. The explosion of discourse that now circulates through the Internet and other social media has garnered analytic attention to new textualizing practices including blogs, text messaging, online news outlets, chat rooms, message boards, spam, Internet dating sites (Gershon 2010; Hill 2009; Shuman 2005:120-148), and scripted discourses in virtual worlds such as Second Life (Boellstorf 2008) as well as more traditional forms of media, such as radio, television, film (e.g., Fairclough 1995b; Meek 2006; Spitulnik 1993, 1996, 2010), and print newspapers (Fenginsen 1999; Graham 2011). The development of orthographies is also relevant here (Cameron 1995; Clark and Ivanič 1997; Irvine and Gal 2000; Jaffe et al. 2012; Rhydwen 1993; Salomon and Niño-Murcia 2011; Schieffelin and Doucet 1994). These studies bring productive insights to questions of accountability, authorship, and anonymity as well as power, authority, and political economies of knowledge production and circulation. READING, ARCHIVING, AND ETHICS

A crucial element of the research process is to make recordings of discourse as it unfolds in live social settings, via the mass media or through socially situated texts. For spoken discourse, typically, this involves using a high-quality digital audio or audiovisual recorder and microphones. (Internal microphones are generally of poor quality. For practical advice regarding recording techniques in social settings and other useful tools for linguistic anthropology research, see Society for Linguistic Anthropology [2012]; Thieberger’s Handbook of Linguistic Fieldwork [2012] also has a number of articles that deal with technical and practical how-to problems in recording discourse

14_107-Bernard.indb 404

6/9/14 10:03 AM

12: Discourse-Centered Methods

405

as well as its ethical entailments.) This technology, as Tedlock notes (1983, 38–39; see also Brady 1999; Gelatt 1977; Goodwin 1993; Ives 1995), has transformed fieldworkers’ ability to record oral texts. It enables them to attend to features of oral performance that could not be captured through elicitation or dictation. As digital video recording equipment has become more affordable and portable, researchers have increasingly incorporated this technology into their fieldwork so that visual–kinesthetic aspects of discursive practices can be incorporated into analyses (e.g., Farnell 1995a, 2011; Goodwin and Goodwin 2006; M. Goodwin 2006; Haviland 1993, 2000; Kendon 1989; Streeck et al. 2011). Since the 1990s, scholars increasingly complement written analyses with audio or audiovisual recordings using personal or journal websites (e.g., Bowen 1991a, 1991b; Farnell 1995; Graham In press; Sammons and Sherzer 2000; Journal of Linguistic Anthropology). Enhancing scholarship with audio and visual samples does not obviate issues of transcription; see section on transcription below. Carrying a recording device in the field as much as possible can provide a researcher with many samples and a variety of types of socially occurring discourse for later analysis. Recording devices may present problems, however, and a researcher must take into account ethical concerns about privacy, access to restricted information, and rights to ownership (Michaels 1994, 21–47; Gaudio [2009] offers an example of discourse-centered research where it was nearly impossible to make recordings of socially situated speech). In addition, researchers must consider the ways in which, like the presence of the ethnographer, the presence of recording devices may significantly alter the events themselves. In other communities, recording devices may be widely used by members themselves, and this can facilitate researchers’ use of documentary technologies (see Pink 2007 [2001]). In many communities, expressive forms such as songs and other forms performance are considered intellectual property (Seeger 1991, 1997, 2004a; see also Brown 2003; Coombe 1998, 2011; Goodman 2005; Verdery and Humphrey 2004). Intellectual property issues associated with field recordings are increasingly likely to confront researchers. This raises the issue of archiving data and the ethical entailments of archiving as well as who has access. As public granting agencies increasingly exert pressure on researchers to make their data accessible to broad publics, archives must establish appropriate policies relating to access. For example, the Archives of the Indigenous Languages of the Americas (AILA), at the University of Texas at Austin, has dealt with these issues by establishing various levels of access in accordance with community restrictions. The advent of affordable recording technologies has in some cases, opened up possibilities for collaboration (e.g., Graham et al. 2009; Michaels 1994; also Pink 2007 [2001]) and often enables research subjects to own and operate their own recording devices. Since outside researchers no longer have a monopoly over recording technologies, they are now challenged more than ever to think about the power vested in control of documentary technologies (see Ginsburg 2007). New technologies have also stimulated changes in the form or use of expressive genres (see Abu-Lughod 1993; Caton 1990; Manuel 1993; also Miller 2007; Turner 1991). Participants may be self-conscious or conscious of addressing audiences beyond those immediately present when they know they are being recorded, and their language use can reflect this. Consultants and collaborators may approach recording devices as

14_107-Bernard.indb 405

6/9/14 10:03 AM

406

Brenda Farnell and Laura R. Graham

political opportunities and as means to address wider audiences (see Graham 1995). People may forget that they are being recorded and, after several minutes of self-conscious interaction, they may settle into familiar communicative patterns and activities will proceed as usual (Briggs 1986; Darnell 1998 [1974]; Graham 1995; Haviland 1996; Hymes 1981; Tedlock 1983, 18, 302–11; Urban 1996a). Members of a community can become accustomed to the constant presence of digital technology in much the same way that people often become accustomed to seeing ethnographers writing things in notebooks. Individuals have also been known to comment on the tape recorder or camera’s absence when it’s not present and wonder why an event is not being recorded. It is not normally permissible to make recordings without the consent of participants in any communicative event, and an ethnographer may be excluded from an event if carrying such devices. Some would say that when rules are not clear, the choice to record or not is a question of professional ethics. Others would argue that it is never permissible. See Shuy (1986) and Murray and Murray (1992) for discussions of the legality of surreptitious recording. For discussion of ethical dimensions of fieldwork, see American Anthropological Association (2012), Anthropology News (1993, 4, 5, 7, 9; 2001, 4), Castañeda (2006), Fleuhr-Lobban (1998), Kovats-Bernat (2002), and Shannon (2007). In some situations, small, unobtrusive devices may be allowed and, if not, one option is to explore the possibility of recording an elicited or “staged” performance at another time. This will, of course, be different in many ways, but is not thereby rendered inauthentic nor is it without considerable analytic value. The presence of the ethnographer and recording devices will almost certainly affect people and their interactions, but reflexive consideration of these factors, rather than being denied, can be incorporated into analyses with interesting results (see Darnell 1998 [1974]). Labov (1972) recognized the problem of how to observe people’s interaction without their being self-conscious as an “observer’s paradox” in sociolinguistic research and developed various fieldwork techniques to attempt to transcend this dilemma. Certainly, good field recordings of contextually situated interaction either audio or audiovisual are essential to discourse-focused work (but see Gaudio 2009). This doesn’t negate the usefulness of field notes, however, which are often essential companions to recorded data. For example, field notes can be used to document important contextual observations that can later be incorporated into transcriptions. Alternately, comments can be spoken into the tape recorder/video camera itself. A useful technique is to correlate numbers on an audio/video recorder counter with notes written in field notes (e.g., Schieffelin 1990, 30). Sometimes ethnographers speak directly into a second voice recorder, but then they risk disturbing the data they are collecting; speaking softly during pauses or intervals during an event may be productive. In and of themselves, field notes are insufficient as records because they record metadiscursive observations or descriptions; they don’t provide records adequate to analyze the richness and nuances of discourse as performed. Tedlock (1983, 38–39), for example, observed that the Zuni texts he recorded using a tape recorder were nearly twice as long as those dictated to Benedict. Among the reasons for this is that narratives given for dictation tended to be condensations of what a performer would tell in

14_107-Bernard.indb 406

6/9/14 10:03 AM

12: Discourse-Centered Methods

407

a normal, spontaneous situation and because a responsive audience was lacking. At the time field recordings are made, a researcher may be unaware of components that subsequent analysis of audio/video reveal as significant carriers of social meaning (see discussion of the two audio recorder/two video method below). PROCESSES OF TRANSLATION AND TRANSCRIPTION

Although the presence of web-enhanced journals, websites, and digital web-based archives provide increased access to primary data, digital audio and visual recordings have neither replaced nor reduced the need for adequate methods of written transcription. On the contrary, electronic technologies have facilitated the development of written transcription systems more adequate to the analytic tasks at hand. The processes of transcription and translation are themselves interpretive analytic tasks that are invested with power (see Bucholtz 2000); they are re-presentations of performance that are necessarily informed by explicit and implicit theoretical assumptions (see Hymes 1981; Ochs 1979b; Sherzer 1992; Sherzer and Urban 1986a, 11–12; Swann 1992; Tedlock 1983; Woodbury 1985). As Ardener (1989 [1978]) reminds us, particular events that are registered depend on our modes of registration and specification—that is, the means by which they are apperceived. He advises that we should know as much as possible about these modes because “our definition of . . . the events depends upon the modes of registration available to us” (p. 87). The techniques we use to complete the tasks of transcription and translation influence the interpretation of texts and the ways in which meanings are understood to emerge from socially situated discourse. Similarly, the phonological patterns or grammatical categories of the translator‘s language may affect the transcription and translation of texts (see Sherzer 1987; Whorf 1956). MacMahon (1996, 831–37) provides an excellent history of the interest in and development of phonetic transcription systems that can be traced back, in England and North America, to the late sixteenth century. Interest in Asian languages and cultures at the end of the eighteenth century led to the first questions about the ability of characters from the Roman alphabet to adequately represent spoken pronunciation in foreign tongues. In other contexts, the need to incorporate information about pronunciation into dictionaries, interest in transcribing local dialects of European languages, and intensive Christian missionary activity in Africa and elsewhere also led to specific methods of transcribing speech (MacMahon 1996).  A group of modern language teachers in the International Phonetic Association incorporated many of these efforts into the International Phonetic Alphabet (IPA) in 1888. Although it has undergone subsequent changes and additions, the IPA remains the major phonetic alphabet in use today. It comprises an extensive set of symbols and diacritics derived from the Roman alphabet and has symbols for consonants, vowels, and some prosodic features such as stress and intonation. It’s worth remembering that in the era of dictation (an era stretching back to the making of Homeric texts and lasting until well into the Boasian era), the only language-recording technology was writing (see Goddard 1996; Mithun 1996). Before the invention of the tape recorder, important collections of spoken language and other vocal expressive forms were recorded on wax cylinders, glass acetate, and

14_107-Bernard.indb 407

6/9/14 10:03 AM

408

Brenda Farnell and Laura R. Graham

aluminum phonograph discs. Materials recorded with these technologies are deposited in the Archives of Traditional Music at Indiana University and in the Smithsonian Institution in Washington, DC. In addition to collections made by Boas and his students such as Radin and Fletcher, Haugen’s study of Norwegian language change in the Midwest (1969) and Turner’s collection of Sea Island Gullah (1969) material included samples of situated discourse. The advent of the tape recorder transformed the fieldworker’s ability to record oral discourse (Tedlock 1983, 38–39), and linguistic anthropologists and folklorists began to turn their theoretical attention to features of oral performance that could not be captured through elicitation or dictation. The examination of native poetic structure (ethnopoetics) and the adequate translation of verbal art emerged as central concerns. At this point, analytical limitations of conventional transcriptions began to emerge. Ironically perhaps, the very success of phonetic notation systems had led investigators to limit their definitions of language to those segmentable components that could be represented with the Roman alphabet and its extensions. New technologies and innovative forms of transcription have exploded limiting theoretical assumptions about language and enabled anthropologically oriented linguists to expand notions of linguistic functionality. With the explicit goal of exploring native poetic structures, Hymes (1981), Tedlock (1983), and others began to expand the notational repertoire to include conventional poetic and dramatic devices. Figure 12.1 illustrates how line breaks were introduced to indicate pauses, inviting the reader to listen more closely to the rhythms of speech and silence in performance. Upper-case letters mark louder speech, and marginal parentheses guide the reader’s attention to voice quality and volume. A vowel followed by long dashes indicates vowel lengthening; letters spilling downward on the page indicate descending glissando; split lines indicate tonal changes in chants. Tedlock called for punctuating according to rising and falling contours of oratory and shaped lines and stanzas according to the stops and starts of dramatic timing. His explicit aim was to create “open or performable texts” that “open the ear” to the sounds of vocal performance (1983, 62). Students of “the ethnography of speaking” contributed to these developments, choosing from these transcription options according to their analytic purposes. Papers in Edwards and Lampert (1993) discuss a number of transcription and coding systems for discourse analysis from contrasting theoretical perspectives, highlighting the point that a transcription format should be related to the goals of analysis. No transcription can ever be so detailed and precise as to provide for the full recreation of the sound (or vision) of a recording: The hypothetical goal of total notation is a positivist’s dream. Individuals and communities can also exert influence over transcription conventions. For example, Becker (1983) describes how his Burmese consultant objected to phonetic transcription and insisted on using the Burmese writing system because it related to Buddhist understandings of the language. Brandt (1981) describes an extreme case where members of a speech community entirely rejected the writing of their languages. Transcription and translation themselves must be viewed as contextually situated emergent dialogic acts, imbued with power that may generate novel meanings. Accordingly, these tasks have also become the focus of productive and provocative analytic

14_107-Bernard.indb 408

6/9/14 10:03 AM

12: Discourse-Centered Methods

409

Figure 12.1. Transcription designed to convey features of oral performance such as volume, pause structure, utterance length, and pitch (from The Spoken Word and the Work of Interpretation by Dennis Tedlock. Copyright © 1983 University of Pennsylvania Press. Reprinted by permission of the publisher.).

inquiry (see Haviland 1996; Urban 1996a; see also papers in Sammons and Sherzer 2000; Tedlock 1983). Recent studies of roles of native informants in language-focused work challenge researchers to reflect on the situated nature of producing transcriptions, translations, and analyses (see Bernard and Salinas 1989). Anthropologists have long recognized the significance individual informants and their social positions play in research endeavors. In discourse-focused work, native consultants play important and unique roles. Not only can their social position influence whether they will transcribe or translate the speech of certain individuals (see Graham 1995, 13–16), but relationships between speakers and consultants may influence the meanings that are understood to be encoded in texts and even how speech itself is transcribed. Graham (2011) demonstrates the power relations involved when print news editors transform quoted speech into print and ways transcriptions influence interpretations and perceptions of speakers. Haviland (1996) examines how Tzotzil Mayas from the highlands of Chiapas, who have little experience with reading and writing, transform aspects of recorded

14_107-Bernard.indb 409

6/9/14 10:03 AM

410

Brenda Farnell and Laura R. Graham

conversation in producing native written texts of multiparty talk. He notes that natives change written texts in ways that normalize significant pragmatic features of conversational situations. This includes erasing context-specific elements of speech such as quotative particles and evidentials that relate utterances to previous conversation. Natives also smooth “interactional edges,” eliminate false starts, disfluencies, asides, and Spanish loan words from their transcriptions. Urban (1996a) observes that there may be predictable regularities in the ways that native transcriptions of recorded speech differ from original utterances. He proposes that when replication occurs in relatively deliberate contexts (as in transcription), the copy may include segmentable forms that don’t occur in the original but that encode meanings that are only pragmatically inferable from the original. He also notes that native transcribers may delete metadiscursive instructions, especially indications of mistakes or deviations from an intended text (as in a myth), and may correct mistakes indicated by such remarks. Urban proposes that power relations between the speaker of the original text and the copier (or transcriber) may influence the production of a written text. He says that when the relationship between the originator and copier is more symmetrical and egalitarian, there will be a greater divergence between the copy and the original; the copier is also more likely to respond to the originator. In contrast, where there is greater power differential, transcribers attempt to render versions that are more faithful to the original text. These observations have important consequences because the ways in which a text is transcribed influences its subsequent translation and analysis. With a corpus of transcribed and translated texts in hand, researchers may examine formal discourse features and grammatical patterning to discover ways in which such devices are implicated in broader social and cultural phenomena. Scholars have examined such features as evidentials, deictics, reported speech, metadiscursive particles and phrases, semantic and grammatical patterning as well as formal features of performance such as the organization of talk. In addition to such microconstituents, formal analysis may focus on macro structures such as the formal organization of entire narrative texts (see Hymes 1981; Urban 1986; also Wutich et al., this volume) or on relationships between entire genres (see Bauman 2004; Bauman and Briggs 1990; Briggs and Bauman 1992; Graham 1986; Seeger 1986). Work with native assistants in transcription and translation, including metacommentary and elicitation, helps tease out further significant features. Through detailed attention to such formal elements, as well as examination of the ways in which creative indexical reference conveys additional social meaning in communicative acts, researchers look for ways in which social phenomena, such as relations of power, kinship, identity, gender relations, aesthetics, ideologies, and affect are discursively constituted and negotiated. COLLABORATIVE TRANSCRIPTION AND ANALYSIS

To better understand native perspectives in instances of socially situated speech, discourse-focused research has attempted to directly involve native speakers in tran-

14_107-Bernard.indb 410

6/9/14 10:03 AM

12: Discourse-Centered Methods

411

scription and analysis. In transcribing and analyzing the speech of Kaluli children, for example, Schieffelin worked with the mothers of children she had recorded (1990, 30–32). The mothers helped with initial transcription by repeating back their own and others’ speech as they listened to the tape-recorded samples. Schieffelin notes that this was especially helpful in glossing special forms children used, identifying speakers in multiparty talk, and specifying the history and ownership of objects involved in the interactions. The women also provided important metalinguistic commentary that made explicit Kaluli notions about proper language use. Schiefflin took careful notes on the women’s comments and incorporated them into the extensive contextual notes she had made while recording. This metacommentary became part of extensive text annotations, and Schiefflin used it to generate further questions about interactional patterns, classifications of speech acts and events, and notions of what constituted “good talk” (1990, 30). She notes that for every hour of recorded speech, she spent an additional 20 hours preparing the annotated and translated transcript. Given the recognition that discourse is both saturated with and constitutive of social meaning, this is neither surprising nor atypical for discourse-centered work. To obtain further insights into the recorded interactions, Schieffelin incorporated a reliability check into her work (1990, 30–31). This involved relistening and reviewing the transcriptions with a young man who hadn’t participated in the interactions. Schieffelin originally conceived of this exercise as a means to check her initial transcriptions, but she discovered that it was a way of “extending and enriching the contextualizing ethnographic and linguistic information” (p. 31) that the mothers had provided. Work with the young man revealed that devices such as prosody, voice quality, and affect-marked affixes convey significant affective information and that pragmatic uses of word order can disambiguate utterances and clarify issues. To avoid influencing consultants’ interpretations of video-recorded data, Erickson and Shultz (1982, 59–63) developed a technique that allowed consultants to stop the tape wherever they found items worthy of attention and to comment on them in interactions they were watching. They found that consultants usually chose to stop at points where discourse topics or conversational routines had just been completed. Graham (1995, 15–16) worked with Xavante assistants in ways similar to Schieffelin but used two tape recorders as she was transcribing and translating field recordings with native speakers. While one recorder was used to play back recorded instances of contextually situated discourse, the other recorded the consultants’ explanations and responses to questions Graham posed about the interactions and language use. Using two recorders enables the investigator to return repeatedly to informants’ responses to the original taped utterances. It also permits the researcher to discover discrepancies between forms that appear in an initial recording and what informants repeat for transcription (see Urban 1996a). Such discrepancies may be due to a straightforward difference between citation form (pronunciations given in elicitation or interview contexts) and those used in actual practice, but not necessarily. Factors such as a difference in age, class, or gender may be at work, and comparisons may reveal social and pragmatic meanings associated with such alternative

14_107-Bernard.indb 411

6/9/14 10:03 AM

412

Brenda Farnell and Laura R. Graham

forms. It was in using this technique, for example, that Graham discerned significant pragmatic phenomena such as features associated with the speech of elders and alternative patterns of person marking and verb morphology used in speech to, or about, certain kin and affines. These forms, Graham (1990) ascertained, are not typically given in elicitation situations. Likewise, in his work on Warao ritual wailing, Briggs (1993) found that women’s comments while collaboratively transcribing their own laments was crucial for grasping fine dimensions of collective composition and performance, in particular the relationship between minute acoustic features and affect. Affect is another area of significant crossover between contemporary linguistic anthropology and sociocultural anthropology. In a similar manner, Farnell made important discoveries about Nakoda (Assiniboine) conceptions of the integrated nature of speech and signing and the primacy of movement in indigenous classifications by videotaping her own interaction with informants during collaborative transcription and translation sessions as they watched and listened to their own previously recorded storytelling performances (Farnell 1995a, 247). Farnell had to change her methodological decision to transcribe first the speech and then the action because one of her informants consistently failed to repeat the words he had spoken on the videotape. It became apparent that he was adding material to ensure that meanings contained in the gestural component of his utterances were also included. Only when both speech and action were discussed and transcribed simultaneously was he satisfied that what he had said was all there. COMPUTER-ASSISTED ANALYSES

Discourse-centered scholars employ audio recording and editing software, some of which can be downloaded free from the Internet (e.g., Audacity). Such programs provide graphic representations of waveforms that can be useful for analysis and representation of patterning in amplitude, pause structure, and pitch (through wave-form and spectrograph display). Digitized images can be especially productive in analysis because they enable researchers to capture microfeatures of performance and correlate these with features of the social contexts. They can also perform analytic tasks involving looping, splicing, and juxtaposition of instances of speech or other sounds. For example, it is possible to create a loop of a recorded segment and replay it repeatedly to listen—and see—aspects of sound such as utterance or pause length and amplitude. Sound segments can also be juxtaposed to discern the existence of regularities or microvariations across instances. Digitized sound samples enabled Roseman (1991, 171) to recognize an acoustic iconicity between cicada buzzing, certain birdcalls, and Temiar (Malaysia) bamboo tube drumming, which were essential to efficacious performance. The digitized representations shown in Figure 12.2 illustrate the striking resemblance between the waveforms of sounds Temiar links with performance and associate with evocative power. Likewise, Urban uses digitized samples to represent the regularity of utterance length and amplitude in an excerpt from the Shokleng origin myth to illustrate the syllableby-syllable dialogic nature of the performance (1991, 109; see also 1996b, 201).

14_107-Bernard.indb 412

6/9/14 10:03 AM

12: Discourse-Centered Methods

413

Figure 12.2. Digitized samples of (1) cicadas, (2) the golden-throated barbet (Megalaima fanklini), and (3) the beat of bamboo tubes illustrate an iconicity that is essential to efficacious Temiar performance (from Roseman 1991:170. Reproduced by permission of the University of California Press from Healing Sounds from the Malaysian Rainforest: Temiar Music and Medicine. Copyright © 1991).

Digitized figures also illustrate the regularities of Warao lament performance in Briggs’s analysis (1993). Sherzer uses digitized samples to display visually the parallelism of amplitude (from high to low) that characterizes certain genres of Kuna oral performance as well as pitch and volume patterns (1992, 430, 435–38). In reviewing recent scholarship for the second edition of this chapter, we note that recent publications have made little use of these kinds of representations (but see Reber 2012). Modified Western musical staffs can also be suitable for depicting features of vocal performance. Linguistic anthropologists, often working in collaboration with ethnomusicologists (e.g., Graham 1984, 1995; Sherzer and Wicks 1982) or those with ethnomusicological training (e.g., Feld 2012 [1982]; Seeger 2004b [1987]), have developed innovative ways to represent melodic genres. For example, Figure 12.3 shows a transcription of a Xavante musical performance in which the musical staff has been reduced from five to three lines to highlight the reduced pitch inventory of this musical genre. Notes with indistinct pitch appear as x, and slanting lines before notes indicate a slide up or down to the indicated pitch (transcription by Scruggs in Graham 1995, 120). By contrasting four representational modes of the same stretch of Kuna chanting, Sherzer (1992, 434–35) shows how a variety of transcription modes are suitable for distinct analytic ends (see Figure 12.4). These alternative modes of representation and transcription can be used to highlight or downplay the importance of varied linguistic and stylistic features that serve various expressive and aesthetic ends (see also Hymes 1981; McClendon 1981; Silverstein 1984; Tedlock 1983, 1987; Woodbury 1985, 1987).

14_107-Bernard.indb 413

6/9/14 10:03 AM

Figure 12.3. The musical transcription of Xavante da-ño’re illustrates how modifications of the conventional Western musical notation system may highlight salient features of non-Western musics. Here, the Western musical staff is reduced to call attention to the pitch inventory and notation is designed to indicate sliding pitch and less definite pitch (transcription by T. M. Scruggs in Graham 1994:734–735, 1995:120–121. Reproduced by permission of the American Anthropological Association from American Ethnologist, 1994. Not for sale or further reproduction).

14_107-Bernard.indb 414

6/9/14 10:03 AM

12: Discourse-Centered Methods

415

THE EMBODIMENT OF DISCOURSE

The shift to performance and the inclusion of context in linguistic analyses also opened up theoretical space within which to add the embodiment of social actors to the notion of language-in-use (Farnell 1995a, 9). As mentioned above, several investigators have recognized that the visual-kinesthetic components of discourse, such as gestures, postures, facial expressions, eye-gaze, and spatial orientation, are meaningful components of linguistic utterances in social contexts (e.g., Duranti 1992; Goffman 1961, 1974; Goodwin 1981; Hanks 1990; Haviland 1993, 2006; Keating 1998; Kendon 1980, 1983; Sherzer 1973; Studdert-Kennedy and Whitney Goodall 1992; Wiget 1987). Sign languages and discourse within d/Deaf communities have also garnered some ethnographic attention (see LeMaster and Monaghan 2002; Monaghan et al. 2003; Senghas and Monaghan 2002; Washabaugh 1981, 1986). Until recently, a way of making the analysis of body movements in spatial contexts part of the normal ethnographer’s toolkit had not been developed. In the 1970s, Birdwhistell’s (1970) kinesics promised a solution, but failed because it was without an adequate post-Cartesian conception of embodied person. Bodily movements were still seen as behaviors rather than signifying acts—that is, as agentic, semiotic practices that

14_107-Bernard.indb 415

6/9/14 10:03 AM

Figure 12.4. Four representational modes of the same stretch of Kuna chanting illustrate a variety of transcription modes uniquely suitable for distinct analytic ends. Part 1 provides a segmental transcription of sounds, morphemes and words in which line and verse structure is based on musical and pause patterns; Part 2 is a graph that captures pitch and tempo; Part 3 is a SoundEdit display that illustrates amplitude and pause patterns; and Part 4 is pitch tracking (of format structures) that shows the falling pitch and value that is characteristic of this genre, as well as long pauses between lines. (Parts 1, 3, and 4 from Sherzer 1992: 434–436 reproduced by permission of the Smithsonian Institution Press; Part 2 reproduced by permission of Joel Sherzer and Sammie Wicks and Latin American Music Review, of the University of Texas Press. Copyright © 1982).

14_107-Bernard.indb 416

6/9/14 10:03 AM

12: Discourse-Centered Methods

417

are shared expressive resources that require translation from one culture to another. Kinesic analyses and attempts to create an adequate transcription system dissolved under the weight of ever-increasing microscopic behavioral detail. Also absent from earlier conceptions was an adequate analytic framework for dealing with the multidimensional spaces within which humans move; spaces are simultaneously physical, conceptual, moral, and ethical (Williams 1995, 52). Williams’s semasiological theory of human action, with its central concept of the “action sign,” has provided investigators with new theoretical resources more adequate to this task (see Williams 1975, 1982, 1995, 1996, 2004 [1991]). From a semasiological perspective, discourse is necessarily embodied because persons are conceived as subjects empowered to perform signifying acts with both vocal signs (speech) and action signs (bodily movements). In addition, a new emphasis on acquiring a literacy specific to the medium (i.e., the ability to read and write body movement) has been achieved

14_107-Bernard.indb 417

6/9/14 10:03 AM

418

Brenda Farnell and Laura R. Graham

through a movement transcription system called Labanotation or the Laban script (Farnell 1994; Page 1990, 1996; Williams and Farnell 1990). Farnell (1999) offers an overview of theoretical developments in the anthropology of human movement systems (see also Journal for the Anthropological Study of Human Movement http://jashm .press.illinois.edu/). An alphabetic script for writing body movement must solve the problem of how to represent all the parts and surfaces of the body with two-dimensional graphic signs (see Figure 12.5). It must also make finite in some way the space(s) in which the body moves. Figure 12.6 illustrates how the Laban script uses simple set theory to represent spatial directions viewed from the actor’s perspective. Locating this imaginary

Figure 12.5. Labanotation: graphic signs for body parts and surfaces (from Farnell 1995a. Reproduced by permission of the University of Texas Press).

14_107-Bernard.indb 418

6/9/14 10:03 AM

12: Discourse-Centered Methods

419

Figure 12.6. Labanotation: graphic signs for specifying spatial direction (from Farnell 1995a. Reproduced by permission of the University of Texas Press).

three-dimensional “cross of axes” at each joint of the body achieves a remarkable economy of graphic signs: The same scheme specifies the direction of individual limbs and smaller body parts and provides a framework for indicating the direction of pathways for the whole body (as when a person moves from one location to another). These graphic signs for body parts and spatial directions are arbitrary but iconically motivated to assist reading fluency. Movement necessarily takes place through time and the flow of action and direction of reading in texts written with the Laban script is upward. Placing symbols in different columns on a vertical staff organizes the flow of action; a central dividing line separates left and right sides of the body. The Laban script can be divided into bars like a musical score if metrical timing is required, or it can be used without if

14_107-Bernard.indb 419

6/9/14 10:03 AM

420

Brenda Farnell and Laura R. Graham

not—the length of spatial symbols indicating relative timing instead. Farnell’s ethnography of the simultaneous production of spoken Nakoda and Plains Sign Language in the storytelling practices of Nakoda elders illustrates this approach (1995a, 1995c). Figure 12.7 is an example of a transcription of speech and gestural signs from a storytelling performance that was first videotaped and then transcribed and translated in collaboration with the storyteller.

Figure 12.7. Assiniboine storytelling with Plains Sign Talk and Nakota: p. 1 of the Labanotated score (from Farnell 1995a:83. Reproduced by the University of Texas Press).

14_107-Bernard.indb 420

6/9/14 10:03 AM

12: Discourse-Centered Methods

421

Concepts of the body, systems of spatial orientation, and concepts of time differ considerably across languages and cultures, and the Laban script has proven flexible enough to take these into account (see Farnell 1994, 1996; Page 1996; Williams 1995; Williams and Farnell 1990). For example, Farnell (1995a, 141–54) found that understanding Nakoda conceptions of the cardinal directions was, among other things, central to understanding deictic utterances in speech and gesture (see Haviland 1993). This applied to unmarked forms of everyday discourse as well as the performance of spoken/signed narratives. Such a conception was incorporated into the transcription by placing a spatial orientation key at the start of the movement score; this tells the reader that this particular conception is in operation throughout, much like the key of C# minor might operate at the start of a musical score. We know as much as we do about the speech components of discursive practices because we remove them from the flow of real time by writing them down. Alphabetic scripts are a clear mode of specification for components of spoken languages at a phonological level. A script designed to produce movement texts establishes similar conditions for the specification and registration of the bodily, spatial, and dynamic components of actions. This makes translation into words unnecessary for creating ethnographically appropriate descriptions of actions. In addition, when actions and words integrate, as is most often the case in social interaction, both can be given equal analytical weight. Although linguists have attempted to develop phonetic transcription systems for sign languages (e.g., Kennedy et al. 1997; Peterson 2011; Stokoe et al. 1965), many anthropological researchers of gesture and signed languages have yet to grasp the analytic advantages of movement literacy (see Duranti 1997, 144–54; Senghas and Monaghan 2002). Inadequate modes of recording and representing embodied action for the purposes of linguistic analysis include word glosses, still photographs, drawings, and video (Farnell 1994, 2012). Analysts of sign language discourse have been assisted by software such as SignStream and MediaTagger, which make word gloss transcriptions/translations linked to video data easier to manipulate (Senghas and Monaghan 2002) but do not resolve the movement transcription issue. In addition, embodied aesthetic sensibilities such as voice quality and melody can underlie notions of propriety, correctness, appropriateness, and honorification in language use. Laderman (1987), for example, illustrates how only “beautiful voices” are efficacious in Malay ritual. Irvine (1992) demonstrates how Wolof speakers ideologically associate vocally embodied affectivity with honorification. INTERSUBJECTIVITY AND COLLABORATION WITH NATIVE SPEAKERS

Ethnographic knowledge production is necessarily intersubjective and collaboratively produced (Pink 2007 [2001], 30–31). Since meanings of utterances are embedded within the context in which they occur, it is important to note not only what explicit propositions are made, but also other features of the environment that might have some relationship to an utterance’s meanings. Many studies show how meaning emerges from context; others show how the internal structure of narratives reflect embeddness in larger interactive processes (C. Goodwin and Duranti 1992, 11; M. Goodwin 1982a, 1982b; Heath 1983).

14_107-Bernard.indb 421

6/9/14 10:03 AM

422

Brenda Farnell and Laura R. Graham

Since consultants are always active participants, working in collaboration with them during transcription, translation, and interpretation is crucial to discovering presupposed knowledge, interpretive frames, and the relevance of prior discourse. Basso (1979, 1990), Briggs (1986, 1988), and Hymes (1981), among others, illustrate the advantages of long-term collaborations with consultants. Farnell (1995a), Graham (1995), and Valentine (1995) explicitly address ways in which consultant and community concerns can influence the shape and scope of research. The methods used will depend on research goals as well as practical and political realities of the fieldwork situation. Collaboration itself is a socially constructed relationship that develops throughout fieldwork. Insofar as the goals of communities shape research agendas, linguistic anthropologists practice engaged anthropology (see Black 2013; Zentella et al. 2007; see also Low and Merry 2010). CONCLUDING REMARKS

Discourse-centered approaches in linguistic anthropology position speech and related expressive acts, including textualizing practices, as the principal means by which social actors create and recreate social life. They thus open up new ways to conceive of culture. Discourse-centered approaches at the heart of linguistic anthropology increasingly influence sociocultural anthropology, particularly as the theoretical focus has shifted away from structure and toward practice. With socially situated discourse as the starting point, discourse-centered approaches provide specific analytic tools and mechanisms for understanding dynamic social processes and the ways in which power and authority are constructed and wielded in the Foucauldian sense of discourse (1972). Some scholars make a useful distinction between Discourse (“big D” discourse) and discourse (“little d” discourse) to capture this difference (Conley and O’Barr 2005; Gee 2011a). Ways in which wider social phenomena will be encoded in discourse can rarely be anticipated before ethnographic fieldwork is well underway. They are usually discovered during fieldwork and necessarily vary according to such factors as interactions between the ethnographer and the individuals and community with whom she or he works as well as formal elements of the discourse itself. Moreover, just as meanings and interpretations are intersubjectively co-produced among interactants in any ethnographic situation, anthropological understandings of discourse forms are achieved with consultants/collaborators as well as in dialogue with academic colleagues and the intellectual capital of linguistic anthropology. As we have shown, discourse-centered research entails ongoing relationships between technologies, method, and theory. In an ever-evolving dynamic process, new theoretical questions and insights challenge researchers to devise new methods and adopt or devise new technologies that, in turn, stimulate more theoretical questions and so on. Since theory and method are closely intertwined and are part of the dialogical processes through which we, as social scientists, construct ethnographic knowledge with our collaborators within fields of power, we can expect that new methods of research will continue to arise.

14_107-Bernard.indb 422

6/9/14 10:03 AM

12: Discourse-Centered Methods

423

ACKNOWLEDGMENTS

We thank Russ Bernard and Jane Hill for their helpful suggestions and the opportunity for collaboration. We also thank Charles Briggs, Clarence Gravlee, Joel Sherzer, Bambi Schieffelin, Greg Urban, Bonnie Urciuoli, and anonymous reviewers for their comments on earlier versions. Each author thanks the other for patience, good humor, and intellectual companionship. Our contributions are shared; our names appear in alphabetical order. REFERENCES

Abu-Lughod, L. 1993. Writing women’s worlds: Bedouin stories. Berkeley: University of California Press. Agha, A. 1996. Honorifics. Annual Review of Anthropology 22: 473–88. Agha, A. 2007. Language and social relations. Cambridge: Cambridge University Press. Ahearn, L. 2001. Invitations to love: Literacy, love letters and social change in Nepal. Ann Arbor: University of Michigan Press. Alim, H. S. 2006. Rock the mic right: The language of hip hop culture. New York: Routledge. American Anthropological Association. 2012. http://www.aaanet.org/committees/ethics/ethcode .htm (accessed February 6, 2013). Anthropology News. 1993. The ethics of fieldwork. Volume 34. Washington, DC: American Anthropological Association. Anthropology News. 2001. Ethics and fieldwork across anthropology’s subfields. 42: 26. Washington, DC: American Anthropological Association. Ardener, E. 1989 [1978]. Some outstanding problems in the analysis of events. In Edwin Ardener: The voice of prophecy and other essays, ed. M. Chapman, 86–104. Oxford: Basil Blackwell. Austin, J. L. 1962. How to do things with words, 2nd ed., J. O. Urmson and M. Sbisa, eds. Cambridge, MA: Harvard University Press. Bakhtin, M. M. 1981. The dialogic imagination. C. Emerson and M. Holquist, trans. Austin: University of Texas Press. Bakhtin, M. M. 1986. Speech genres and other late essays. V. W. McGee, trans. Austin: University of Texas Press. Basso, E. 1995. The last cannibals: A South American oral history. Austin: University of Texas Press. Basso, K. 1972. “To give up on words”: Silence in Western Apache culture. In Language and social context, ed. P. P. Giglioli, 67–86. Harmondsworth, UK: Penguin Books. Basso, K. 1979. Portraits of the whiteman: Linguistic play and cultural symbols among the Western Apache. New York: Cambridge University Press. Basso, K. 1990. Western Apache language and culture. Tucson: University of Arizona Press. Bauman, R. 1977. Verbal art as performance. Rowley, MA: Newbury House. (Reprinted in 1984 by Waveland Press.) Bauman, R. 1983. Let your words be few: Symbolism of speaking and silence among 17th century Quakers. Cambridge: Cambridge University Press. Bauman, R. 1992. Identity and interaction: A sociocultural linguistic approach. In Rethinking context: Language as an interactive phenomenon, ed. C. Goodwin and A. Duranti, 125–45. Cambridge: Cambridge University Press. Bauman, R. 2004. A world of others’ words: Cross-cultural perspectives on intertextuality. Malden, MA: Blackwell. Bauman, R., and C. Briggs. 1990. Poetics and performance as critical perspectives on language and social life. Annual Review of Anthropology 19: 59–88.

14_107-Bernard.indb 423

6/9/14 10:03 AM

424

Brenda Farnell and Laura R. Graham

Bauman, R., and C. Briggs. 2003. Voices of modernity: Language ideologies and the politics of inequality. Cambridge: Cambridge University Press. Bauman, R., and J. Sherzer, eds. 1989 [1974]. Explorations in the ethnography of speaking, 2nd ed. Cambridge: Cambridge University Press. Becker, A. L. 1983. Biography of a sentence: A Burmese proverb. In Text, play, and story: The construction and reconstruction of self and society, ed. E. M. Bruner, 185–210. Proceedings of the American Ethnological Society, American Anthropological Association. (Reprinted in 1995 in Beyond translation: Essays toward a modern philology, Ann Arbor: University of Michigan Press.) Benveniste, Emile. 1971a [1956]. Subjectivity in language. In Problems in general linguistics, trans. M. E. Meek, 217–22. Miami: University of Miami Press. (Originally published in 1956 in For Roman Jakobson, The Hague: Mouton.) Benveniste, Emile. 1971b [1956]. The nature of pronouns. In Problems in general linguistics. M. E. Meek, trans. Miami: University of Miami Press. (Originally published in 1956 in For Roman Jakobson, The Hague: Mouton.) Bernard, H. R., and J. Salinas Pedraza. 1989. Native ethnography: A Mexican Indian describes his culture. Newbury Park, CA: Sage. Birdwhistell, R. 1970. Kinesics and context: Essays in body motion communication. Philadelphia: University of Pennsylvania Press. Black, S. P. 2013. Linguistic anthropology in 2012: Language matter(s). American Anthropologist 115: 269–82. Boas, F. 1940. Race, language and culture. New York: Macmillan. Boellstorff, T. 2008. Coming of age in Second Life: An anthropologist explores the virtually human. Princeton: Princeton University Press Bourdieu, P. 1977a. Outline of a theory of practice. R. Nice, trans. Cambridge: Cambridge University Press. Bourdieu, P. 1977b. The economics of linguistic exchanges. Social Science Information 16: 645–68. Bourdieu, P. 1991. Language and symbolic power. Cambridge: Harvard University Press. Bowen, J. 1991a. Gayo speech and song. Audiotape. New Haven, CT: Yale University Press. Bowen, J. 1991b. Sumatran politics and poetics: Gayo history, 1900–1989. New Haven, CT: Yale University Press. Brady, E. 1999. A spiral way: How the phonograph changed ethnography. Jackson: University of Mississippi Press. Brandt, E. 1981. Native American attitudes toward literacy and recording in the Southwest. Journal of the Linguistic Association of the Southwest 4:185–95. Brenneis, D. 1984. Straight talk and sweet talk: Political discourse in an occasionally egalitarian community. In Dangerous words: Language and politics in the Pacific, ed. D. L. Brenneis and F. R. Myers, 69–84. New York: New York University Press. Brenneis, D. 1986. Shared territory: Audience, indirection, and meaning. Text 6: 339–47. Briggs, C. 1986. Learning how to ask? A sociolinguistic appraisal of the role of the interview in social science research. Cambridge: Cambridge University Press. Briggs, C. 1988. Competence and performance: The creativity of tradition in Mexicano verbal art. Philadelphia: University of Pennsylvania Press. Briggs, C. 1992. “Since I am a woman, I will chastise my relatives”: Gender, reported speech and the (re)production of social relations in Warao ritual wailing. American Ethnologist 19: 336–61. Briggs, C. 1993. Personal sentiments and polyphonic voices in Warao women’s ritual wailing: Music and poetics in a critical and collective discourse. American Anthropologist 95: 929–57. Briggs, C., ed. 1996. Disorderly discourse: Narrative, conflict and inequality. New York: Oxford University Press.

14_107-Bernard.indb 424

6/9/14 10:03 AM

12: Discourse-Centered Methods

425

Briggs, C. 2001. Anthropology of discourse. In International encyclopedia of social and behavioral sciences, Vol. 6, ed. N. J. Smelser and P. B. Baltes, 3732–36. Amsterdam: Elsevier. Briggs, C. 2005. Communicability, racial discourse, and disease. Annual Review of Anthropology 34: 269–91. Briggs, C. 2007. Anthropology, interviewing and communicability in contemporary society. Current Anthropology 48: 551–80. Briggs, C., and R. Bauman. 1992. Genre, intertextuality and social power. Journal of Linguistic Anthropology 2: 131–72. Briggs, C., with C. Mantini-Briggs. 2003. Stories in the time of cholera: Racial profiling during a medical nightmare. Berkeley: University of California Press. Brown, M. 2003. Who owns native culture? Cambridge, MA: Harvard University Press. Brown, R., and A. Gillman. 1960. The pronouns of power and solidarity. In Style in language, ed. T. Sebeok, 253–76. Cambridge, MA: MIT Press. Bucholtz, M. 2000. The politics of transcription. Pragmatics 32: 1439–65. Bucholtz, M., A. C. Liang, and L. A. Sutton. 1999. Reinventing identities: The gendered self in discourse. New York: Oxford University Press. Bucholtz, M., and S. Trechter. 2008. Introduction: Bringing language into whiteness studies. Journal of Linguistic Anthropology 11: 3–21. Calton, C. 2012. Language ideology in the American Sign Language classroom. Deaf Studies Today! 2008 and 2010 Conference Proceedings, ed. B. Eldredge, K. Morton, and J. Legg. Orem: Utah Valley University. Cameron, D. 1995. Verbal hygiene. London: Routledge. Castañeda, Q. E. 2006. Ethnography in the forest: An analysis of ethics in the morals of anthropology. Current Anthropology 21: 121–45. Caton, S. 1990. “Peaks of Yemen I summon”: Poetry as cultural practice in a North Yemeni tribe. Berkeley: University of California Press. Chafe, W. 1993. Seneca speaking styles and the location of authority. In Responsibility and evidence in oral discourse, ed. J. Hill and J. Irvine, 72–87. Cambridge: Cambridge University Press. Chapman, M. 1982. Semantics and the Celt. In Semantic anthropology, ed. D. Parkin, 122–43. London, Academic Press. Chernela, J. 2011. The second world of Wanano women: Truth, lies, and back-talk in the northwest Brazilian Amazon. Journal of Linguistic Anthropology 21: 193–210. Chun, E. W. 2004. Ideologies of legitimate mockery: Margaret Cho’s revoicings of mock Asian. Pragmatics 14: 263–90. Clark, R., and R. Ivanič. 1997. The politics of writing. London: Routledge. Conley, J., and J. O’Barr. 2005. Just words: Law, language and power. Chicago: Chicago University Press. Coombe, E. 1998. The cultural life of intellectual properties: Authorship, appropriation, and the law. Durham, NC: Duke University Press. Coombe, E. 2011. Making and unmaking intellectual property: Creative production in legal and cultural perspective. Chicago: University of Chicago Press. Darnell, R. 1998 [1974]. Correlates of Cree performance. In Explorations in the ethnography of speaking, 2nd ed., ed. R. Bauman and J. Sherzer, 315–36. New York: Cambridge University Press. Davies, B., and R. Harré. 1990. Positioning: The discursive production of selves. Journal for the Theory of Social Behavior 20: 43–63. Du Bois, J. 1993 [1987]. Meaning without intention: Lessons from divination. Papers in Pragmatics 1: 80–122. (Reprinted in 1993, Responsibility and evidence in oral discourse, Cambridge: Cambridge University Press.)

14_107-Bernard.indb 425

6/9/14 10:03 AM

426

Brenda Farnell and Laura R. Graham

Duranti, A. 1986. The audience as co-author. Text 6: 239–47. Duranti, A. 1992. Language and bodies in social space: Samoan ceremonial greetings. American Anthropologist 94: 657–91. Duranti, A. 1993. Intentions, self, and responsibility: An essay in Samoan ethno-pragmatics. In Responsibility and evidence in oral discourse, ed. J. Hill and J. Irvine, 24–47. Cambridge: Cambridge University Press. Duranti, A. 1997. Linguistic anthropology. Cambridge: Cambridge University Press. Duranti, A., and C. Goodwin, eds. 1992. Rethinking context: Language as an interactive phenomenon. Cambridge: Cambridge University Press. Dureau, C. 2012. Translating love. Ethos 40: 142–63. Edwards, J. A., and M. D. Lampert. 1993. Talking data: Transcription and coding in discourse research. Hillsdale, NJ: Lawrence Erlbaum. Ellis, J. 1994. Language, thought, and logic. Evanston, IL: Northwestern University Press. Erickson, F., and J. Shultz. 1982. The counselor as gatekeeper: Social interaction in interviews. New York: Academic Press. Fairclough, N. 1989. Critical discourse analysis: The critical study of language. Harlow, UK: Longman. Fairclough, N. 1995a. Language and power. London: Longman. Fairclough, N. 1995b. Media discourse. London: Edward Arnold. Fairclough, N. 2003. Analyzing discourse: Textual analysis for social research. London: Routledge. Farnell, B. 1994. Ethno-graphics and the moving body. Man: Journal of the Royal Anthropological Institute 29: 929–74. Farnell, B. 1995a. Do you see what I mean?: Plains Indian sign talk and the embodiment of action. Austin: University of Texas Press. Farnell, B. 1995b. Introduction. In Human action signs in cultural context: The visible and the invisible in movement and dance, ed. B. Farnell, 1–28. Metuchen, NJ: Scarecrow Press. Farnell, B. 1995c. Wiyuta: Assiniboine storytelling with signs. CD-ROM. Austin: University of Texas Press. Farnell, B. 1996. Movement writing systems. In The world’s writing systems, ed. P. Daniels and W. Bright, 855–79. Oxford, England: Oxford University Press. Farnell, B. 1999. Moving bodies, acting selves. Annual Review of Anthropology 28: 341–73. Farnell, B. 2001. Rethinking verbal and non-verbal in discursive performance. Textus 14: 417–36. Farnell, B. 2011. Theorizing the body in visual anthropology. In Made to be seen: Perspectives on the history of visual anthropology, ed. M. Banks and J. Ruby, 136–58. Chicago: University of Chicago Press. Farnell, B. 2012. Dynamic embodiment for social theory: “I Move Therefore I Am.” London: Routledge. Farnell, B., and C. Varela. 2008. The second somatic revolution. Journal for the Theory of Social Behavior 38: 215–40. Feld, S. 2002. Bells and winter festivals of Greek Macedonia. Washington, DC: Smithsonian Folkways Recordings. Feld, S. 2012 [1982]. Sound and sentiment: Birds, weeping, poetics, and song in Kaluli expression, 2nd ed. Philadelphia: University of Pennsylvania Press. Feld, S., and A. Fox. 1994. Music and language. Annual Review of Anthropology 23: 25–53. Fenigsen, J. 1999. “A broke-up mirror”: Representing Bajan in print. Cultural Anthropology 14: 61–87. Fishman, J. 1966. Language loyalty in the United States. Amsterdam: Mouton. Fishman, J. 1968. Readings in the sociology of language. The Hague: Mouton. Fishman, J. 1970. Sociolinguistics: A brief introduction. Rowley, MA: Newbury House.

14_107-Bernard.indb 426

6/9/14 10:03 AM

12: Discourse-Centered Methods

427

Fleuhr-Lobban, C. 1998. Ethics. In Handbook of methods in cultural anthropology, ed. H. R. Bernard, 173–202. Walnut Creek, CA: Sage. Foucault, M. 1972. The archaeology of knowledge and the discourse on language, trans. A. M. Sheridan Smith. New York: Pantheon Books. Fox, A. 2004. Real country: Music and language in working-class culture. Durham, NC: Duke University Press. French, B. 2010. Maya ethnolinguistic identity: Violence, cultural rights, and modernity in highland Guatemala. Tucson: University of Arizona Press. Gal, S. 1991. Between speech and silence: The problematics of research on language and gender. In Gender at the crossroads of knowledge: Feminist anthropology in the postmodern era, ed. M. Di Leonardo, 175–203. Berkeley: University of California Press. Garfinkel, H. 1967. Studies in ethnomethodology. Englewood Cliffs, NJ: Prentice Hall. Garfinkel, H. 1972. Remarks on ethnomethodology. In Directions in sociolinguistics, ed. J. Gumperz, 301–24. New York: Holt, Rinehart and Winston. Garvey. C. 1984. Children’s talk. Cambridge: Harvard University Press. Gaudio, R. P. 2009. Allah made us: Sexual outlaws in an Islamic African city. Chichester, UK: Wiley-Blackwell. Gee, J. P. 2011a [1999]. An introduction to discourse analysis: Theory and method. New York: Routledge. Gee, J. P. 2011b. How to do discourse analysis: A tool kit. New York: Routledge. Gelatt, R. 1977. The fabulous phonograph 1877–1977, 2nd rev. ed. New York: Macmillan. Gershon, I. 2010. The breakup 2.0: Disconnecting over new media. Ithaca, NY: Cornell University Press. Giddens, A. 1979. Central problems in social theory: Action, structure and contradiction in social analysis. Berkeley: University of California Press. Giddens, A. 1984. The constitution of society. Berkeley: University of California Press. Ginsburg, F. 2007. Shooting back: From ethnographic film to indigenous production/ethnography of media. New York: Wiley. Goddard, I. 1996. The description of the native languages of North America before Boas. In Handbook of North American Indians, Vol. 17, ed. W. C. Sturtevant, 17–42. Washington DC: Smithsonian Institution Press. Goffman, E. 1961. Encounters: Two studies in social interaction. Indianapolis: Bobbs-Merrill. Goffman, E. 1972 [1967]. Interaction ritual. Harmondsworth, UK: Penguin Books. Goffman, E. 1974. Frame analysis. New York: Harper and Row. Goffman, E. 1981. Forms of talk. Philadelphia: University of Philadelphia Press. Goodman, J. 2005. Berber culture on the world stage: From village to video. Bloomington: Indiana University Press. Goodwin, C. 1981. Conversational organization: Interaction between speakers and hearers. New York: Academic Press. Goodwin, C. 1993. Recording human interaction in natural settings. Pragmatics 3: 181–209. Goodwin, C., and A. Duranti. 1992. Rethinking context: An introduction. In Rethinking context: Language as an interactive phenomenon, ed. C. Goodwin and A. Duranti, 1–42. Cambridge: Cambridge University Press. Goodwin, C., and M. Goodwin. 2006. Participation. In A companion to linguistic anthropology, ed. A. Duranti, 222–44. Oxford: Blackwell, Goodwin, C., and J. Heritage. 1990. Conversational analysis. Annual Review of Anthropology 19: 283–307. Goodwin, M. 1982a. “Instigating”: Storytelling as a social process. American Ethnologist 9: 799–819.

14_107-Bernard.indb 427

6/9/14 10:03 AM

428

Brenda Farnell and Laura R. Graham

Goodwin, M. 1982b. Processes of dispute management among urban black children. American Ethnologist 9: 76–96. Goodwin, M. Harness. 1990. He-said-she-said: Talk as social organization among black children. Bloomington: Indiana University Press. Goodwin, M. Harness. 2006. The hidden life of girls: Games of stance, status and exclusion. Malden, MA: Blackwell. Graham, L. R. 1984. Semanticity and melody: Parameters of contrast in Shavante vocal expression. Latin American Music Review 5: 161–85. Graham, L. R. 1986. Three modes of Shavante vocal expression: Wailing, collective singing, and political oratory. In Native South American discourse, ed. J. Sherzer and G. Urban, 83–118. Berlin: Mouton. Graham, L. R. 1990. The always living: Discourse and the male lifecycle of the Xavante Indians of Central Brazil. Ph.D. diss., Austin: University of Texas. Graham, L. R. 1993. A public sphere in Amazonia? The depersonalized collaborative construction of discourse in Xavante. American Ethnologist 20: 717–41. Graham, L. R. 1995. Performing dreams: Discourses of immortality among the Xavante of central Brazil. Austin: University of Texas Press. Graham, L. R. 2000. The one who created the sea: Tellings, meanings, and inter-textuality in the translation of Xavante narrative. In Translating Native Latin American verbal art: Ethnopoetics and ethnography, ed. K. Sammons and J. Sherzer, 252–71. Washington, DC: Smithsonian Institution Press. Graham, L. R. 2011. Quoting Mario Juruna: Linguistic imagery and the transformation of indigenous voice in the Brazilian print press. American Ethnologist 38: 164–82. Graham, L. R. In press. Do sonho á performance: Discursos de imortalidade nos Xavante, with audio CD. Editora Universidade de São Paulo. Graham, L. R., D. Hernández, and C. Waiásse. 2009. Owners of the water: Conflict and collaboration over rivers. 34 min. Watertown, MA: Documentary Educational Resources. Grimshaw, A. D., ed. 1990. Conflict talk: Sociolinguistic investigations of arguments in conversations. Cambridge: Cambridge University Press. Grimshaw, A. D. 1994. What’s going on here? Complementary studies of professional talk. Norwood, NJ: Ablex Press. Gumperz, J. J. 1982. Discourse strategies. Cambridge: Cambridge University Press. Gumperz, J. J., and D. Hymes, eds. 1964. The ethnography of communication. American Anthropologist, Special Publication 66, Number 6, Part II. Gumperz, J. J., and D. Hymes, eds. 1986 [1972]. Directions in sociolinguistics: The ethnography of communication. Oxford: Basil Blackwell. Hall, K., and M. Bucholtz. 1995. Gender articulated: Language and the socially constructed self. New York: Routledge. Hanks, W. 1990. Referential practice: Language and lived space among the Maya. Chicago: University of Chicago Press. Harré, R., ed. 1984. Personal being. Cambridge, MA: Harvard University Press. Harré, R. 1986a. Mind as a social formation. In Rationality, relativism and the human sciences, ed. J. Margolis, M. Krausz, and R. M. Burien, 91–106. Dortrecht, the Netherlands: Martinus Nijhoff. Harré, R. 1986b. The social construction of emotions. Oxford: Basil Blackwell. Harré, R. 1987. The social construction of selves. In Self and identity: Psychosocial perspectives, ed. K. Yardley and T. Honess, 41–52. New York: John Wiley. Harré, R., and G. Gillett. 1994. The discursive mind. London: Sage. Haugen, E. 1969. The Norwegian language in America. Bloomington: Indiana University Press.

14_107-Bernard.indb 428

6/9/14 10:03 AM

12: Discourse-Centered Methods

429

Haviland, J. 1993. Anchoring, iconicity and orientation in Guugu Yimithirr pointing gestures. Journal of Linguistic Anthropology 3: 3–45. Haviland, J. 1996. Text from talk in Tzotzil. In Natural histories of discourse, ed. M. Silverstein and G. Urban, 45–78. Chicago: University of Chicago Press. Haviland, J. 2000. Early pointing gestures in Zinacantan. Journal of Linguistic Anthropology 8: 162–96. Haviland, J. 2006. Gesture. In A companion to linguistic anthropology, ed. A. Duranti, 197–221. Oxford: Blackwell. Heath, S. 1983. Ways with words: Language, life and work in communities and classrooms. Cambridge: Cambridge University Press. Heller, M. 2011. Paths to post-nationalism: A critical ethnography of language and identity. New York: Oxford University Press. Hill, J. 1993. Hasta la vista baby: Anglo Spanish in the American Southwest. Critique of Anthropology 13: 145–76. Hill, J. 1998. Language, race and white public space. American Anthropologist 100: 680–89. Hill, J. 2002. “Expert rhetorics” in advocacy for endangered languages: Who is listening and what do they hear? Journal of Linguistic Anthropology 12: 119–33. Hill, J. 2008. The everyday language of white racism. Oxford: Wiley Blackwell. Hill, J. 2009. Made from bone: Trickster myths, music and history from the Amazon. Urbana: University of Illinois Press. Hill, J., and J. Irvine. 1993. Responsibility and evidence in discourse. Cambridge: Cambridge University Press. Hill, J., and O. Zepeda. 1993. Mrs. Patricio’s trouble: The distribution of responsibility in an account of personal evidence. In Responsibility and evidence in oral discourse, ed. J. Hill and J. Irvine, 197–225. Cambridge: Cambridge University Press. Hirschkind, C. 2006. The ethical soundscape: Cassette sermons and Islamic counterpublics. New York: Columbia University Press. Holquist, M. 1983. The politics of representation. Quarterly Newsletter of the Laboratory of Comparative Human Cognition 5: 2–9. Hymes, D. 1962. The ethnography of speaking. In Anthropology and human behavior, ed. T. Gladwin and W. C. Sturtevant, 13–53. Washington, DC: Anthropological Society of Washington. Hymes, D. 1964. Introduction: Toward ethnographies of communication. In The ethnography of communication, ed. J. J. Gumperz and D. Hymes. American Anthropological Association Special Publication 66, Number 6, Part II: 1–34. Hymes, D. 1967. Models of the interaction of language and social setting. Journal of Social Issues 23: 8–28. Hymes, D. 1971. Sociolinguistics and the ethnography of speaking. In Social anthropology and language, ed. E. Ardener, 47–94. London: Tavistock Press. Hymes, D. 1974. Foundations in sociolinguistics: An ethnographic approach. Philadelphia: University of Pennsylvania Press. Hymes, D. 1975. The pre-war Prague school and post-war American anthropological linguistics. In Transformational generative paradigm and modern linguistic theory, ed. E. K. F. Koerner, 359–81. Amsterdam: John Benjamins. Hymes, D. 1981. “In vain I tried to tell you”: Essays in Native American ethnopoetics. Philadephia: University of Pennsylvania Press. Inoue, M., ed. 2004. Introduction: Temporality and historicity in and through linguistic ideology. Journal of Linguistic Anthropology, Special issue: 14: 1–5. Inoue, M. 2006. Vicarious language: Gender and linguistic modernity in Japan. Berkeley: University of California Press.

14_107-Bernard.indb 429

6/9/14 10:03 AM

430

Brenda Farnell and Laura R. Graham

Irvine, J. 1989. When talk isn’t cheap: Language and political economy. American Ethnologist 16: 248–67. Irvine, J. 1992. Ideologies of honorific language. Pragmatics 2: 251–62. Irvine, J. 1996. Shadow conversations: The indeterminacy of participant roles. In Natural histories of discourse, ed. G. Urban and M. Silverstein, 131–59. Chicago: University of Chicago Press. Irvine, J., and S. Gal. 2000. Language ideology and linguistic differentiation. In Regimes of language, ed. P. Kroskrity, 35–83. Santa Fe: SAR Press. Ives, E. D. 1995. The tape recorded interview: A manual for field workers in folklore and oral history, 2nd ed. Knoxville: University of Tennessee Press. Jaffe, A. 1999. Ideologies in action: Language politics in Corsica. Berlin: Mouton de Gruyter. Jaffe, A., J. Androutsopoulos, M. Sebba, and S. Johnson, eds. 2012. Orthography as social action: Scripts, spelling, identity and power. The Hague: De Gruyter Mouton. Jakobson, R. 1960. Linguistics and poetics. In Style in language, ed. T. Sebeok, 350–77. Cambridge, MA: MIT Press. Jakobson, R. 1971 [1957]. Shifters, verbal categories and the Russian verb. In Selected writings of Roman Jakobson, ed. R. Jakobson, 130–47. The Hague: Mouton. Jaworski, A., and N. Coupland, eds. 2002. The discourse reader. New York: Routledge. Kaeppler, A. 1995. Visible and the invisible in Hawaiian dance. In Human action signs in cultural context: The visible and the invisible in movement and dance, ed. B. Farnell, 31–43. Metuchen NJ: Scarecrow Press. Keane, W. 1991. Delegated voice: Ritual speech, risk, and the making of marriage alliances in Anakalang. American Ethnologist 18: 311–30. Keating, E. 1998. Power sharing: Language, rank, gender and social space in Pohnpei, Micronesia. Oxford: Oxford University Press. Keating, E. 2000. Space. Journal of Linguistic Anthropology 9: 234–37. Keating, E., and G. Mirus. 2003. American Sign Language in virtual space: Interactions between deaf users and computer-mediated video communication and the impact of technologies on language practice. Language in Society 32: 693–714. Keating, E., and C. Sunakawa. 2010. Participation cues: Coordinating activity and collaboration in complex online gaming worlds. Language in Society 39: 331–56. Kendon, A. 1980. Gesticulation and speech: Two aspects of the process of utterance. In The relationship of verbal and non-verbal communication, ed. M. R. Key, 207–27. The Hague: Mouton. Kendon, A. 1983. Gesture and speech: How they interact. In Nonverbal interaction, ed. J. M. Wieman and R. P. Harrison, 13–45. Sage Annual Reviews of Communication Research, Vol. 11. London: Sage. Kendon, A. 1989. Sign languages of Aboriginal Australia. Cambridge: Cambridge University Press. Kennedy, G., A. Arnold, P. Dugdale et al. 1997. A dictionary of New Zealand sign language. Auckland: Auckland University Press. Kovats-Bernat, C. J. 2002. Negotiating dangerous fields: Pragmatic strategies for fieldwork amid violence and terror. American Anthropologist 104: 208–22. Kroskrity, P., B. Schieffelin, and K. Woolard, eds. 1992. Language ideologies. Pragmatics, Special issue: 2(3). Kuipers, J. 1990. Power in performance: The creation of textual authority in Weyewa ritual speech. Philadelphia: University of Pennsylvania Press. Kulick, D. 1992. Language shift and cultural reproduction: Socialization, self and syncretism in a Papua New Guinean village. New York: Cambridge University Press. Labov, W. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.

14_107-Bernard.indb 430

6/9/14 10:03 AM

12: Discourse-Centered Methods

431

Laderman, C. 1987. The ambiguity of symbols in the structure of healing. Social Science and Medicine 24: 293–301. Lederman, R. 1963. The boundaries of speech and song. Ethnomusicology 7: 1–16. Lederman, R. 1984. Who speaks here? In Dangerous words: Language and politics in the Pacific, ed. D. L. Brenneis and F. R. Myers, 85–107. New York: New York University Press. LeMaster, B., and L. Monaghan. 2002. Sign languages. In A companion to linguistic anthropology, ed. A Duranti, 141–65. Malden, MA: Blackwell. Lippi-Green, R. 1997. English with an accent: Language, ideology, and discrimination in the United States. New York: Routledge. List, G. 1963. The boundaries of speech and song. Ethnomusicology 7: 1–16. Lo, A. 1999. Codeswitching, speech community membership, and the construction of ethnic identity. Journal of Sociolinguistics 3: 461–79. Lo, A., and H. Fung. 2011. Language socialization and shaming. In The handbook of language socialization, ed. A. Duranti, E. Ochs, and B. Schieffelin, 169–89. Malden, MA: Wiley-Blackwell. Lo, A., and A. Reyes, eds. 2009. Beyond yellow English: Toward a linguistic anthropology of Asian Pacific America. New York: Oxford University Press. Low, S., and S. Merry. 2010. Engaged anthropology: Diversity and dilemmas, an introduction to Supplement 2. Current Anthropology 51: S203–S226. Lucy, J., ed. 1990. Reflexive language: Reported speech and metapragmatics. Cambridge: Cambridge University Press. Lutz, C. 1988. Unnatural emotions: Everyday sentiments on a Micronesian atoll and their challenge to Western theory. Chicago: University of Chicago Press. Lutz, C., and L. Abu-Lughod, eds. 1990. Language and the politics of emotion. Cambridge: Cambridge University Press. MacMahon, M. K. C. 1996. Phonetic notation. In The world’s writing systems, ed. P. T. Daniels and W. Bright, 821–46. Oxford: Oxford University Press. Makihara, M., and B. Schieffelin. 2007. Consequences of contact: Language ideologies and sociocultural transformations in Pacific societies. Oxford: Oxford University Press. Mannheim, B., and D. Tedlock. 1995. Introduction. In The dialogic emergence of culture, ed. B. Mannheim and D. Tedlock, 1–32. Urbana: University of Illinois Press. Manuel, P. L. 1993. Cassette culture: Popular music and technology in north India. Chicago: University of Chicago Press. McClendon, S. 1981. Meaning, rhetorical structure, and discourse organization in myth. In Analyzing discourse: Text and talk, ed. D. Tannen, 284–305. Georgetown University Roundtable on Language and Linguistics. Washington DC: Georgetown University Press. McElhinny, B. S. 2007. Words, worlds, and material girls. Berlin: Mouton de Gruyter. McNeill, D. 1985. So you think gestures are non-verbal? Psychological Review 92: 350–71. McNeill, D. 1992. Hand and mind: What gestures reveal about thought. Chicago: University of Chicago Press. McNeill, D., ed. 2000. Language and gesture. Cambridge: Cambridge University Press. Meek, B. A. 2006. And the Injun goes “How!”: Representations of American Indian English in white public space. Language in Society 35: 93–128. Meek, B. A. 2007. Respecting the languages of elders: Ideological shift and linguistic discontinuity in a northern Athapascan community. Journal of Linguistic Anthropology 17: 23–43. Meintjes, L. 2003. Sound of Africa! Making music Zulu in a South African studio. Durham, NC: Duke University Press. Mendoza-Denton, N. 2008. Homegirls: Language and cultural practice among Latina youth gangs. Malden, MA: Blackwell.

14_107-Bernard.indb 431

6/9/14 10:03 AM

432

Brenda Farnell and Laura R. Graham

Mertz, E., and R. Parmentier. 1985. Introduction. In Semiotic mediation, ed. E. Mertz and R. Parmentier, 1–19. New York: Academic Press. Michaels, E. 1994. Bad Aboriginal art: Tradition, media and technological horizons. Minneapolis: University of Minnesota Press. Miller, F. 2007. The moral resonance of Arab media: Audio cassettes, poetry and culture in Yemen. Cambridge, MA: Harvard Middle Eastern Monographs. Mithun, M. 1996. The description of the native languages of North America: Boas and after. In Handbook of North American Indians, Vol. 17, ed. W. C. Sturtevant, 43–63. Washington, DC: Smithsonian Institution Press. Moeran, B. 2007. From participant observation to observant participation: Anthropology, fieldwork and organizational ethnography. Creative Encounters Working Papers #2, Copenhagen Business School. http://openarchive.cbs.dk/bitstream/handle/10398/7038/wp%20 2007-2.pdf?sequence=1 (accessed March 16, 2013). Moerman, M. 1988. Talking culture: Ethnography and conversation analysis. Philadelphia: University of Pennsylvania Press. Monaghan L., K. Nakamura, C. Schmaling, and G. H. Turner. 2003. Many ways to be deaf. Washington, DC: Gallaudet University Press. Mori, J. 1999. Negotiating agreement and disagreement in Japanese: Connective expressions and turn construction. Amsterdam: John Benjamins. Morgan, M. 2007. The real hiphop: Battling for knowledge, power and respect in the underground. Durham, NC: Duke University Press. Mühlhäusler, P., and R. Harré. 1990. Pronouns and people: The linguistic construction of social and personal identity. Oxford: Basil Blackwell. Murphy, W. 1990. Creating the appearance of consensus in Mende political discourse. American Anthropologist 92: 24–41. Murray, T. E., and C. R. Murray. 1992. Legal and ethical issues in surreptitious recording. Publication of the American Dialect Society 76. Tuscaloosa: University of Alabama Press. Myers-Scotton, C. 1993. Social motivations for codeswitching: Evidence from Africa. Oxford: Clarendon Press. Myers-Scotton, C. 2006. Multiple voices: An introduction to bilingualism. Malden, MA: Blackwell. Negrón, R. 2011. Ethnic identification among urban Latinos: Language and flexibility. El Paso, TX: LFB Scholarly Publishing. Nuckolls, J. 1992. Sound symbolic involvement. Journal of Linguistic Anthropology 2: 51–80. Nuckolls, J. 2010. Lessons from a Quechua strongwoman: Ideophony dialogue and perspective. Tucson: University of Arizona Press. Ochs, E. 1979a. Introduction. In Developmental pragmatics, ed. E. Ochs and B. B. Schieffelin, 1–17. New York: Academic Press. Ochs, E. 1979b. Transcription as theory. In Developmental pragmatics, ed. E. Ochs and B. B. Schieffelin, 43–72. New York: Academic Press. Ochs, E. 1988. Culture and language development: Language acquisition and language socialization in a Samoan village. New York: Cambridge University Press. Page, J. 1990. A comparative study of two movement writing systems: Laban and Benesh notation. Master’s Thesis, University of Sydney, Sydney, Australia. Page, J. 1996. Images for understanding: Movement notations and visual recordings. Visual Anthropology 8: 171–96. Perley, B. 2011. Defying Malaseet language death: Emergent vitalities of language, culture and identity in eastern Canada. Lincoln: University of Nebraska Press. Peterson, D. J. 2011. SLIPA: An IPA for signed languages. http://www.dedalvs.com/slipa.html (accessed February 10, 2013).

14_107-Bernard.indb 432

6/9/14 10:03 AM

12: Discourse-Centered Methods

433

Philips, S., S. Steele, and C. Tanz, eds. 1987. Language, gender and sex in comparative perspective. Cambridge: Cambridge University Press. Pink, S. 2007 [2001]. Doing visual anthropology, 2nd ed. London: Sage. Porcello, T. 2005. Wired for sound: Engineering and technologies in sonic cultures. Middletown, CT: Weslyan University Press. Pratt, M. L. 1987. The ideology of speech act theory. Centrum N.S. 1: 5–18. Pressman, J. 1997. Honorification and projection on Saint Barthélemy. Anthropological Linguistics 39: 111–50. Quinn, N. 2005. Finding culture in talk: A collection of methods. New York: Palgrave MacMillian. Radin, P. 1949. The culture of the Winnebago: As described by themselves. Indiana University Publications in Anthropology and Linguistics, Memoir 2. Baltimore: Waverly Press. Reber, E. 2012. Affectivity in interaction: Sound objects in English. Amsterdam/Philadelphia: Johns Benjamins. Rhydwen, M. 1993. Kriol: The creation of a written language and a tool of colonization. In Language and culture in Aboriginal Australia, ed. M. Walsh and C. Yallop, 155–68. Canberra, Australia: Aboriginal Studies Press. Rorty, R., ed. 1992 [1967]. The linguistic turn. Chicago: University of Chicago Press. Rosaldo, M. 1982. The things we do with words: Ilongot speech acts and speech act theory in philosophy. Language in Society 11: 203–35. Rosaldo, R. 1980. Ilongot headhunting 1883–1974: A study in society and history. Stanford, CA: Stanford University Press. Roseman, M. 1991. Healing sounds from the Malaysian rainforest: Temiar music and medicine. Berkeley: University of California Press. Roth-Gordon, J. 2009. The language that came down the hill: Slang, crime and citizenship in Rio de Janeiro. American Anthropologist 111: 57–68. Sacks, H., E. A. Schegloff, and G. Jefferson. 1974. A simplest systematics for the organization of turn-taking conversation. Language 50: 696–735. Sahlins, M. 1976. Culture and practical reason. Chicago: University of Chicago Press. Sahlins, M. 1985. Islands of history. Chicago: Chicago University Press. Salomon, F., and M. Nino-Murcia. 2011. The lettered mountain: A Peruvian village’s way with writing. Durham, NC: Duke University Press. Sammons, K., and J. Sherzer. 2000. Translating Native American verbal art: Ethnopoetics and ethnography of speaking. Washington, DC: Smithsonian Institution Press. Samuels, D. 2004. Putting a song on top of it: Expression and identity on the San Carlos Apache Reservation. Tucson: University of Arizona Press. Sapir, E. 1949. Selected writings of Edward Sapir in language, culture and personality, ed. D. G. Mandelbaum. Berkeley: University of California Press. Saussure, F. de. 1916. Course de linguistique générale. C. Bally and A. Sechehaye, eds. Paris: Payot. Saville-Troike, M. 1989. The ethnography of communication: An introduction, 2nd ed. New York: Basil Blackwell. Schafer, M. R. 1977. The tuning of the world. New York: Knopf. Schafer, M. R. 1994. The soundscape: Our sonic environment and the tuning of the world. Rochester, VT: Destiny Books. Schegloff, E. A. 1972. Sequencing in conversational opening. In Directions in sociolinguistics, ed. J. Gumpertz, 346–80. New York: Holt, Reinhart and Winston. Schegloff, E. A., and H. Sacks. 1973. Opening up closings. Semiotica 7: 289–327. Schieffelin, B. B. 1990. The give and take of everyday life: Language socialization of Kaluli children. New York: Cambridge University Press.

14_107-Bernard.indb 433

6/9/14 10:03 AM

434

Brenda Farnell and Laura R. Graham

Schieffelin, B. B., and E. Ochs. 1986a. Language socialization. Annual Review of Anthropology 15:163–91. Schieffelin, B. B., and E. Ochs. 1986b. Language socialization across cultures. New York: Cambridge University Press. Schieffelin, B. B., and R. C. Doucet. 1994. The “real” Haitian Creole: Ideology, metalinguistics, and orthographic choice. American Ethnologist 21: 176–200. Schieffelin, B. B., K. A. Woolard, and P. V. Kroskrity, eds. 1998. Language ideologies: Practice and theory. New York: Oxford University Press. Searle, J. R. 1969. Speech acts: An essay in the philosophy of language. Cambridge: Cambridge University Press. Seeger, A. 1986. Oratory is spoken, myth is told, and song is sung, but they are all music to my ears. In Native South American discourse, ed. J. Sherzer and G. Urban, 59–82. Berlin: Mouton. Seeger, A. 1991. Singing other peoples’ songs. Cultural Survival Quarterly 15: 36–39. Seeger, A. 1997. Ethnomusicology and music law. In Borrowed power: Essays on cultural appropriation, ed. B. Ziff and P. V. Rao, 52–67. New Brunswick, NJ: Rutgers University Press. Seeger, A. 2004a. The selective protection of musical ideas: The “creators” and the dispossessed. In Property in question: Value transformation in the global economy, ed. K. Verdery and C. Humphrey, 69–84. Oxford, UK: Berg. Seeger, A. 2004b [1987]. Why Suyá sing: A musical anthropology of an Amazonian people. Urbana: University of Illinois Press. (Originally published, Cambridge: Cambridge University Press.) Senghas R., and L. Monaghan. 2002. Signs of their times: Deaf communities and the culture of language. Annual Review of Anthropology 31: 69–97. Shannon, J. 2007. Informed consent: Documenting the intersection of bureaucratic regulation and ethnographic practice. PoLAR: Political and Legal Anthropology Review 30: 229–48. Sherzer, J. 1973. Verbal and nonverbal deixis: The pointed lip gesture among the San Blas Kuna. Language in Society 2: 117–31. Sherzer, J. 1987. A discourse-centered approach to language and culture. American Ethnologist 89: 295–309. Sherzer, J. 1992. Modes of representation and translation of Native American discourse: Examples from the San Blas Kuna. In On the translation of Native American literatures, ed. B. Swann, 426–40. Washington, DC: Smithsonian Institution Press. Sherzer, J., and R. Darnell. 1972. Outline guide for the ethnographic study of speech use. In Directions in sociolinguistics: The ethnography of communication, ed. J. J. Gumperz and D. Hymes, 548–54. New York: Holt. Sherzer, J., and G. Urban. 1986a. Introduction. In Native South American discourse, ed. J. Sherzer and G. Urban, 1–14. Berlin: Mouton. Sherzer, J., and G. Urban, eds. 1986b. Native South American discourse. Berlin: Mouton. Sherzer, J., and S. Wicks. 1982. The intersection of language and music in Kuna discourse. Latin American Music Review 3: 147–64. Shuman, A. 2005. Redemption and empathy in junk-mail narratives. In Other people’s stories: Entitlement claims and the critique of empathy, ed. A. Shuman, 120–48. Urbana: University of Illinois Press. Shuy, R. 1986. Ethical issues in analyzing FBI surreptitious tapes. International Journal of the Sociology of Language 62: 119–28. Sidnell, J., ed. 2009. Conversation analysis: Comparative perspectives. Cambridge: Cambridge University Press. Sidnell, J. 2010. Conversation analysis: An introduction. Chichester, UK: Wiley Blackwell. Sidnell, J., and T., Stivers, eds. 2012. Handbook of conversation analysis. Chichester, UK: WileyBlackwell.

14_107-Bernard.indb 434

6/9/14 10:03 AM

12: Discourse-Centered Methods

435

Silverstein, M. 1976. Shifters, linguistic categories, and cultural description. In Meaning in anthropology, ed. K. H. Basso and H. A. Selby, 11–56. Albuquerque: University of New Mexico Press. Silverstein, M. 1977. Cultural prerequisites to grammatical analysis. In Georgetown University round table on languages and linguistics: Linguistics and anthropology, ed. M. Saville-Troike, 139–51. Washington, DC: Georgetown University Press. Silverstein, M. 1979. Language structure and linguistic ideology. In The elements: A parasession on linguistic units and levels, ed. P. R. Cylne, W. Hanks, and C. L. Hofbauer, 193–248. Chicago: Chicago Linguistic Society. Silverstein, M. 1984. On the pragmatic “poetry” of prose. In Georgetown University roundtable in linguistics, Vol. 18, ed. D. Schiffrin, 181–99. Washington DC: Georgetown University Press. Silverstein, M. 1985. The culture of language in Chinookan narrative texts; or on saying that . . . in Chinook. In Grammar inside and outside the clause: Some approaches to theory from the field, ed. J. Nichols and A. C. Woodbury, 132–71. London: Cambridge University Press. Silverstein, M. 1993. Metapragmatic discourse and metapragmatic function. In Reflexive language: Reported speech and metapragmatics, ed. J. Lucy, 33–58. Cambridge: Cambridge University Press. Silverstein, M. 2001 [1981]. The limits of awareness. Linguistic anthropology: A reader, ed. A. Duranti. Malden, MA: Blackwell. Silverstein, M. 2003. Indexical order and the dialectics of sociolinguistic life. Language and Communication 23: 193–229. Silverstein, M., and G. Urban. 1996. Introduction. In Natural histories of discourse, ed. G. Urban and M. Silverstein, 1–20. Chicago: University of Chicago Press. Society for Linguistic Anthropology. Tools for linguistic anthropologists. http://kit.linguistic anthropology.org/about/ (accessed February 6, 2013). Spitulnik, D. 1993. Anthropology and mass media. Annual Review of Anthropology 22: 293–325. Spitulnik, D. 1996. The social circulation of media discourse and the mediation of communities. Journal of Linguistic Anthropology 6: 161–87. Spitulnik, D. 2010. Radio time sharing and the negotiation of linguistic pluralism in Zambia. Pragmatics 2: 335–54. Stokoe, W., D. C. Casterline, and C. G. Croneberg. 1965. A dictionary of American Sign Language on linguistic principles. Washington, DC: Gallaudet University Press. Streeck, J., C. Goodwin, and C. LeBaron. 2011. Embodied interaction: Language and body in the material world. New York: Cambridge University Press. Studdert-Kennedy, M., and E. Whitney Goodall. 1992. Gestures, features and segments in early child speech. Haskins Laboratory Status Report on Research. SR 111/112: 89–102. Swann, B., ed. 1992. On the translation of Native American literatures. Washington, DC: Smithsonian Institution Press. Tannen, D., ed. 1982. Analyzing discourse: Text and talk. Washington, DC: Georgetown University Press. Tannen, D. 1984. Conversational style: Analyzing talk among friends. Norwood, NJ: Ablex Press. Tannen, D. 1994. Gender and discourse. New York: Oxford University Press. Tannen, D. 1995. Waiting for the mouse: Constructed dialogue in conversation. In The dialogic emergence of culture, ed. B. Mannheim and D. Tedlock, 198–218. Urbana: University of Illinois Press. Taylor, C. 1985. Human agency and language. Cambridge: Cambridge University Press. Tedlock, D. 1983. The spoken word and the work of interpretation. Philadelphia: University of Pennsylvania Press.

14_107-Bernard.indb 435

6/9/14 10:03 AM

436

Brenda Farnell and Laura R. Graham

Tedlock, D. 1987. Hearing a voice in an ancient text: Quiché Maya poetics in performance. In Native American discourse, ed. J. Sherzer and A. Woodbury, 140–75. Cambridge: Cambridge University Press. Testa, R. 1988. Interruptive strategies in English and Italian conversation: Smooth versus contrastive linguistic preferences. Multilingua 7: 285–312. Thibault, P. J. 1997. Re-reading Saussure: The dynamics of signs in social life. London: Routledge. Thieberger, N. 2012. The Oxford handbook of linguistic fieldwork. Oxford, UK: Oxford University Press. Turner, L. D. 1969. Africanisms in the Gullah dialect. New York: Arno. Turner, T. 1991. Representing, resisting, rethinking: Historical transformations of Kayapo culture and anthropological consciousness. In Colonial situations. History of anthropology, Vol. 7, ed. G. Stocking, 285–313. Madison: University of Wisconsin Press. Urban, G. 1984. Speech about speech and speech about action. Journal of American Folklore 97: 310–28. Urban, G. 1985. Semiotics of two speech styles in Shokleng. In Semiotic mediation, ed. E. Mertz and R. Parmentier, 311–29. New York: Academic Press. Urban, G. 1986. Semiotic functions of macro-parallelism in the Shokleng origin myth. In Native South American discourse, ed. J. Sherzer and G. Urban, 15–58. Berlin: Mouton. Urban, G. 1988. Ritual wailing in Amerindian Brazil. American Anthropologist 90: 385–400. Urban, G. 1989. The “I” of discourse. In Semiotics, self, and society, ed. G. Urban and B. Lee, 27–51. New York: Mouton. Urban, G. 1991. A discourse-centered approach to culture: Native South American myths and rituals. Austin: University of Texas Press. Urban, G. 1996a. Entextualization, replication, and power. In Natural histories of discourse, ed. G. Urban and M. Silverstein, 21–44. Chicago: University of Chicago Press. Urban, G. 1996b. Metaphysical community: The interplay of the senses and the intellect. Austin: University of Texas Press. Urban, G. 2001. Metaculture: How culture moves through the world. Minneapolis: University of Minnesota Press. Urban, G. 2006. Metasemiosis and metapragmatic. Encyclopedia of language and linguistics, 2nd ed., ed. K. Brown, 88–91. Amsterdam: Elsevier. Urciuoli, B. 1995. The indexical structure of visibility. In Human action signs in cultural context, ed. B. Farnell, 189–215. Metuchen NJ: Scarecrow Press. Urciuoli, B. 1996. Exposing prejudice: Puerto Rican experiences of language, race and class. Boulder, CO: Westview Press. Urciuoli, B. 2009. Talking/not talking about race: The enregisterments of culture in higher education discourses. Journal of Linguistic Anthropology 19: 21–39. Valentine, L. 1995. Making it their own: Severn Ojibwe communicative practices. Toronto: University of Toronto Press. van Dijk, T. 1991. Racism and the press. London: Routledge. van Dijk, T. 1993. Elite discourse and racism. Newbury Park, CA: Sage. van Dijk, T. 2008. Discourse and power. Houndsmills, UK: Palgrave. Varela, C. R. 1995. Ethogenic theory and psychoanalysis: The unconscious as a social construction and a failed explanatory construct. Journal for the Theory of Social Behavior 25: 363–85. Varela, C. R., and R. Harré. 1996. Conflicting varieties of realism: Causal powers and the problems of social structure. Journal for the Theory of Social Behavior 26: 313–25. Verdery, K., and C. Humphrey. 2004. Property in question: Value transformation in the global economy. Oxford, UK: Berg

14_107-Bernard.indb 436

6/9/14 10:03 AM

12: Discourse-Centered Methods

437

Verschueren, J. 1985. What people say they do with words. Norwood, NJ: Ablex. Verschueren, J. 2004. Notes on the role of metapragmatic awareness in language use. In Metalanguage: Social and ideological perspectives, ed. A Jaworski, N. Coupland, and D, Galinski, 54–73. Amsterdam: Mouton de Gruyter. Voloshinov, V. N. 1973. Marxism and the philosophy of language. New York: Seminar Press. Washabaugh W. 1981. Sign language in its social context. Annual Review of Anthropology 10: 237–52. Washabaugh W. 1986. Five fingers for survival. Ann Arbor, MI: Karoma. Watson-Gegeo, K. A., and G. White, eds. 1990. Disentangling: Conflict discourse in Pacific societies. Stanford, CA: Stanford University Press. Webster, A. 2010. Imagining Navajo in the boarding school: Laura Tohe’s No Parole Today and the intimacy of language ideologies. Journal of Linguistic Anthropology 20: 39–62. White, G. M., and J. Kirkpatrick, eds. 1985. Person, self and experience: Exploring Pacific ethnopsychologies. Berkeley: University of California Press. Whitehead, N. L. 2009. Post human anthropology. Identities: Global Studies in Culture and Power 16: 1–32. Whorf, B. Lee. 1956. Language thought, and reality. Cambridge, MA: MIT Press. Wiget, A. 1987. Telling the tale: A performance analysis of a Hopi coyote story. In Recovering the word: Essays in Native American literature, ed. B. Swann and A. Krupat, 297–338. Berkeley: University of California Press. Wilce, J. 2009. Medical discourse. Annual Review of Anthropology 38: 199–215. Williams, D. 1975. The role of movement in selected symbolic systems. D. Phil., Oxford University, Oxford, UK. Williams, D. 1982. Semasiology. In Semantic anthropology, Vol. 22, ed. D. Parkin, 161–81. London: Academic Press. Williams, D. 2004 [1991]. Ten lectures on theories of the dance, 2nd ed. Urbana: University of Illinois Press. Williams, D. 1995. Space, intersubjectivity and the conceptual imperative: Three ethnographic cases. In Human action signs in cultural context: The visible and the invisible in movement and dance, ed. B. Farnell, 44–81. Metuchen, NJ: Scarecrow Press. Williams, D., ed. 1996. Special Issue: Human Action Signs. Visual Anthropology 8(3/4). Williams, D. 2004. Anthropology and the dance: Ten lectures, 2nd ed. Urbana: University of Illinois Press. Williams, D., and B. Farnell. 1990. A beginning text on movement writing for non-dancers. Canberra: Australian Institute for Aboriginal Studies. Woodbury, A. 1985. The functions of rhetorical structure: A study of central Alaskan Yupik Eskimo discourse. Language in Society 14: 153–90. Woodbury, A. 1987. Rhetorical structure in a central Alaskan Yupik Eskimo traditional narrative. In Native American discourse, ed. J. Sherzer and A. Woodbury, 176–239. Cambridge: Cambridge University Press. Woolard, K. 1989. Double talk: Bilingualism and the politics of ethnicity in Catalonia. Stanford, CA: Stanford University Press. Woolard, K. 1992. Language ideologies: Issues and approaches. Pragmatics 2: 235–49. Woolard, K. 1998. Introduction: Language ideology as a field of inquiry. In Language ideologies practice and theory, ed. K. Woolard, 3–47. New York: Oxford University Press. Woolard, K. 2006. Codeswitching. In A companion to linguistic anthropology, ed. A. Duranti, 73–94. Oxford: Blackwell. Woolard K., and B. B. Schieffelin 1994. Language ideology. Annual Review of Anthropology 23: 55–82.

14_107-Bernard.indb 437

6/9/14 10:03 AM

438

Brenda Farnell and Laura R. Graham

Zeitlyn, D. 2004. The gift of the gab—Anthropology and conversation analysis. Anthropos 99: 451–68. Zentella, A. C. 1997. Growing up bilingual: Puerto Rican children in New York. Malden, MA: Blackwell. Zentella, A. C., ed. 2005. Building on strength: Language and literacy in Latino families and communities. New York: Teachers College Press. Zentella, A. C., B. Urciuoli, and L. R. Graham. 2007. Problematic language assessment in the U.S. Census. Anthropology News 48: 10–11.

14_107-Bernard.indb 438

6/9/14 10:03 AM

CHAPTER 13

Visual Anthropology Fadwa El Guindi

The first Handbook of Method in Cultural Anthropology (Naroll and Cohen 1970) contained a brief overview of developments in visual anthropology. By 1998, visual anthropology had become a much more important part of the field and warranted a full chapter (El Guindi 1998). This chapter updates that earlier chapter and has been reorganized and reconceptualized. Visual anthropology is conceptualized within the anthropology tradition influencing the quality of data by adding new ways of data gathering and creative means of analysis. A genealogy of conceptual/methodological developments is represented in Figure 13.1. I have also added an analysis of part of a conversation between Gregory Bateson and Margaret Mead (Brand 1976, N.d.) in which they talk about the nature of the visual project and its value for anthropology. This is presented in a section called For God’s Sake, Margaret. HISTORY

It is common to assume that the advent of cinematic technology was first adopted for cinema (art and commercial recreation), but actually it was first used for scientific purposes. Early ethnographers, for example, exploited the new technology of moving

Figure 13.1. Geneology of visual anthropology (modified from El Guindi, 2004).

439

14_107-Bernard.indb 439

6/9/14 10:03 AM

440

Fadwa El Guindi

pictures to expand their capacity for data gathering and, indeed, the technology contributed to discovery. Those early adopters gave anthropology much to appreciate, but the use of visual tools is still resisted. Photography was and is the focus of the Colliers’ life project. John Collier, Jr.’s (1967) classic book, Visual Anthropology: Photography as a Research Method, put photography at center stage in anthropology. The book was modified and coedited by John Collier, Jr. and Malcolm Collier (Collier and Collier 1986), and remains a primary source on photography in anthropology. The Colliers’ photography lifts people’s lives and lands them on an anthropological ground against a landscape of artistic beauty. The other medium, film (or moving pictures), has grown in importance and has gained the most attention. Emilie de Brigard (1995) wrote an account on the development of film for anthropology (but also photography) that became the guiding historical essay in the field. It is still cited widely. In her account, and pervasively among anthropologists, the visual medium was perceived from the perspective of filmic and cinematographic developments. De Brigard traced visual developments back to the advent of modern photographic and sound technology, an approach that gives primacy to the technical development of cinematographic tools. This pioneering account was published in the first major—and still classic—collection of works on visual anthropology, Principles in Visual Anthropology (Hockings 1975). That volume foreshadowed the emergent visual field within anthropology. New editions of this sourcebook were published in 1995 and 2003. In a review of the 1995 edition, Schneider concurs that Hockings’s 1975 Principles “constituted the first attempt to bring together in a single volume the writings of major practitioners and theoreticians of ethnographic film and photography . . . [which] aimed to establish visual anthropology as a recognized subdiscipline of anthropology” (Schneider 1997, 704–5; emphasis added). Indeed, recognition is essential to move the field forward, although this is still a work in progress. An ideological divide grew in visual anthropology mirroring the one that developed in cultural anthropology at large, with a distinction between cultural studies and cultural anthropology. The former came to encompass narrative approaches and was promoted as being more humanistic, while the latter promotes field research and participant observation ethnography as the basis of data collection and analysis. Anderson (2003) raises several issues regarding the categorization of the genre of ethnographic film that emerge out of the orientation leaning toward film as art as contrasted with systematic ethnographic work in anthropology. According to this trend, ethnographic film was to be grouped with cinema, to be studied according to premises from a cultural studies perspective that draws inspiration from literary theory. As a practitioner of “observational cinema,” MacDougall (1995) favored filming topics about culture by running an unengaged camera. This was criticized by recognized ethnographic filmmakers (see, e.g., Heider 1976; Loizos 1993; Rollwagen 1988b; Rouch 1995a; Ruby 1975) who supported ethnographic film, in contrast with observational cinema, because it is engaged with the subject, was grounded in ethnography, and was informed by anthropological knowledge. Most ethnographic filmmakers were anthropologists or worked closely with anthropologists, so that mastery of culture and language were essential aspects of their projects.

14_107-Bernard.indb 440

6/9/14 10:03 AM

13: Visual Anthropology

441

Loizos (1993) was unambiguous. He called on anthropologists to “unlearn the idea that formal conceptual analysis rules the academy-and rules alone” (1993, 64). MacDougall (1978, 405) and Loizos (1993) separately called for the visual project to be considered a humanistic endeavor in cinema to show people as more “rounded” than cultural systematics (manifested in exchange patterns and kinship systems) shows them to be. Loizos wrote that in the 1980s, more anthropologists came to see their informants “as more rounded than they had been as producers of kinship systems, economic data, myths and cosmologies” (p. 80). Why are people more rounded when they recite a poem than when they exchange necklaces? My recent research in Qatar on kinship (El Guindi 2011a, 2012a, 2012b, 2012c) reveals the significance of studying kinship as kinship. Qatari students immediately perk up in class when the topic shifts to kinship (ranging from genealogy charting to discussion of descent structures). They behave at their most well rounded when talking about or engaged in “kinshipping.” Poetry, painting, novels, and songs have always been included, along with sex, food, child-rearing practices, and so on in the cultural repertoire of traditional ethnographies. In his analysis of Indonesian cinema, Heider (1991) showed how cinema, too, can be studied as a window on national culture, as a cultural element for analysis, not as an approach to visual anthropology as some trends suggest. Portraying kinship is a challenge for visual anthropologists but, in my view, one that needs to be met. Jay Ruby and Marcus Banks stressed the visible subject in visual anthropology (Banks and Ruby 2011; and see El Guindi [2001a, 2001b] for a perspective beyond the visible). El Guindi stressed going beyond the visual as merely the visible and the orientation of the visual medium as art (El Guindi 2001b). Peter Crawford (himself not a filmmaker) collaborated on books about both sides of the issue: one with David Turton called Film as Ethnography (Crawford and Turton 1992) and another with Simonis on narrative and aesthetics (Crawford and Simonsen 1992). Decades earlier, David MacDougall (1978, 405) dismissed ethnographic film totally, saying that it “cannot be said to constitute a genre, nor is ethnographic film-making a discipline with unified origins and an established methodology.” These are two separate points: first, whether to consider ethnographic film to be a genre; and second, whether ethnographic film has unified origins and established methodology. These points constitute much of what has been raised in debate and discussion. There are anthropologists, however, who do consider methods worthy of an entire monograph, even when their orientation leans toward art (Marcus 2001). Loizos (1993) suggested that ethnographic film should be looked at from the cinematic point of view, as a kind of documentary cinema, lest it become “narrowly concerned with ghetto culture called ‘ethnographic films’ ” (p. 1). Instead of becoming a ghetto culture of cinema, ethnographic film was and is thriving as a popular genre inside and outside anthropology, inside and outside the classroom, made so particularly by John Marshall through his films on the Kung San (Marshall 1962, 1966; 1969a, 1969b, 1971, 1974 [films]) and Timothy Asch through his films with Napoleon Chagnon on the Yanomami (Asch and Chagnon 1969, 1971, 1974a, 1974b [films]). The genre of documentary film and photography comes out of journalism and cinema, which makes it methodologically and epistemologically different from anthropology. Ethnographic film and photography, particularly visual ethnography (El Guindi

14_107-Bernard.indb 441

6/9/14 10:03 AM

442

Fadwa El Guindi

1986 [film], 1993, 2004, 185–249), are anthropology conceptually and ethnography empirically. A documentary film documents an event or a story constructed in terms of premises and principles coming out of journalism or cinema. Perhaps this is where claims to objectivity become pertinent. A journalist wants to provide an accurate story. An anthropologist discovers, explains, and produces knowledge. The two genres— documentary film and ethnographic film—are different. The next section, Genealogy of Visual Anthropology, discusses the role and significance of the anthropological character of visual anthropology. GENEALOGY OF VISUAL ANTHROPOLOGY

The term “visual anthropology” was coined after World War II and became associated with using cameras to make records about culture (Worth 1980, 7). After initial frustrations for those wanting to add visual tools to writing tools in anthropology, by the 1990s, photography and filming had become standard tools for ethnographic researchers. In Europe, the focus was almost exclusively on what came to be called ethnographic film; in the United States, various visual formats and media developed for teaching, recording, research, and analysis, which became part of visual anthropology. Instead of classifying the medium by physical format, still or moving photography, video or digital format, and so on, I differentiate among visual developments according to use and value for anthropology. This abstract classification becomes a framework against which we can classify or group visual anthropological projects organized as a genealogy of traditions. All forms share the quality of necessary culture knowledge, mastery of a people’s language, and systematic analysis. They are all suitable, even recommended, for classroom teaching (Asch 1975; Heider et al. 2006) and public lecturing. The grouping constitutes: (1) the visual medium as a recording tool of data for analysis and/or archival purpose; (2) the visual medium for elicitation and discovery; (3) the visual medium used for experimental culture reconstructions; and (4) the visual medium as ethnography (visual ethnography). Photography and film, and later digital formats, have all been used in these different ways. In a series of publications (El Guindi 1988, 1990b, 1993, 1996, 1998, 2004, 193–246), I introduced the term visual ethnography, referring to anthropological film grounded in prior systematic ethnographic research and demonstrating anthropological analysis. This notion of visual ethnography developed out of concern for the nature and extent of the anthropological process in film. It was not proposed as an alternative to ethnographic film but as a parallel to it. I highlight here the two phases of production and construction and explore three films to demonstrate the points that seemed missing in emphasis regarding film in anthropology.

Visual Ethnography The first film is Jean Rouch’s Les Maîtres Fous (Rouch 1954 [film]). Karl Heider saw an aspect of symbolic analysis in a segment of Jean Rouch’s editing of Les Maîtres Fous. According to Heider (1983), Rouch used montage to show the referent of a symbol: [W]hen the egg is smashed on the head of the possessed man playing the governor general, Rouch cuts, interrupting the possession ceremony of the Hauka, to a shot of the

14_107-Bernard.indb 442

6/9/14 10:03 AM

13: Visual Anthropology

443

real governor-general of the colony in full regalia to show us the white ostrich plume streaming down his hat and tells us, visually as well as verbally, that the egg is meant to symbolize the feather (this cross-cutting has a parallel in a written analysis of a myth). (Heider 1983, 5)

Heider notes how Rouch used flashbacks in the final sequence to juxtapose the men as they appear in their everyday contexts with their exalted forms in their possession states. In other words, his film was an analyzed ethnographic presentation and had value for its development of the visual medium as ethnography, not only as data or a record. The second film is Timothy Asch’s film The Ax Fight (Asch and Chagnon 1971 [film]). The Ax Fight represents a turning point in Asch’s career as filmmaker (for a more extended discussion, see El Guindi 2004, 102–3). At that time, the integrated teamwork between Asch and anthropologist Napoleon Chagnon was at its peak. Asch was previously influenced by the Marshall tradition of filmmaking known as sequence film, recording ethnographic events in their “natural” state. The motivation for the Marshall films was to provide anthropology teachers and students with visual tools to improve teaching and enhance learning. These films are still used in classrooms. Visual anthropologists had concerns about whether films can stand on their own in teaching or public viewing (i.e., without companion print publications). There was also a reaction against films being used in classrooms as “babysitting” devices without the presence of an anthropologist to discuss its context and ethnography. After teaming up with Chagnon to film the Yanomami, Asch developed his filmmaking beyond Sequence Film, referring to short event segments, to complete ethnographic events (for more on the difference, see El Guindi 1974, 97–103). Construction of The Ax Fight was innovative, becoming a landmark in visual anthropology. It has two parts: The first showed the chronological events leading to an ax fight among two groups in the Yanomamo village of study. The second part revealed elements hidden from the filmmaker, anthropologist, and viewer. After finding out what the fight was about, the second part revealed the anthropologist’s explanation of the conflict, while revealing other ethnographic aspects. Although such a film would not be easily grabbed by PBS television, it was of great value to anthropology. The third film is El Sebou’, which is one of three, El Sebou’ (El Guindi 1986 [film]), Ghurbal (El Guindi 1995 [film]), and El Moulid (El Guindi 1990a [film]) that can all be classified as visual ethnography. I discuss the making of El Sebou’ at length as to production, construction, and viewing (1988, 1990b, 1993, 1998, 2004, 193–246), in developing visual ethnography as a genre. In terms of incorporating analysis in the film itself, El Sebou’ is closer to Rouch’s Les Maitres Fous than to Asch’s The Ax Fight. The Ax Fight dealt with a spontaneous event and showed both event and explanation of the event. El Sebou’ was filmed after I first studied the Egyptian ritual of birth for two years. Based on data I gathered using ethnographic techniques and analysis of these data, production and construction of El Sebou’ proceeded and revealed results of these in the film itself. The film represents the ethnography of the ritual of El Sebou’. In the genealogy represented in Figure 13.1, visual ethnography is included as one of three genres classified. The anthropologically relevant visual productions include (1) ethnographic film; (2) research film; and (3) visual ethnography. Ethnographic

14_107-Bernard.indb 443

6/9/14 10:03 AM

444

Fadwa El Guindi

film is divided into filming others and filming selves. I classify filming others as a genre that includes sequence film (Asch 1975), Jean Rouch films (considered to be in a league of their own) (Rouch 1979, 1995a, 1995b), and feminist films (like films on the Maasai by Curlin and Llewellyn-Davies [Curling and Llewelyn-Davies 1974, 1975, 1984a, 1984b (films)] and on Greece by Susannah M. Hoffman [Hoffmann 1976 (film)]). The other major classification, filming selves, consists of culture reconstruction, as in Asen Balikci’s classic Netsilik films (Balikci and Mary-Rousseliere 1967) [films]), experimental, as among the Navajo (Adair and Worth 1967), and applied visual anthropology (Biella 1988, 1993). DOMINATION OF ETHNOGRAPHIC FILM

In anthropology, the two main visual media were employed early on in visual projects: film and photography. The most prominent systematic field research project producing a substantial, systematic record of visual data using both mediums was that of Margaret Mead and Gregory Bateson. The pioneering chapter by Emilie de Brigard is mostly about film, although it also discusses what became known as “research film,” which centered photography and other modes of nonverbal communication in the field. Other classical overviews of the field include works by Karl Heider (1976), Peter Loizos (1993), and edited volumes by Crawford and Turton (1992) and Rollwagen (1988a). Rollwagen’s chapter in his edited volume was “The Role of Anthropological Theory in Ethnographic Film-Making.” The most common name for film in anthropology became ethnographic film, and turning film (i.e., moving pictures)—as Davey (2008) showed, in a content analysis of the first twenty volumes of the journal Visual Anthropology—into the visual genre most talked about in anthropology. Internally, in visual anthropology, John Marshall and Timothy Asch distinguished what they called sequence film from ethnographic film. In the genealogy presented here, I classify visual projects on the basis of whether they are filming others or filming selves, filming for data or for illustration of ethnographic activity, filming for description or for analysis, in addition to the new digital multiple format media. More current developments can be flexibly incorporated in this genealogy without much modification to the overall structure. Despite all the debate that ensued against ethnographic film, it remains the most popular visual medium utilized in general teaching of anthropology, whether for illustrating examples of cultural traditions (Heider et al. 2006), analyzing primate behavior (Zeller 1992), or making methodological points about the ethnographic interview (El Guindi 1995 [film]). The introductory book by Heider et al. (2006) attests to the continuing usefulness of film as a teaching tool in cultural anthropology. Most books, much discussion, international conferences, and worldwide festivals focused on film rather than photography, such that ethnographic film dominated the field and became almost equivalent to visual anthropology, although this was more the case in Europe, including England, than in the United States. Technological developments have impacted the use of the two mediums of photography and film in anthropology. The significance of these advances is discussed next.

14_107-Bernard.indb 444

6/9/14 10:03 AM

13: Visual Anthropology

445

SOUND AND PICTURE

Synchronous sound for portable cameras was developed in the 1960s. Even after highly portable synch-sound cameras were developed, few ethnographers jumped at the opportunity to use this technology. Exceptions were Jean Rouch in France and John Marshall in the United States. Rouch quickly adopted technological advances to innovate in cinema technique expanding exploration of the human condition. Loizos identified four qualities of Rouch’s contributions: documentation, collaboration, interrogation, and improvisation and fantasy. For Rouch, Loizos said (1993): The “camera is not confined to the role of a passive recording instrument,” as in observational cinema, but becomes “rather an active agent of investigation and the camera user can become an interrogator of the world” (p. 46). Loizos, documentary film journalist by training, was bewildered by the reluctance of anthropologists to film when they had quickly adopted still cameras and tape recorders for field research. The remoteness of the places where much fieldwork was done might have been a factor in the reluctance to use the technology. In the earliest days of film, cameras were bulky and difficult to handle. Equipment was difficult to handle, cameras were fixed on tripods, and film was of low exposure suitable only for shooting in broad daylight or with use of artificial light, and synchronized sound technology had not yet been invented. Even in the early days of photography and film, some archaeologists, primatologists, and ethnologists used still and moving pictures technologies for gathering data, for cross-checking facts, and for building records (Blackman 1986, 1992; Caldarola 1987; Edwards 1992; Faris 1992; Scherer 1992). Cinematic experiments in ethnology may have been isolated and fragmented, but the limits of the technology did not stop some anthropologists from exploring the possibilities and going beyond the material limits to record and archive data. Starting in 1894, Franz Boas spent over 40 years studying the Kwakiutl of the Northwest Coast and used still photography in the field (Jacknis 1984, 1992). In 1930, at the age of 70, Boas went on what would be his last field trip to the Kwakiutl accompanied by Russian anthropologist Yulia Averkieva. Boas took along a motion picture camera and wax cylinder sound-recording machine. He shot 16-mm film of dances, games, manufacturing, songs, music, and other aspects of Kwakiutl life. The footage was meant to be for research and archiving, but was posthumously edited into film divided into thematic segments by Bill Holm and with the assistance of several Kwakiutl informants. The tworeel production from Boas’s footage was completed in 1973 (Boas 1973). Ruby (1983) suggests that Boas would not have had sufficient technical knowledge to realize that he could not synchronously record sound and image in the 1930s and that he probably assumed he could (1983, 27, 29). The first use of visual tools with ethnographic intent is often credited to Baldwin Spencer (Cohen 2001) when he took the Edison cylinder recorder and a Warwick camera to central Australia in 1901 (MacDougall 1995, 117). Other early users were Margaret Mead and Gregory Bateson (1930, 1952, 1954). But Luc de Heusch (de Heusch 2001) gives the ultimate pioneer’s award to Alfred Haddon, who, in 1898 took his camera on the Cambridge University Expedition to Torres Strait, to Franz Boas, who in 1930 used a small 16-mm camera on his final expedition to the Kwakiutl, and to Marcel Griaule who, a few years later, directed 35-mm shootings by

14_107-Bernard.indb 445

6/9/14 10:03 AM

446

Fadwa El Guindi

professional filmmakers of Au Pays des Dogon and Sous les Masques Noirs (1938a, 1957) and later, to Gregory Bateson and Margaret Mead used photography and film recording during their field studies in Bali and New Guinea. (de Heusch 2001, 3)

On the Torres Strait expedition, El Guindi (1998, citing Banks 1996) writes that Haddon’s team: was equipped with a variety of recording tools, from W. H. R. Rivers’s genealogical method, to photography, to wax-cylinder sound recording, to the Lumière motion camera. Haddon collected over 7,000 feet of film (chiefly of ceremonies) and made a number of wax cylinders. (El Guindi 1998, 462–63)

For more ancestors, see Jacknis (1990 on Mooney). Rouch (1979, 1995a) describes in dramatic language these early days with the camera: One almost needed to be crazy to try using a tool as forbidding as the camera . . . the first clumsy attempts . . . in Marcel Griaule’s Au pays des Dogons (Griaule 1935) and Sous le masque noir (Griaule 1938b) we can understand their discouragement with the result of their efforts. Their admirable documentation was put through the filmmaking machine. (Rouch 1995a, 84; emphasis added)

But Rouch was reacting to more than bulky filming equipment. He was frustrated about the absence of academic funding for such filming, making such projects vulnerable to accepting funds from business sources. Rouch says that it is “absurd to try to mix research and business” (1995a, 84), and blames lack of research funding for “wild, insensitive editing, oriental music, commentary in the style of a sportscast” (1995a, 84)—a travesty that Mead and Bateson escaped as they had funding from academic sources, such as American universities. Rouch considers this to be a factor in the success of the series produced by Mead and Bateson (1930 [films]). Synch sound added an important dimension to the quality of film and experience of filmmaking in anthropology. Colette Piault (1994) vividly describes the viewing experience when first encountering the effect of synch-sound in a film: [I]n 1972, we saw at a festival in Venice . . . To Live with Herds (MacDougall 1972) shot by David and Judith MacDougall of the Jie in East Africa. For me, and probably for many other film-makers, it was a shock and a revelation: the sync-sound had appeared and for the first time the dialogues had been recorded and translated . . . we discovered . . . what people . . . were talking about. It was one of the first films—so far as I know, the first one shown in Europe—to use subtitles (in this case, English) to render indigenous speech. . . . This film marked an important step and influenced many film-makers. Sync-sound brought the possibility of listening to the voice of the other directly . . . gave the feeling of closeness and intimacy which was so missing in the preceding films. (Piault 1994, 5–6)

In the 1960s, 16-mm color film also became widely accessible and faster film enabled filming in poor light (inside houses, huts, evening rituals, etc.). These changes made possible more intimacy, more flexibility, more spontaneity, and more shooting-style innovations. Rouch demonstrated the profound impact of the changes in

14_107-Bernard.indb 446

6/9/14 10:03 AM

13: Visual Anthropology

447

both his participatory camera and catalytic shooting style. Emily de Brigard (1975) considers Jean Rouch one of a few who not only benefited from, but in the 1960s led to change in technology combining the use of synchronous-sound filming with shooting by hand-held camera. The invention of video pushed ethnographic film further forward, opening new visual opportunities and wider use for ethnographic film. Today, multimedia technology combines simultaneous and interactive media and formats, which can allow for maximum flexibility in application and for methodological and analytic rigor. There are more digitizing instruments and new software, such as HyperResearch and ELAN (from the Max Planck Institute). ELAN is available on http://tla.mpi.nl/ tools/tla-tools/elan/. These tools support the tagging and analyzing of visual data in the same way that textual data is analyzed across the social sciences (see Wutich et al., this volume). As Bishop and Bishop (2012) put it: This is an exciting time in visual anthropology. The proliferation of inexpensive imaging devices, editing software, and distribution avenues has opened purely visual discourse to an enormous number of people. Digital images and videos are electronically exchanged; people who once were the subjects of anthropological films make them; videos comment on other videos. This review section is part of a conversation, an opportunity to sample and reflect on the range of subjects, approaches, and concerns of the vast stream of anthropology in visual media. This includes not only films but also photography, museum displays, the Internet, and experimental work with visual imagery in the wider anthropological community. (p. 146)

The visual medium expands the empirical base for anthropological analysis. It permits a combination of multiple data forms combined in interactive ways within one format. It allows the linking of elements and data forms for a more adequate description, leading to new relations and connections via comparison and analogy, a process fundamental to analysis. New insights are revealed. Visual tools have now become integral to the teaching of anthropology in academic institutions around the world, in documentation, and in data gathering and analysis by anthropologists and others in the social sciences (El Guindi 2004). It is also used in applied work and to achieve advocacy goals (Chalfen and Michael 2007; Pink 2007). Sarah Pink calls for an applied visual anthropology (Pink 2007, 2011) as an alternative to ethnographic film. “Early advocates an applied visual anthropology,” she writes, “have criticized the sub-discipline’s emphasis on ethnographic film-making” (Pink 2007, 4). Clearly, there is no subdiscipline of applied visual anthropology, nor does it constitute an alternative to ethnographic film since the two are of different ontological status, one representing a genre of film and the other the applicability of anthropology to people’s lives. “Applied,” “practicing,” “engaged,” and “public” are among the terms used to refer to the relevance of anthropology to people’s lives. Good anthropology, visual or print, will inevitably be of applied use. While visual tools gradually entered the ordinary anthropological project, this integration has not automatically led to the mainstreaming of the field. It was not until the 1980s that the process toward mainstreaming began—that is, taking first steps using

14_107-Bernard.indb 447

6/9/14 10:03 AM

448

Fadwa El Guindi

formal channels within the larger discipline of anthropology to integrate visual methods, thus leading to visual anthropology achieving some recognition by the discipline. The next section presents a brief summary of developments leading to the present. MAINSTREAMING AND PROFESSIONALIZING THE VISUAL

In 1964, a conference was held on filmmaking in anthropology. Other than that, there were virtually no programs, publications, or regular meetings on ethnographic film or visual anthropology. In subsequent years, the Program in Ethnographic Film (PIEF) was established, and by 1973, it had become the Society for the Anthropology of Visual Communication (SAVICOM). This society functioned under the wing of the American Anthropological Association. Jay Ruby, along with a number of anthropologists and researchers set up the National Anthropological Film Center (directed by E. Richard Sorenson) at the Smithsonian Institution. The PIEF Newsletter, begun by Jay Ruby and Carroll Williams in 1969, continued as the Society for the Anthropology of Visual Communication (SAVICOM) Newsletter. This was later incorporated into the Anthropology Newsletter of the AAA. In 1974, SAVICOM began a more formal publication series, edited by Sol Worth. Ethnographic films have been reviewed in the American Anthropologist since 1965. Since 1966, ethnographic film sessions have been a regular feature of the AAA annual meetings. But this was insufficient to convince most anthropologists of the importance of visual methods in their projects. From 1990 to 1994, as film review editor, I began the process of broadening the review section, arguing that visual anthropology rather than ethnographic film more accurately described the intellectual, theoretical, and methodological coverage of visual activities. The section was renamed with the succession of Harald Prins as editor for visual anthropology reviews in 1994. Today, editorship of visual anthropology reviews is managed by a filmmaker–anthropologist team, John Bishop and Naomi Bishop. In 1969, Jay Ruby began the annual Conference on Visual Anthropology at Temple University in Philadelphia. In 1984, the Society for Visual Anthropology (SVA) was formed as a constituent section of the American Anthropological Association. This was crucial in mainstreaming visual anthropology as a field. The SVA was instrumental in shifting activity from ethnographic film to a more balanced coverage of research areas. SVA’s official published statement is inclusive of diverse media forms and different spheres of culture, including the material covering a wider range of methods and approaches to the visual medium. The SVA produced a regular newsletter—now Visual Anthropology Review—out of which a select group of articles were published (Taylor 1994). On the tenth anniversary of the annual SVA Film/Video Festival (El Guindi and Williams 1995, xv–xvi), the SVA published a list of all the films and videos given awards during the preceding years, along with a detailed scholarly commendation (Blakely and Williams 1995, vii–viii; Williams et al. 1995). That year, the AAA published another listing of films (Heider and Hermer 1995). In the first-and-still-classic collection of articles on visual anthropology, Hockings (1975) lamented that “of the various English handbooks now available on research methodology, only one devotes as much space as two pages (out of a thousand) to some applications of cinematography in anthropology.” Hockings was referring to his two-

14_107-Bernard.indb 448

6/9/14 10:03 AM

13: Visual Anthropology

449

page overview in the first Handbook of Method in Cultural Anthropology (Naroll and Cohen 1970). The second handbook (Bernard 1998) contained a full chapter on visual methods (El Guindi 1998) and since then, more anthropologists have been adopting visual methods in their regular methods classes. There were developments outside the United States, too. In the Netherlands, filming before World War II was incorporated as part of the expeditions to the colonies and was sponsored by the Colonial Institute of Amsterdam. J. C. Lamster was the first to use film in Java and Bali, recording work on agricultural plantations and on various traditions and customs. According to Nijland (2002), the purpose was to educate the Dutch population about the colonies. The Colonial Institute’s film collection was finally housed in the Nederlands Filmmuseum. In Germany, the Institut für den Wissenschaftlichen Film was reorganized after World War II. German anthropologists resumed filming in Melanesia, Africa, and Europe, emphasizing scientific purity (Husmann 1983; Koloss 1983). The institute’s program produced “Rules for Film Documentation in Ethnology and Folklore” in 1959. According to the rules, filmmaking must be done by persons with sound anthropological training or supervision. Further, an exact log must be kept, recording authentic events. Finally, the filming must be done without using dramatic camera angles or movement and must be edited for representativeness. In 1952, the first systematic anthropological film archive, a scientific encyclopedia in film form, was established at Gottingen, Germany (Wolf 1972b [film]) Konrad Lorenz and others assembled and arranged several thousand films on anthropological and biological subjects. Each film consisted of a single thematic unit, such as dance, work, or ritual, grouped according to different cultures (Taureg 1983; Wolf 1972a) and arranged to facilitate comparisons of behavior across cultures (Wolf 1972b). The majority of the films are silent. Most also have a printed document with background and technical information. The document is in the language of its author (German, French, or English—the official languages of the encyclopedia). Work on an ethnohistorical atlas of the Soviet Union began in the 1960s. By the mid-1970s, the classical methods of collecting ethnographic materials had been complemented with ethnographic filming. A number of Soviet research institutions have used ethnographic filming. Filming began to be used for scientific purposes at the State Ethnographical Museum of the Estonian Soviet Socialist Republic in 1961(Peterson 1975). In 1970, a Japanese archive was established at Tokyo. In 1972, Japanese anthropologists, TV journalists, and artists formed the Japanese Committee of Film on Man, in collaboration with Jean Rouch. Its aim was to support the production of ethnographic films in collaboration with the National Institute of Ethnological Studies (Hockings 1988; Hockings and Omori 1988; Ushijima 1988; Ushiyama 1975). In 1974, the National Anthropological Film Center was started at the Smithsonian (see Sorenson 1995). This led to the Human Studies Film Archives, which houses a large collection of film and visual material (Wintle and Homiak 1995). Egypt hosted its first academic conference on visual anthropology in 2010 in Cairo, sponsored by the American University in Cairo. Many of the participants were practicing artists. The focus was on the production of knowledge. Papers from

14_107-Bernard.indb 449

6/9/14 10:03 AM

450

Fadwa El Guindi

that conference are in Sabae and Westmoreland (2012), including a paper in which I stressed the analytic value of stills for knowledge production. DOMINATION BY ETHNOGRAPHIC FILM

In non-Western countries, there was great interest in the visual, and many developed strong cinematic traditions. Egypt (Abdullah 1984; al-Bindari 1981; al-Hadari 1989) and India (Sahay 1983) had a powerful early cinematic tradition of realistic fiction and social documentary, but no tradition of ethnographic films developed. Ethnographic films were viewed as being products of colonial influences in which colonial filmmakers were filming the colonized—the West orientalizing the East (Amin 1989; Said 1978)—and anthropology’s colonial roots were being revisited through visual anthropology. Skepticism about visual anthropology was similar to that expressed about anthropology in general. There was distrust of Western projects grounded in what was perceived as colonialist agendas. Adolfo Colombres situates the ethnographic film genre, like anthropology itself, within colonial encounters and dominating relationships. To him, Malinowski and Flaherty were romantics who, in the 1920s, escaped civilization by going to remote lands and introduced methods of recording, thus sidestepping the political context of a colonial situation. Colombres points out that Robert Flaherty admitted that he wasn’t interested in the demise of the people he was filming—one brought about by white domination. Rather, Flaherty’s goal was to demonstrate “their originality and majesty before whites annihilated not only their identity, but the people themselves” (Colombres 1985a, 12, 1985b). In other words, Flaherty’s sentimental nostalgic view of culture would freeze a reconstructed pre-Contact “noble savagery.” Marks’s view of the 1895 footage by Felix Regnault supports the claim of the colonial orientation of anthropological film. First, the film depicted tribal peoples, so fixing its subject matters. Second, the last of four sequences, in which the Madagascans carry the photographer on the palanquin, evokes the image of servile native bearers carrying the dominant European photographer (Marks 1995, 339). In 1946, Jean Rouch, inspired by Robert Flaherty, Dziga Vertov, and Jean Vigo and a secondhand camera, set out to film reality. Colombres observes that when Rouch filmed that year in Niger, he was a member of the French colonial society, launching a spiritual adventure inside colonially dominated French West Africa. In general, Rouch avoided politics and the political context. In Moi, Un Noir (Rouch 1957 [film]), about immigrants from Niger to the Ivory Coast, Rouch gives voice, for the first time, to the colonized so that they could express their view of the world. But Colombres asks (1985b, 17): To whom does Rouch designate the responsibility of the African consciousness? Was it to an immigrant Nigerian from British West Africa who danced well, as was expected of an African black? The film, Colombres stated, confirms the racist stereotypes of the colonizer and is a cinema of the exotic, an important component of colonialism. Others echo this same perspective. In India, Singh (1992) criticizes anthropological studies of India, saying that they overstress divisions and tribalism, particularly during the colonial period, when ethnography was concerned primarily with tribes. This legacy continued. Colonial ethnography, Singh writes (p. 9), “created the categories of

14_107-Bernard.indb 450

6/9/14 10:03 AM

13: Visual Anthropology

451

caste and tribe—simplifying a very complex structure (of) India as a feeling, as a vision, as a dream shared by all of us.” Singh proposes research and filming that would show an understanding of the nature of Indian pluralism as a: melting pot, a mosaic, a fishing net into which have been drawn peoples and races . . . (reflecting) the unity of the people, shaped by geography and environment, by history and culture, that developed as communities and regions have interacted over time. This process is called unity in diversity, some call it diversity in unity. (pp. 8–10)

Singh’s alternative approach to the study of India consists of a comprehensive study in which community, as identified and defined by the people themselves, is the unit of study. The project would include visual documentation by films (Singh 1992, 11, 12). Roy and Jhala (1992, 28) see a role for visual anthropology in India in that it “could initiate discourse across the illiteracy barrier and provide a platform from which the cultures of India could gain both ‘voice’ and ‘representation.’ ” They regard “technological feasibility, political desirability, international example, and the promise of international cooperation as incentives for undertaking visual anthropology in India” (p. 21). Singh (1992, 14) supports development of a visual anthropology that respects premises of cultural integrity and finds great value in scientific visual documentation for influencing consciousness and intervention strategies during crises in a way that fiction film can’t. While a strong critical perspective about film in general developed in Latin America, Egypt, India, and elsewhere, no methodological framework for ethnographic film has been formulated that deals with the premises underlying the critique. Filmmakers in those areas concentrated on fiction and, in a somewhat limited way, adopted the documentary form in three genres: journalistic, propaganda, and folkloric (the latter presented as ethnographic). Reverse visual anthropology is, perhaps, one attempt to subvert domination. Three decades after Jean Rouch began filming in Africa, Manthia Diawara, head of the African Studies Program at New York University and commentator on West African film, experimented with the idea of “visual anthropology in reverse.” Diawara filmed Rouch in Paris to see if “shared anthropology,” a phrase coined by Rouch, could shed light on cross-cultural relations between the powerful and the disempowered (see Diawara 1989). In a review of Diawara’s work (1995 [film]), Michael Fischer (1997) observes that Rouch had proposed a journey through the sculptures of Paris as his way of presenting the “public Rouch.” Diawara came to realize that “he was taking me on a tour of my own French education and showing me how much of it I still carry.” Diawara recalls that he was made to recite in front of one sculpture the childhood verse of Monsieur Fox and Monsieur Crow, the fable by La Fontaine. Fischer asks (1997): Is this an unsuccessful effort by the filmmaker, a political blockage in Rouch, or an example of the wily Rouch’s upstaging, inverting the power relation between filmmaker and film subject, the very thing that Diawara has been warned about and warns us about both early and late in the film? (p. 142)

A twist on filming the colonized is illustrated by the case in which John Collier worked with Bernard Siegel on an ethnography of Picuris Pueblo, a dwindling pueblo

14_107-Bernard.indb 451

6/9/14 10:03 AM

452

Fadwa El Guindi

of the Tewa language group. Collier’s contribution was to make photographs for Siegel to use in photo interviewing. This makes it a valuable project for both keeping a record of a people and the use of a method: photo interviewing. Both presumed that ceremonial dances constituted the heart of Pueblo religious life. San Lorenzo Day, a summer fiesta, took place during the research, and Collier photographed the theatrically exquisite Deer Dance. During photo interviewing, however, Collier and Siegel learned that the dance held a relatively low place in the day’s ceremony. As one collaborator said: “We do this to please the white people” (Collier 1988, 90). This discovery shows the value of photo interviewing for eliciting data, which can also tell us something about the effect of hegemony on a people who perform their culture. The visual method assisted in revealing this observation. A more recent example comes from Dianne Stadhams in Pink (2007, 119–42), which describes her project on tourism in Gambia of West Africa. Below a picture of children, a caption states: “When I grow up, I want to be a tourist” (Stadhams 2007, 119). While some Gambians gain economic benefits from tourism, it seems that watching the growing tourism in their country causes children to dream of careers as tourists. A key issue identified by Stadhams (2007, 132) for her project of building a television program is “what tourism means to and for Gambians.” This reported comment by Gambian children relates to the overall effect of consumer-based industries such as tourism on the construction of dreams of Gambians for their country’s future. In my review of the volume I ask: Has power from participation resulted in improved lives of ordinary people? Do we have instruments to measure quality of life? Some Gambians are happy that tourism is bringing work and spreading wealth, but what about young Gambians whose vision of the future is being distorted by dreaming of careers as lazy tourists? This is where doing good anthropology can benefit knowledge and future lives (El Guindi 2011b). FOR GOD’S SAKE, MARGARET

Gregory Bateson and Margaret Mead worked as a team and used film and photography in the field, integrating both media in their systematic, comparative ethnography. Mead made their project’s priority explicit (Jacknis 1988), saying that they weren’t filming “for the purpose of making documentary films and photographs for which one decides a priori upon the norms and then gets the Balinese to go through these behaviors in suitable lighting” (Bateson and Mead 1942, 49; emphasis added). The reference here is to the relative staging of behaviors and events to accommodate the limitations of technology (light conditions). In a seminal conversation between Gregory Bateson and Margaret Mead, subsequently published (Brand 1976), important issues regarding the use of the visual as a medium in anthropology were argued in the liveliest manner. Note that the visual project of Mead and Bateson was accomplished by the two as a team—independently credentialed, prominent, prolific, and accomplished anthropologists who used scientific rigor in their methods and offered profound insights in their analyses (Bateson 1943, 1963, 1972, 1979; Bateson and Mead 1935, 1942; McQuown and Bateson 1971). A segment of the conversation was excerpted and reprinted in the journal Studies in the Anthropology of Visual Communication with the title “Margaret Mead and Gregory

14_107-Bernard.indb 452

6/9/14 10:03 AM

13: Visual Anthropology

453

Bateson on the Use of the Camera in Anthropology” (Mead and Bateson 1977). This reassigned title is techno-centric and reduces what I consider to be a rich multifaceted conversation to a technical level—not at all what the profound conversation is about, a conversation I recommend as required reading in any visual anthropology class. In one segment of the conversation and elsewhere (Mead 1963), Mead observes how one can forget that other things are happening when the camera is pointing in one direction, or that other things are happening outside the frame (for an analysis of the conversation, see El Guindi 2004, 61–73). The focus on some behavior of momentary interest might sacrifice other, potentially important data. The issue concerned the disadvantages of leaving the camera running from one position versus moving it around to capture related action (on this, see Heider 1983). This has given rise to spurious arguments about objectivity, selectivity, and representationality. Sorenson (1995; also see Sorenson and Jablonko 1975) notes that: A peculiar myth that has developed in recent years is that anthropological films cannot be scientific because their content is always governed by selective interests. This absurd notion ignores the degree to which selectivity and special interest underlie all scientific inquiry. Method is crucial. In order for visual records to be a valid scientific resource, they need to be shaped by the scientific methodological considerations that govern the investigation of nonrecurring phenomena. (p. 496)

Rather, the point to be made from Margaret Mead’s remark is about the importance she gave to visual research data and their use for discovery. Heider (1983) discusses research footage (not intended for inclusion in a finished film) used by an ethnographer to capture an image of behavior for careful frame-by-frame analysis. He notes how research with film and videotape allowed researchers like Bateson, Birdwhistell, Lomax, and Kendon to demonstrate how important information is continually being expressed and communicated in whole bodies and in both sides of the conversation and how this kind of research has had a direct impact on thinking about ethnographic film (see Heider 1983, 2–10). Adam Kendon shot his footage of greeting behavior in wide angle to pick up unexpectedly early stages of a greeting sequence (Kendon and Ferber 1973). Significantly, Kendon (1980, 1990, 1991, 2004) has taken his research on gestures beyond anthropology, stimulating debate among linguists and cognitive scientists. THE ANTHROPOLOGY IN THE VISUAL

It became a challenge to reveal to anthropologists at large the anthropology in the visual. In fact, as evident in this overview and in others, there is a unified origin and a sustained interest in establishing sound methods for creating visual data. Marcel Griaule sustained Regnault’s concept of ethnographic filming as a scientific activity concerned with traditional ethnographic subjects. He distinguished three film types: archive footage for research, training films for anthropology courses, and public education films, occasionally including works of art (Griaule 1957). Andre Leroi-Gourhan (1948) applied the adjective ethnological to film and introduced another tripartite classification: the research film, the “exotic” travel film (to be abhorred as superficial and exploitative), and the “film of environment,” produced with

14_107-Bernard.indb 453

6/9/14 10:03 AM

454

Fadwa El Guindi

no scientific aim but deriving an ethnological value from its exportation. Chanock and Sorenson (1975, 432) refer to the research film method, which provides identified and annotated visual records, unedited, not designed to impose preconceived ideas, and focused on films based on ethnographic understandings of a culture. Loizos (1993) offered a grouping of films from specific periods as innovations and modalities. The innovations he identified are: (1) production technology; (2) diverse subject matter; (3) widened range of strategies of argument used by filmmakers—that is, films that combine several modes of representation; and (4) enhanced ethnographic contextualizing devices. Loizos proposes the following modalities: documentation, explanatory, explanation rejected (film modality that rejects conceptual explanation root and branch), and context enrichment. MacDougall (1978, 405) found one distinction useful: ethnographic footage (raw material that comes out of a camera, like field notes, used for a variety of purposes including the making of films) and ethnographic films (structured works made for presentation to an audience). He further divided ethnographic footage into two major forms: research footage to serve specific scientific inquiries and record footage made to provide more general documents for archiving and future research footage for research purposes (see Omori 1988). To Heider (1983, 5), “ethnographic films must themselves be ethnographically accountable” and “the better the ethnographicness, the better the cinema.” Yasuhiro Omori (1988) focused his discussion on footage film and ethnographic film. He draws the analogy from print ethnography and sees a correspondence between footage film and field notes, on the one hand, and ethnographic film and monograph, on the other. He states that within visual anthropology, films take two main forms. First, there are simple footage films taken as field notes. Second, there is the ethnographic film (like a monograph) that is shot in a comprehensive way and organized around a theme related to the entire culture being studied (Omori 1988, 192). Footage films are short films recording a technical process or scenes of human behavior within a group. The monograph film tends to be longer and has a story constructed on a specific theme. Omori characterized the difference as part/whole, analytic/interpretive, scientific/ethnographic, or differential length (pp. 192, 194, 196). Gotthard Wolf considered three kinds of scientific film: the research film, the scientific documentation film, and the university instructional film (1961, 16–20). Increasingly, there are visual studies that focus on popular media, home photography, and public culture (Chalfen 1975, 1987, 1992; El Guindi 1996; Ginsburg 1995; Pink 2007; Ruby 1981). These progressive developments placed filming squarely within anthropology rather than documentary cinema—a position expressed by many anthropologists to this day. Worth (1966, 1968, 1969, 1972, 1980, 1981a, 1981b) referred to ethnographic film as a set of signs to study the behavior of a people, used either as a recording of data about culture or as data of culture. Ruby (1975) proposed four criteria for ethnographic films. They should be: (1) about whole cultures, or definable portions of cultures; (2) informed by explicit or implicit theories of culture; (3) explicit about the research and filming methods they had used; and (4) use a distinctively anthropological lexicon. In

14_107-Bernard.indb 454

6/9/14 10:03 AM

13: Visual Anthropology

455

a now-classic book, Balikci (1989) discussed degrees of ethnographicness on the basis of six premises used “to define what is ethnographic in a film” (pp. 33–34). Rollwagen (1988a, 1988b) stressed the disciplinary framework within which ethnographic film is situated, ethnography being a scientific description conducted within a theoretical framework. This puts film/photography within anthropology, not cinema, and entails the premise that the visual medium is integral to research or, in the case of ethnographic film, is based on a significant amount of research prior to filming (Rollwagen 1988b, 983). Clearly, ethnographic film isn’t merely involvement in other cultures. When Heider (1976) wrote his now iconic statement about ethnographic film, that satisfactory ethnographic films are those revealing “whole bodies, and whole people, in whole acts,” he sketched a system for discussing different attributes that contribute to the ethnographicness of a film, making clear to anthropology that ethnographic film is integral to the scientific anthropological project (p. 75; emphasis added). In contrast, Loizos (1993) situated film in cinema, which draws inspiration from literary theory (and which challenges objectivity and facts) rather than in the social science tradition of anthropology. Margaret Mead was certain about the place of the visual. To Mead (1962), visual tools were used for research data and discovery: Film materials . . . have made it possible to explore ways to tap the theoretical insights of other disciplines through the use of visual materials and of providing a continuing resource for the exploration of new hypotheses as the behavior, recorded on film, can be viewed repeatedly in the light of other new materials. (p. 138)

The questions of concern regarding the visual in anthropology are: Do different visual forms generate primary data or do they illustrate ethnographic observations gathered by regular ethnographic observation? The goal, as many visual anthropologists agree, is to contextualize the visual in good anthropology and visual data in sound ethnography. So, while the importance of the visual medium is gradually being recognized in mainstream anthropology in terms of method and theory, the significant issue remains the quality of knowledge production. It goes without saying that anthropologists are (or should be) concerned with the anthropological quality of knowledge production irrespective of the medium used, print or visual. The issue is ethnographic adequacy. CONCLUSION

From Felix Regnault (1931), to Alfred Haddon (1895), to John Collier, Jr. (1967, 1988) and Margaret Mead (1975), among many others, the quest was for anthropologists to take visual tools seriously. All of these scholars called for using visual tools in research, a kind of contribution they themselves demonstrated in projects of record and discovery of knowledge. They demonstrated the importance of the visual as both research data and a tool for research in anthropology. This appeal was famously made by Mead (1995), who criticized ethnographic inquiry that came to “depend on words, and words, and words” (p. 5), and who admonished anthropologists for their passivity and resistance to using pictures in field research.

14_107-Bernard.indb 455

6/9/14 10:03 AM

456

Fadwa El Guindi

This critique and appeal is as relevant today as ever in anthropology around the world. The visual as a methodological scholarly tool is close to being absent in the very societies, like Egypt, that have successfully produced fiction for over a century. (For a discussion of the state of Arab documentary film, see Abdullah 1984.) To enhance the future of visual anthropology, research must continue to build on its rich methodological heritage. Today’s anthropology students must be able to imagine what Margaret Mead would have done with today’s new possibilities of digitizing instruments and new software. But imagining what Mead and Bateson would have done with such tools requires us not only to learn how to use them but to master anthropology and do good ethnography of the kind done by these pioneers even before modern visual tools were available. As Bateson (1979) said: [W]e have been trained to think of patterns . . . as fixed affairs. It is easier and lazier that way but, of course, all nonsense. In truth, the right way to begin to think about the pattern which connects is to think of it as primarily . . . a dance of interacting parts. (p. 13)

Scholars and practitioners of art and literature may be suspicious of a scientific approach, to the systematics in the accumulation of knowledge, but the key to meaningful knowledge, as Bateson said, is “the dance of interacting parts.” Visual production of knowledge might just be the methodological tool leading to this key. FILMS

Asch, T., and N. A. Chagnon. 1969. The feast. DER. 16mm; color. Asch, T., and N. A. Chagnon. 1971. The ax fight. DER. 13'; color. Asch, T., and N. A. Chagnon. 1974a. A father washes his children. DER. 13'; color. Asch, T., and N. A. Chagnon. 1974b. A man called “Bee”: Studying the Yanomamo. DER. 16mm; color. Balikci, A., and G. Mary-Rousseliere. 1967. The Netsilik Eskimo Series: National Film Board of Canada. Seven films. Boas, F. 1973. (filmed 1930). The Kwakiutl of British Columbia. (Edited by B. Holm.) University of Washington Burke Museum. Silent, b&w. Curling, C., and M. Llewelyn-Davies. 1974. Masai women. Granada Television (Disappearing World Series). Distributed by Shanachie Entertainment Corp. Color. Curling, C., and M. Llewelyn-Davies. 1975. Masai manhood. Granada Television (Disappearing World Series). Distributed by Shanachie Entertainment Corp. Color. Curling, C., and M. Llewelyn-Davies. 1984a. Diary of a Maasai village. BBC-TV. Granada Television International. Distributed by Shanachie Entertainment Corp. Color. Curling, C., and M. Llewelyn-Davies. 1984b. The Women’s olamal: The social organisation of a Maasai fertility ceremony. BBC-TV. Granada Television International. Distributed by Shanachie Entertainment Corp. Color. Diawara, M. 1995. Rouch in reverse. Parminder Vir. California Newsreel. 51'; color. El Guindi, F. 1986. El Sebou’: Egyptian birth ritual. DER. 27'; color. El Guindi, F. 1990a. El Moulid: Egyptian religious festival. DER. 38'; color. El Guindi, F. 1995a. Ghurbal. DER. 30'; color. Griaule, M. 1935. Au pays des dogons. Comité du Film Ethnographique (Paris). 15"; b&w. Griaule, M. 1938b. Sous le masque noir. NSFL. 50'; color. Hoffmann, S. M. 1976. Kypseli: Women and men apart—A divided reality. Berkeley, University of California Extension Media Center. Color.

14_107-Bernard.indb 456

6/9/14 10:03 AM

13: Visual Anthropology

457

MacDougall, D. 1972. To live with herds. Berkeley, Extension Media Center. 16mm, b&w. Marshall, J. 1962. A joking relationship. DER. 13'; b&w. Marshall, J. 1966. N/um Tchai: The ceremonial dance of the !Kung Bushmen. DER. 25'; color. Marshall, J. 1969a. An argument about a marriage. DER. 18'; color. Marshall, J. 1969b. A curing ceremony. DER. 8'; b&w. Marshall, J. 1971. Bitter melons. DER. 30'; color. Marshall, J. 1974. The meat fight. DER. 14'; color. Mead, M., and G. Bateson. 1930. Character formation in different cultures series. NYU. Six films; b&w. Mead, M., and G. Bateson. 1952. Childhood rivalry in Bali and New Guinea. NYU. B&w. Mead, M. and G. Bateson. 1954 (filmed 1930s). Bathing babies in three cultures. PSU. 9'; b&w. Rouch, J. 1954. Les maîtres fous. Films de la Pleiade. 35'; color. Rouch, J. 1957. Moi, un noir: Films de la Pleiade. 80'; color. Wolf, G. 1972b. Encyclopaedia Cinematographica 1952–1972. Gottingen: Institut für den Wissenschaftlichen Film (IWF). REFERENCES

Abdullah, M. 1984. Tagarub fi al-cinema al-tasgiliyya (Experiments in documentary cinema) (Arabic). Cairo: Al-Hay’a Al-Misriyya Al-’Amma Lil Kitab. Adair, J., and S. Worth. 1967. The Navajo as filmmaker: A brief report of research in the crosscultural aspects of film communication. American Anthropologist 69: 76–78. al-Bindari, M. 1981. Al-cinema al-tasgiliyya fi misr hatta akhir sanat 1980 (Arabic) (Documentary cinema in Egypt through 1980). Cairo: Dar al-Sha’b Publishers. al-Hadari, A. 1989. Tarikh al-sinema fi Masr (Arabic) (History of cinema in Egypt). Cairo: N.a. Amin, S. 1989. L’eurocentrisme, trans. R. Moore. New York: Monthly Review Press. Anderson, K. T. 2003. Toward an anarchy of imagery: Questioning the categorization of films as “ethnographic.” Journal of Film and Video 55: 73–87. Asch, T. 1975. Using film in teaching anthropology: One pedagogical approach. In Principles of visual anthropology, ed. P. Hockings, 385–420.The Hague: Mouton. Balikci, A. 1988. Anthropologists and ethnographic filmmaking. In Anthropological filmmaking: Anthropological perspectives on the production of film and video for general public audiences, ed. J. Rollwagen, 31–46. New York: Harwood Academic Publishers. Balikci, A. 1989. The Netsilik Eskimo. Prospect Heights, IL: Waveland Press. Balikci, A. 1990. Bambi e il caribu. In Immagine e culture, ed. R. Cedrini, 36–44. Palermo, Italy: Flaccovio Editore. Banks, M., and J. Ruby, eds. 2011. Made to be seen: Perspectives on the history of visual anthropology. Chicago: University of Chicago Press. Bateson, G. 1943. Cultural and thematic analysis of fictional films. Transactions of the New York Academy of Sciences 5: 72–78. Bateson, G. 1963. Exchange of information about patterns of human behavior. In Information storage and neural control, ed. W. Fields and W. Abbott, 173–86. Springfield, IL: Charles C. Thomas. Bateson, G. 1972. Steps to an ecology of mind: Collected essays in anthropology, psychiatry, evolution, and epistemology. San Francisco: Chandler Publishing Company. Bateson, G. 1979. Mind and nature: A necessary unity. New York: E. P. Dutton. Bateson, G., and M. Mead. 1935. Fieldnotes and relevant papers. Washington, DC: Library of Congress, Library Holding, Manuscript Division. Bateson, G., and M. Mead. 1942. Balinese character: A photographic analysis, Vol. 2. New York: Special Publications of the New York Academy of Sciences.

14_107-Bernard.indb 457

6/9/14 10:03 AM

458

Fadwa El Guindi

Bernard, H. R., ed. 1998. Handbook of methods in cultural anthropology. Walnut Creek, CA: AltaMira. Biella, P. 1988. Filming AIDS in the barrio: A study in the pedagogy of the oppressed. Presented at the 87th Annual Meeting of the American Anthropological Association, Phoenix, Arizona. Biella, P. 1993. Beyond ethnographic film: Hypermedia and scholarship. In Anthropological film and video in the 1990s, ed. J. R. Rollwagen, 131–76. Brockport, NY: The Institute, Inc. Bishop, J., and N. Bishop. 2012. An opportunity to sample and reflect. American Anthropologist 114: 146. Blackman, M. B. 1986. Visual ethnohistory: Photographs in the study of culture history. In Studies in Third World societies, Publication 35, ed. D. Weidman, G. Williams, and M. Zamora, 137–66. Williamsburg, VA: William and Mary College. Blackman, M. B. 1992. Of “peculiar carvings and architectural devices”: Photographic ethnohistory and the Haida Indians. In Anthropology and photography 1860–1920, ed. E. Edwards, 137–42. New Haven, CT, and London: Yale University Press in association with The Royal Anthropological Institute. Blakely, T. D., and J. S. Williams. 1995. Anthropological excellence in film: Ten years of award winners in the SVA/AAA Film and Video Festival. Society for Visual Anthropology Publications. Arlington, VA: Society for Visual Anthropology, a section of the American Anthropological Association. Brand, S. 1976. For God’s sake, Margaret. Co-Evolutionary Quarterly 10: 32–44. Brand, S. N.d. For God’s sake, Margaret: Conversation with Gregory Bateson and Margaret Mead. http://www.oikos.org/forgod.htm (accessed March 9, 2013). Caldarola, V. 1987. The generation of primary photographic data in ethnographic fieldwork and the problem of objectivity. In Visual explorations of the world, ed. M. Taureg and J. Ruby, 217–27. Aachen, Germany: Edition Heredot. Chalfen, R. 1975. Cinema naivete: A study of home moviemaking as visual communication. Studies in the Anthropology of Visual Communication 2: 87–103. Chalfen, R. 1987. Snapshot versions of life. Bowling Green, OH: Bowling Green State University Popular Press. Chalfen, R. 1992. Picturing culture through indigenous imagery: A telling story. Manchester, UK: Manchester University Press. Chalfen, R., and R. Michael. 2007. Combining the applied, the visual and the medical: Patients teaching physicians with visual narratives. In Visual interventions: Applied visual anthropology, ed. S. Pink, 53–70. New York: Berghahn Books. Chanock, F. O., and E. R. Sorenson. 1975. Research films and the communications revolution. In Principles of visual anthropology, ed. P. Hockings, 431–38. The Hague: Mouton. Cohen, H. 2001. Early ethnographic film and documentary: Points of contact. In Origins of visual anthropology: Putting the past together. Conference Proceedings. Gottingen, German: IWF—Knowledge and Media. Collier, J., Jr. 1967. Visual anthropology: Photography as a research method. New York: Holt, Rinehart, & Winston. Collier, J., Jr. 1988. Visual anthropology and the future of ethnographic film, ed. J. R. Rollwagen, 73–96. Chur, Switzerland: Harwood Academic Press. Collier, J., Jr., and M. Collier. 1986. Visual anthropology: Photography as a research method. Albuquerque: University of New Mexico Press. Colombres, A. 1985a. Cine, antropología y colonialismo. Buenos Aires: Ediciones del Sol: CLACSO. Colombres, A. 1985b. Prologo. In Cine, antropología y colonialismo, ed. A. Colombres, 11–54. Buenas Aires: Ediciones del Sol: CLACSO.

14_107-Bernard.indb 458

6/9/14 10:03 AM

13: Visual Anthropology

459

Crawford, P. I., and J. K. Simonsen. 1992. Ethnographic film aesthetics and narrative tradition: Proceedings from NAFA2. Aarhus, Denmark: Intervention Press. Crawford, P. I., and D. Turton. 1992. Film as ethnography. Manchester, UK: Manchester University Press. Davey, G. 2008. Twenty years of Visual Anthropology. Visual Anthropology 21: 189–201. de Brigard, E. 1995. The history of ethnographic film. In Principles of visual anthropology, ed. P. Hockings, 13–45. Berlin: Mouton. de Heusch, L. 2001. The prehistory of ethnographic film. In Origins of visual anthropology: Putting the past together, ed. E. Beate and H. Rolf. Conference Proceedings. Gottingen, Germany: IWF—Knowledge and Media. Diawara, M. 1989. Oral literature and African film: Narratology in Wend Kuuni. In Questions of third cinema, ed. J. Pines and P. Willemen, 199–221. London: British Film Institute. Edwards, E. 1992. Anthropology and photography 1860–1920. New Haven, CT, and London: Yale University Press in association with The Royal Anthropological Institute. El Guindi, F. 1988. The making of El Sebou’. Visual Anthropology 1: 499–507. El Guindi, F. 1990b. El Sebou’: Film study guide. Los Angeles: El Nil Research. El Guindi, F. 1993. Charting content, freezing structure: A methodological base for visual ethnography. In Anthropological film and video in the 1990s, ed. J. R. Rollwagen, 11–36. Brockport, NY: The Institute, Inc. El Guindi, F. 1996. Film study guide: Egyptian Celebration of Life Series—El Sebou’. Los Angeles: El Nil Research. El Guindi, F. 1998. From pictorializing to visual anthropology. In Handbook of methods in cultural anthropology, ed. H. R. Bernard, 459–512. Walnut Creek, CA: AltaMira. El Guindi, F. 2001a. Beyond the visible: Method & theory in visual anthropology. In Origins of visual anthropology: Putting the past together, ed. E. Beate and H. Rolf. Conference Proceedings. Gottingen, Germany: IWF—Knowledge and Media. El Guindi, F. 2001b. Beyond picturing culture: A critique of a critique. American Anthropologist 103: 522–27. El Guindi, F. 2004. Visual anthropology: Essential method and theory. Walnut Creek, CA: AltaMira. El Guindi, F. 2007. Back to the future of visual anthropology: From rock art, to visual ethnography, to Powerpoint. In Memories of the origins of ethnographic film, ed. B. Engelbrecht, 437–58. Frankfurt: M. Peter Lang Verlag. El Guindi, F. 2011a. Kinship by suckling: Extending limits on alliance. Anthropologicheskii Forum (Forum for Anthropology and Culture). Peter the Great Museum of Anthropology and Ethnography (Kunstkamera). Russian Academy of Sciences. Special Forum on Kinship, Forum No. 15: 381–84. El Guindi. F. 2011b. Review of visual interventions: Applied visual anthropology. The Journal of the Royal Anthropological Institute (Man) 17: 678–79. El Guindi, F. 2012a. Milk and blood: Kinship among Muslim Arabs in Qatar. Anthropos A.107: 545–55. El Guindi, F. 2012b. Suckling as kinship. Anthropology Newsletter (American Anthropological Association) 53. El Guindi, F. 2012c. Suckling, adoption and the incest taboo: Significance of studying kinship for social science and for Qatar. Annual Research Forum Proceedings (Bloomsbury Qatar Foundation Journals): 201242P. El Guindi, F., and J. Williams. 1995. The Society for Visual Anthropology Film/Video Festival: A short history. Program, xv–xvi. American Anthropological Association, 94th Annual Meeting, Washington, DC.

14_107-Bernard.indb 459

6/9/14 10:03 AM

460

Fadwa El Guindi

Faris, J. C. 1992. Photography, power and the southern Nuba. In Anthropology and photography 1860–1920, ed. E. Edwards, 211–17. New Haven, CT, and London: Yale University Press in association with The Royal Anthropological Institute. Fischer, M. M. J. 1997. Raising questions about Rouch. American Anthropologist 99: 140–43. Ginsburg, F. 1995. The parallax effect: The impact of aboriginal media on ethnographic film. Visual Anthropology Review 11: 64–76. Griaule, M. 1938a. Masques Dogon. Paris: Memoires de l’Institut d’Ethnologies, Musée de l’Homme. Griaule, M. 1957. Methode de l’ethnographie. Oarus: Presses Universitaires de France. Haddon, A. C. 1895. Photography and folklore. Folklore 6: 222–24. Heider, K. G. 1976. Ethnographic film. Austin: University of Texas Press. Heider, K. G. 1983. Fieldwork with a cinema. Studies in Visual Communication 9: 2–10. Heider, K. G. 1991. Indonesian cinema: National culture on screen. Honolulu: University of Hawaii Press. Heider, K. G., P. A. R. Blakely, and T. D. Blakely. 2006. Seeing anthropology: Cultural anthropology through film, 4th ed. Boston: Allyn & Bacon. Heider, K. G., and C. Hermer. 1995. Films for anthropological teaching, Vol. 29. Arlington, VA: American Anthropological Association. Hockings, P. 1975. Principles of visual anthropology. The Hague: Mouton. Hockings, P. 1988. Ethnographic filming and the development of anthropological theory. In Cinematographic theory and new dimensions in ethnographic film, ed. P. Hockings and Y. Omori, 205–24. Osaka: National Museum of Ethnology. Hockings, P., and Y. Omori. 1988. Cinematographic theory and new dimensions in ethnographic film. Osaka: National Museum of Ethnology. Husmann, R. 1983. Film and fieldwork: Some problems reconsidered. In Methodology in anthropological filmmaking: Papers of the IUAES-Intercongress Amsterdam 1981, ed. N. C. R. Bogaart and H. W. E. R. Ketelaar, 93–112. Gottingen, Germany: Edition Herodot. Jacknis, I. 1984. Franz Boas and photography. Studies in Visual Communication 10: 2–60. Jacknis, I. 1988. Margaret Mead and Gregory Bateson in Bali: Their use of photography and film. Cultural Anthropology 3: 160–77. Jacknis, I. 1990. James Mooney as an ethnographic photographer. Visual Anthropology 3: 2–3. Jacknis, I. 1992. George Hunt, Kwakiutl photographer. In Anthropology and photography 1860–1920, ed. E. Edwards, 143–51. New Haven, CT, and London: Yale University Press in association with The Royal Anthropological Institute. Kendon, A. 1980. Gesticulation and speech: Two aspects of the process of utterance. In The relationship of verbal and nonverbal communication, ed. M. R. Key, 207–27. The Hague: Mouton. Kendon, A. 1990. Conducting interaction: Patterns of behavior in focused encounters. Cambridge: Cambridge University Press. Kendon, A. 1991. Some considerations for a theory of language origins. Man 26: 199–221. Kendon, A. 2004. Gesture: Visible action as utterance. Cambridge: Cambridge University Press. Kendon, A., and A. Ferber. 1973. A description of some human greetings. In Comparative ecology and behavior of primates, ed. R. P. Michael and J. H. Crook, 591–668. London: Academic Press. Koloss, H.-J. 1983. The ethnological film as a medium of documentation and as a method of research. In Methodology in anthropological filmmaking: Papers of the IUAES-Intercongress Amsterdam 1981, ed. N. C. R. Bogaart and H. W. E. R. Ketelaar, 87–92. Gottingen, Germany: Edition Herodot. Leroi-Gourhan, A. 1948. Cinema et sciences humaines—Le film ethnologique existe-t-il? Revue de Geographie Humaine et d’Ethnologie No. 3: 42–51.

14_107-Bernard.indb 460

6/9/14 10:03 AM

13: Visual Anthropology

461

Loizos, P. 1993. Innovation in ethnographic film: From innocence to self–consciousness, 1955–85. Chicago: University of Chicago Press. MacDougall, D. 1978. Ethnographic film: Failure and promise. Annual Review of Anthropology 7: 405–25. MacDougall, D. 1995. Beyond observational cinema. In Principles of visual anthropology, ed. P. Hockings, 115–32. Berlin: Mouton. Marcus, B. 2001. Visual methods in social research. London: Sage. Marks, D. 1995. Ethnography and ethnographic film: From Flaherty to Asch and after. American Anthropologist 97: 339–47. McQuown, N. A., and G. Bateson. 1971. The natural history of an interview, Nos. 95–98. Department of Photoduplication, Joseph Regenstein Library, University of Chicago. Mead, M. 1962. Retrospects and prospects. Anthropology and human behavior. In Anthropology and human behavior, ed. T. Gladwin and W. Sturtevant, 115–49. Washington, DC: American Anthropological Society of Washington. Mead, M. 1963. Anthropology and the camera. In The encyclopedia of photography, ed. W. D. New York: Greystone Press. Mead, M. 1995. Visual anthropology in a discipline of words. In Principles of visual anthropology, 2nd ed., ed. P. Hockings, 3–10. Berlin: Mouton. Naroll, R., and R. Cohen, eds. 1970. Handbook of method in cultural anthropology. New York: Natural History Press. Nijland, D. J. 2002. A history of ethnographic film, video and multimedia in the Netherlands. In Tales from Academia: History of Anthropology in the Netherlands, ed. H. Vermeulen and J. Kommers, 893–935. (Nijmegen Studies in Development and Cultural Change, 40). Saarbrücken, Germany: Verlag für Entwicklungspolitik. Omori, Y. 1988. Basic problems in developing film ethnography. In Cinematographic theory and new dimensions in ethnographic film, ed. P. Hockings and Y. Omori, 191–201. Osaka: National Museum of Ethnology. Peterson, A. Y. 1975. Some methods of ethnographic filming. In Principles of visual anthropology, ed. P. Hockings, 185–90. The Hague: Mouton. Piault, C. 1994. Some thoughts on verbal and non-verbal expression in anthropological films. In Evolution, technique, strategies, problems, Durban, South Africa, 1994, 1–11. The Campbell Collections and Centre for Oral Studies, University of Natal. Pink, S., ed. 2007. Visual interventions. Oxford: Berghahn Books. Pink, S. 2011. Images, senses and applied visual anthropology. Visual Anthropology 24: 437–54. Regnault, F.-L. 1931. Le role du cinema en ethnographie. La Nature 59: 304–6. Rollwagen, J. R. 1988a. Anthropological filmmaking: Anthropological perspectives on the production of film and video for general public audiences. Chur, Switzerland: Harwood Academic Publishers. Rollwagen, J. R. 1988b. The role of anthropological theory in “ethnographic film-making.” In Anthropological filmmaking: Anthropological perspectives on the production of film and video for general public audiences, ed. J. R. Rollwagen, 287–315. New York: Harwood Academic Publishers. Rouch, J. 1979. La camera et les hommes. In Pour une anthropologie visuelle, ed. C. de France, 53–71. Paris: Mouton. Rouch, J. 1995a. The camera and man. In Principles of visual anthropology, 2nd ed., ed. P. Hockings, 79–98. Berlin: Mouton. Rouch, J. 1995b. Our totemic ancestors and crazed masters. In Principles of visual anthropology, ed. P. Hockings, 217–32. Berlin: Mouton.

14_107-Bernard.indb 461

6/9/14 10:03 AM

462

Fadwa El Guindi

Roy, R., and J. Jhala. 1992. An examination of the need and potential for visual anthropology in the Indian social context. In Visual anthropology and India: Proceedings of a seminar, ed. K. S. Singh, 16–30. Calcutta: Anthropological Survey of India. Ruby, J. 1975. Is an ethnographic film a filmic ethnography? Studies in Visual Communication 6: 104–11. Ruby, J. 1981. Seeing through pictures: The anthropology of photography. Camera-Lucida: The Journal of Photographic Criticism 1: 19–32. Ruby, J. 1983. An early attempt at studying human behavior with a camera: Franz Boas and the Kwakiutl–1930. In Methodology in anthropological filmmaking: Papers of the IUAES Intercongress, Amsterdam, 1981, ed. N. C. R. Bogaart and H. W. E. R. Ketelaar, 25–38. Gottingen, Germany: Editions Herodot. Sabae, H., and M. Westmoreland, eds. 2012. Visual productions of knowledge: Toward a different Middle East, Vol. 31. Cairo: The American University in Cairo Press. Sahay, K. N. 1983. Ethnographic films in India: Prospect, priorities and proposals. In Methodology in anthropological filmmaking: Papers of the IUAES-Intercongress Amsterdam 1981, ed. N. C. R. Bogaart and H. W. E. R. Ketelaar, 49–60. Gottingen, Germany: Edition Herodot. Said, E. W. 1978. Orientalism. New York: Pantheon Books. Scherer, J. C. 1992. The photographic document: Photographs as primary data in anthropological enquiry. In Anthropology and photography 1860–1920, ed. E. Edwards, 32–41. New Haven, CT, and London: Yale University Press in association with The Royal Anhropological Institute. Schneider, A. 1997. Visual anthropology: Some additions to an essential sourcebook, Review of Principles of Visual Anthropology by Paul Hockings, 1995, 2nd ed. The Mouton World Anthropology series, Berlin: Mouton. Current Anthropology 38: 704–5. Singh, K. S. 1992. Cultural policy, cultural pluralism and visual anthropology: An Indian perspective. In Visual anthropology and India: Proceedings of a seminar, ed. K. S. Singh, 7–15. Calcutta: Anthropological Survey of India. Sorenson, E. R. 1995. Visual records, human knowledge, and the future. In Principles of visual anthropology, ed. P. Hockings, 493–506. Berlin: Mouton. Sorenson, E. R., and A. Jablonko. 1975. Research filming of naturally occurring phenomena: Basic strategies. In Principles of visual anthropology, ed. P. Hockings, 151–66. The Hague: Mouton. Spencer, P. 1975. Review of The Masai. Royal Anthropological Institute News 9: 1–6. Spencer, P. 1988. The Maasai of Matapato. Manchester, UK: Manchester University Press. Stadhams, D. 2007. Look to learn. The role for visual ethnography in the elimination of poverty. In Visual interventions, ed. S. Pink, 119–42. New York. Oxford: Berghahn Books. Taureg, M. 1983. The development of standards for scientific films in German ethnography. In Methodology in anthropological filmmaking: Papers of the IUAES-Intercongress, Amsterdam 1981, ed. N. C. R. Bogaart and H. W. E. R. Ketelaar, 61–86. Gottingen, Germany: Edition Herodot. Taylor, L. 1994. Visualizing theory: Selected essays from V.A.R. 1990–1994. New York: Routledge. Ushijima, I. 1988. The situation of visual anthropology in Japanese universities. In Cinematographic theory and new dimensions in ethnographic film, ed. P. Hockings and Y. Omori, 135–50. Osaka: National Museum of Ethnology. Ushiyama, J. 1975. Anthropological programming in Japanese television. In Principles of visual anthropology, ed. P. Hockings, 445–47. The Hague: Mouton. Williams, J. S., J. S. Williams, A. Jablonko et al. 1995. SVA/AAA Film and Video Festival: A short history revisited. In Anthropologcial excellence in film: Ten years of award winners in

14_107-Bernard.indb 462

6/9/14 10:03 AM

13: Visual Anthropology

463

the SVA/AAA Film and Video Festival, ed. T. D. Blakely and J. S. Williams, vii–viii. Arlington, VA: Society for Visual Anthropology, a section of the American Anthropological Association. Wintle, P., and J. P. Homiak. 1995. Guide to the collections of the Human Studies Film Archives. Washington, DC: National Museum of Natural History, Smithsonian Institution. Wolf, G. 1961. Der systematischen filmischen Bewegungsdokumentation. In Der Film in Dienste der Wissenschaft: Festschrift zur Einweihung des Neubaues für das IWF, ed. G. Wolf. Gottingen, Germany: IWF. Wolf, G. 1972a. Encyclopaedia cinematographica 1952–1972. Gottingen, Germany: IWF. Worth, S. 1966. Film as a non-art: An approach to the study of film. The American Scholar 35: 322–34. Worth, S. 1968. Cognitive aspects of sequence in visual communication. AV Communication Review 16: 121–45. Worth, S. 1969. The development of a semiotic film. Semiotica 1: 282–21. Worth, S. 1972. A semiotic of ethnographic film. Program in Ethnographic Film (PIEF) Newsletter 3: 8–12. Worth, S. 1980. Margaret Mead and the shift from “visual anthropology” to the “anthropology of visual communication.” Studies in Visual Communication 6: 15–22. Worth, S. 1981a. The development of a semiotic of film. In Studying visual communication, ed. L. Gross, 181–21. Philadelphia: University of Pennsylvania Press. Worth, S. 1981b. Studying visual communication. Philadelphia: University of Pennsylvania Press. Zeller, A. 1992. Grooming interactions over infants in four species of primates. Visual Anthropology 5: 63–86.

14_107-Bernard.indb 463

6/9/14 10:03 AM

14_107-Bernard.indb 464

6/9/14 10:03 AM

CHAPTER 14

Ethnography of Online Cultures Jeffrey G. Snodgrass

With the growing influence of the Internet, an ever larger proportion of social interaction either happens online or is impacted by life online. In some cases of interest to anthropologists, life offline goes online, as when indigenous or diasporic groups represent themselves to themselves and others in online blogs and bulletin boards (Landzelius 2006). In other cases, life online goes offline, as when mobile and increasingly smart cell phones bring cyberspace into offline contexts (Horst and Miller 2006). In still other situations, networked game spaces like World of Warcraft (WoW) and less clearly game-like online places such as Second Life function, in certain ethnographers’ understanding, as nearly self-contained and autonomous virtual worlds demonstrating structures and processes not explained by, or even fully connected to, life offline (Boellstorff 2008). Beginning in the 1990s, a science and technology studies offshoot of anthropologists began documenting the social life of the Internet ethnographically (Fischer 2007). These scholars of the digital age sometimes saw in computers the liberatory possibility for a new post-human subject and network society, where newly minted cyborgs might combine the best of the human and the machine (Castells 1996; Haraway 1991; Hayles 1999; Negroponte 1996; Turkle 1995, 2011). Or they could celebrate the social connectivity, communicative interactivity, and creative freedom seemingly facilitated by Web 2.0: “a term used to distinguish contemporary social media (wikis, blogs, embedded videos) from their immediate predecessors, the static web pages and message forums that had characterized what was retroactively dubbed Web 1.0 technologies” (Coleman 2010b, 489). Still, scholarly optimism was invariably contested both within these studies and from other sources (e.g., Hindman 2009; Mosco 2004), intermittently tempered by sober reminders of a continuing digital divide, with new media and communications technologies accessible only to some (Warschauer 2004). More recently, anthropologists continue to document the diverse ways computermediated interactions both transform and also reaffirm offline social experience, either improving or compromising offline existence (or, indeed, doing both simultaneously). Consciously evoking ancestral sites like Mead’s Samoa, intrepid ethnographers now routinely venture into newly emerging online cultures like Second Life (Boellstorff 2008). Despite referencing island societies and even anthropological mythology, these researchers do not generally peddle ethnographic exotica, but instead demonstrate how the Internet explains much of interest in contemporary life and, indeed, potentially serves as a metaphor for society’s new digital “natives” and “tribes” (Adams and Smith 2008). 465

14_107-Bernard.indb 465

6/9/14 10:03 AM

466

Jeffrey G. Snodgrass

Likewise, anthropological scrutiny is not confined to relatively self-contained Internet-based virtual worlds like Second Life and WoW, which provide backdrops and experiences approaching the depth and sensorial richness of offline life. Rather, ethnographers now regularly document social-networking sites like Facebook as well as multifarious blogs, webcasts, Internet hacks and scams, fan-sites, games, chatrooms, and bulletin boards, all of which intersect both with each other and with the offline world. Indeed, such is the energy of ethnographers in documenting social activity on and through the Internet—the largest, most visible, and most important of these new technologically mediated social spaces—that anthropological scholars summarize only with difficulty a now burgeoning literature of life online (Coleman 2010b; Wilson and Peterson 2002). In this chapter, I will sketch strategies cultural anthropologists and closely related others have used to document the social life of groups having an important online component. Excellent methodological reflections on how to study life online include: Arnason (2011), Burrell (2009), Garcia et al. (2009), Hine (2000, 2005), Markham (1998), Markham and Baym 2009, Miller and Slater (2000), and Murthy (2008). See also Boellstorf et al.’s (2012) illuminatingly practical handbook of methods more narrowly focused on conducting ethnography in virtual worlds. This chapter focuses on the strategies anthropologists have used to document life online, though it draws on insights of scholars from other fields. It thus encompasses approaches like so-called netnography (Inter[net] + eth[nography]), a term consumer market researchers influenced by anthropological methods use to describe their documentation of Internet activity (Jupp 2006; Kozinets 2002, 2006a, 2006b). The chapter concentrates on the ethnography of life online. The term “ethnography” refers to a case study account of a particular group’s practices and points of view, which relies largely on qualitative field-based methods like participant observation (Agar 1980, 1986; Spradley 1979, 1980). Immersed as participant observers in naturalistic social settings, ethnographers typically engage in long-term face-to-face interactions with members of a particular group, complementing first-hand experience and self-reflection with more distanced and objective observations of everyday life. Although usually centered around qualitative methods, ethnographic projects of life online, like their offline counterparts, can be enhanced by incorporating more structured data collection methods and systematic analytical strategies (Aunger 2004; Gravlee et al. 2012; Handwerker 2001). This chapter is inspired by ethnographic thought and practice developed and honed in the offline, physical world: Malinowski’s (1922) early emphasis on the importance of fieldwork in the context of his research in the Trobriand Islands, Geertz’s (1973) arguments for the importance of “thick description” and capturing the “native’s point of view,” Agar’s (1980) concept of the anthropological fieldworker as a “professional stranger,” Aunger’s (2004) relatively recent attempt to integrate both the “writing cultures” critique of ethnographic authority (Clifford and Marcus 1986), and structured methods like cultural consensus analysis (Romney et al. 1986) into a revamped “reflexive ethnographic science.” As such, this chapter will suggest that ethnographers of life online need to adjust standard anthropological methods and perspectives to the unique technologically mediated nature of the Internet, rather

14_107-Bernard.indb 466

6/9/14 10:03 AM

14: Ethnography of Online Cultures

467

than breaking entirely with such methods and perspectives (Boellstorff et al. 2012; Kozinets 2002, 2006a, 2006b). Overall, the chapter focuses on how to ethnographically document the social activity of groups whose members communicate and interact at least in part through computer-mediated technologies like the Internet. It reveals how digital technologies can enhance and complement even traditional (offline) ethnographic practice. However, the focus is on the former and not the latter, that is, on an “ethnography of life online,” rather than on so-called online, digital, or virtual ethnography (for the latter, where online data plays a typically supplementary rather than primary role, see Murthy 2008). With this in mind, I now turn to a number of interrelated issues related to ethnographic data collection, followed by a consideration of the analysis and presentation of online data, a discussion of research ethics, and brief concluding remarks. DOES AN ETHNOGRAPHY OF LIFE ONLINE NECESSARILY UNFOLD ONLINE?

In his study of Second Life, Boellstorff (2008) argues for studying so-called virtual worlds from entirely within their online persistent realities. He points out that these places exist in their own right and are not merely expressions of offline life. Suggesting that virtual realities like Second Life establish their own contexts, he points out that most of their residents do not meet their fellow residents offline. Thus, he reasons, “actual-world sociality cannot explain virtual-world sociality” (2008, 63). In contrast, Malaby (2009) conducts his own study within the actual-world of San Francisco Bay area Linden Lab, the corporate creators of Second Life. Malaby’s study consists of face-to-face interactions with Linden Lab employees in real-world settings. He himself worked as an employee of the lab, doing light work for them in their actual-world office, but, as he says, “nothing that put me in contact with Second Life users” (2009, 3). Thus, ethnographies of life online can take very different forms, depending on a researcher’s agenda. And ethnographers of life online need to make tactical decisions about how best to track their research subjects, wherever these subjects might wander. This could be through webs of interlinked online fan sites, blogs, and newsgroups devoted to Second Life, offline in Linden Lab, or somewhere complexly in between, as when subjects access Second Life through their smartphones while sipping lattes in an actual-world café with an offline group of friends, some of them themselves also simultaneously in Second Life. In fact, many ethnographies of online activity differently balance their documentation of online as compared to offline contexts, again pointing to the variations of research agendas and also the flexibility of ethnography to respond to differing study goals and research subject habits. Magnet’s (2007) scrutiny of the texts and photos of models and members of the website Suicide Girls (www.suicidegirls.com)—“a commercial site which features the online journals, profiles and nude photographs of young, heavily tattooed, punk women” (p. 577)— unfolds almost wholly online within this site’s heavily trafficked web pages and forums. Coleman (2009, 2010a) locates her studies of open source software designers and hackers partly online, visiting, for example, software designer and hacker Internet forums. She also participates in person in offline hacker conferences with informants who are often thought to be quintessential digital subjects. In her study of Everquest, Taylor (2006) plays this “massively

14_107-Bernard.indb 467

6/9/14 10:03 AM

468

Jeffrey G. Snodgrass

multiplayer online role-playing game” (MMORPG or MMO for short), thus conducting much of her research in-game and online. But she also studies gamers in actualworld gamer conferences and conventions. Nardi (2010), too, conducted her fieldwork largely within the MMO, WoW. She also spent a lot of time poring over “WoW-related websites, blogs, forums, wikis, and news articles” (p. 30), studying how these online sites fit together and communicated with each other, an important ethnographic goal. Still, her research also unfolds in actualworld Internet cafés in China. Miller and Slater (2000, 5) say in their influential study of the Internet in Trinidad: “If you want to get to the Internet, don’t start from there.” They suggest instead that “we need to treat Internet media as continuous with and embedded in other social spaces” (p. 5). Thus, they conduct offline household surveys and, like Nardi, observe the social life of actual-world Internet cafés. Likewise, Burrell (2009, 188–89) noted the way online activity in Accra, Ghana was nested in a series of offline contexts, extending from the Internet café forming the focus of her study “to homes, churches, schools, foreign countries, into the future (if only imagined), and back to the Internet café.” This situation led this ethnographer to think of her field site as a complex network of online and offline spaces, people, and objects. Burrell came to believe that ethnographers of online activity could not hope to exhaustively study all the relevant nodes in this network. Instead, researchers must choose the spaces to include and exclude, which leads each ethnographer to study a slightly different slice of reality, or, in Burrell’s case, a somewhat idiosyncratic intersection of Internet sites and offline urban Accra realities. In this sense, and invoking points first made in reference to more traditional physical-world ethnographies (Gupta and Ferguson 1997), Burrell suggests that ethnographic field sites are, in some important sense, constructed rather than discovered. Together, these examples reinforce the idea that there is no single rule as to how ethnographers of life online should balance—or even construct—their mix of online as compared to offline observations. Rather, each researcher needs to think about the nature and habits of their object of study and also about their research question. For example, a growing body of anthropological and interdisciplinary research shows that Internet spaces like Second Life and online games like WoW and Everquest form communities and even relatively self-contained virtual worlds, providing as they do nearly complete sensorial and social experiences with little need to reference the outside world (Bainbridge 2010; Boellstorff 2008; Castronova 2005; Corneliussen and Rettberg 2008; Golub 2010; Malaby 2009; Nardi 2010; Pearce and Artemesia 2009; Snodgrass et al. 2011a, 2011b, 2011c; Snodgrass et al. 2012, 2013; Taylor 2006). Boellstorff (2008), for example, studies a reality that in some important sense exists independently with its own internal structures and processes. This gives a researcher like him at least the choice to study this online environment as a self-contained reality. Indeed, Boellstorff (2008, 61) says that, at least initially, he had no other specific research topic in mind than to learn what was the ethnographic potential of virtual worlds, again justifying his decision to locate his study entirely online, as a kind of methodological experiment.

14_107-Bernard.indb 468

6/9/14 10:03 AM

14: Ethnography of Online Cultures

469

Though working within a relatively self-contained virtual world, researchers can still track that world offline, as did Malaby (2009) in his own study of Second Life. Again, this decision was motivated by the nature of his research. Responding to a literature on labor, information technologies, and business life (Freeman 2000; Ross 2003), Malaby explored questions of corporate creativity, production, governance, and labor relations. He was interested in particular in how Linden Labs hoped to unleash their creative energies by freeing themselves from older bureaucratic structures and business models. This pushed him into the corporate spaces of this company, rather than the virtual spaces of their creation. Ethnographers of virtual world places like Second Life should realize that certain dimensions of that world might seem largely self-contained (like play groups unfolding in fantasy realms, where the members have never met offline in the physical world), others less so (as when actual-world politicians bring their agendas and campaigns into Second Life), and researchers will thus need to adjust their methods accordingly. Further, researchers cannot easily treat online life as self-contained—in the sense of largely, if not totally, disconnected from offline life and concerns—when the Internet serves as a medium for expressing so-called real-life concerns. This is certainly the case in the Internet cafés described in Nardi (2010), Miller and Slater (2000), and Burrell (2009). Initially rooted at least partly in offline reality, an ethnographer naturally moves into other offline spaces connected to these cafés. Likewise, diaspora communities can use the Internet to keep in touch with dispersed members, leading some researchers to wonder about the offline activities and issues that bind these groups together (Landzelius 2006). Members of the Occupy Wall Street (and elsewhere) movement utilize social media like Twitter—and the “#Occupy” (hash)tag in particular—to organize offline protest in physical space (Juris 2012). Iranian bloggers and protesters use cell phones and online social media to protest state-imposed realities, with such protests met by the state’s own online measures (Doostdar 2004; Sreberny and Khiabany 2010). Here, too, it seems important to document the offline contexts and activities that shape the online responses. The Internet may provide third-place respites from the first and second places of home and work (Steinkuehler and Williams 2006). In these situations, researchers might be driven into offline reality to better understand the pressures that are met with particular Internet responses. Indeed, exploring relationships between online and offline contexts in these situations, where social interactions are no longer face-to-face and spatially bound, allows anthropologists to further expand their understandings of community (Gupta and Ferguson 1997; Smith and Kollock 1999). Sometimes a research question itself will focus on intersections and even conflicts between online and offline realities. In my own study of WoW, I wondered how gamers’ mental health might be influenced, for better or for worse, by online play (Snodgrass et al. 2011a, 2011b, 2011c; 2012, 2013). The game can provide an important temporary respite from the offline world. But it can also be a time-consuming source of distress, interfering with actual-world lives. Methodologically, this led me to observe a mix of online, offline, and hybrid online–offline spaces. For example, I conducted interviews

14_107-Bernard.indb 469

6/9/14 10:03 AM

470

Jeffrey G. Snodgrass

and observations in players’ homes, hoping to observe how game-play might either enhance or interfere with actual-world commitments and relationships. Researchers need to be aware that no matter what choices they make, they will never exhaust the complex intersection of offline and online spaces that ground and shape respondents’ experience. For example, though I interviewed and observed players in Colorado in their actual-world homes, I was not able to do so similarly in Europe and East Asia, where some resided. In some cases, the influence of offline contexts on online realities is less visible. For example, Gershon (2010, 3) notes that researchers of online activity need to be aware of what she calls “media ideologies,” or, “beliefs about how a medium communicates and structures communication.” These ideologies, which are partially situated in offline contexts, direct people to use the Internet and other media in particular ways, as when they push users to see as appropriate and even attractive the possibility of ending offline romantic relationships via online electronic media like Facebook. Gershon is thus careful to balance her scrutiny of online behavior with a consideration of media ideologies that are not confined to the Internet. In choosing which online and offline spaces to include and exclude, ethnographers should think carefully about their object of study and the nature of their research question, which ideally go hand in hand. Studying offline contexts will invariably enrich a study of online life, showing how Internet behavior is informed by actual-world relationships, networks, and identities (or vice versa). Still, the time cost might be too high according to a particular project’s constraints, especially if the research question directs one elsewhere. Ethnographers need to remember that they cannot hope to explore all relevant online and offline sites. Rather, they will need to decide which sites—be they online, offline, or mixed—most clearly and efficiently reveal the social behavior of interest to the study. As such, based on the nature of the group in question and the study’s aims, ethnographers can argue for locating ethnography entirely online (Boellstorff 2008) or mostly offline (Malaby 2009). More likely, researchers will balance their study with a mix of online and offline observations. For ethnographers, finding this balance should take place before or in the early stages of fieldwork. Once this balance is struck, which can change as a project unfolds, researchers can situate themselves as participant observers, which brings us to the next section. IMMERSION IN A NATURALISTIC WEB SETTING: ONLINE PARTICIPANT OBSERVATION

As in offline situations, ethnographers of the Internet need to decide how to balance observation of online social life with participation in that reality. Ethnography of life online provides researchers with opportunities to immerse themselves in both online and offline places. This section will focus on the challenges and opportunities related to immersion in the former, given that the latter is treated elsewhere in this handbook. Participant observation exists along a continuum, from more or less exclusive observation to almost full participation. Though controversial, a researcher’s observing of activities without participating or even formally announcing him- or herself as a researcher—what some might equate with surreptitious Internet “lurking,” a term I

14_107-Bernard.indb 470

6/9/14 10:03 AM

14: Ethnography of Online Cultures

471

avoid because its negative and even nefarious connotations rob it of serious consideration as a research strategy—is, according to some ethnographers, a viable and advantageous methodological choice in online contexts (Garcia et al. 2009, 58–60). Online worlds, in particular, offer researchers the possibility of accessing detailed social behavioral data with little to no actual participation. And non–participant observation can be carried out even more easily in online as compared to offline contexts, where one’s non-interacting presence is more visible and thus might disturb naturally occurring behavior. Indeed, some have argued that disclosing one’s researcher presence might compromise both respondents’ anonymity and disturb the “naturalism” of online contexts, pointing to the potential superiority of purely observational over more participatory methods (Soukup 1999). Though a proponent of fuller participation in online communities, Kozinets (2002) learned a lot simply by eavesdropping on online conversations in an online coffee newsgroup, as did Magnet (2007) in the online Suicide Girls web pages and forums, Coleman (2009) in software developer blogs and bulletin boards, and Nardi (2010) in WoW-related websites. Such sites can be data mined for details that flesh out the social lives of these forums’ members. In fact, the Internet allows for relatively easy retrieval and storage of such data, as when Nardi (2010) (and I) use the “/chatlog” command within WoW to save members’ communications with each other via various in-game chat channels, thus accessing and storing even those conversations we either simply passively “overheard” or were not paying direct attention to as they unfolded. Further, initial non–participant observation allows ethnographers to learn about a particular online community, helping researchers make more informed choices about what is the right group and context for their study. For example, non–obtrusive observations might reveal whether a site is vibrant or not, shown by the number of members and posts. It can also give researchers a feel for the demographic makeup of a particular online site, which might be important if a study required choosing a group representative of a segment of Internet users or of the online population as a whole. Likewise, researchers can build up knowledge of an online community in a non-participating phase, learning the group’s particular language, norms, and customs. This might allow researchers, if and when they decide to finally interact with others (it should be kept in mind that ethnographers can easily move between more observation- and participation-heavy phases of research), to speak and act more like an insider, forming better first impressions and potentially allowing for better eventual acceptance into the online community. Having an extended observational rather than participatory period of time in an online reality also allows ethnographers to develop meaningful and relevant research questions and hypotheses. Similar to Boellstorff (2008) in his account of Second Life, Nardi (2008, 27), in her study of WoW, says: “When I began my study, I had no hypotheses or precise research questions.” She refers to this as a “go with the flow” pattern (p. 27), letting her research questions emerge naturally from the setting rather than from her own preconceptions and agendas. These are extreme examples, as most anthropologists do start with research questions and hypotheses in mind, even if they are uncomfortably vague and eventually abandoned for more intriguing ones as the research develops. (In fact, Boellstorff and

14_107-Bernard.indb 471

6/9/14 10:03 AM

472

Jeffrey G. Snodgrass

Nardi both developed more specific questions in time and give detailed and useful advice on how to do so in their jointly authored handbook on ethnography of virtual worlds [Boellstorff et al. 2012; see especially Chapter 4].) But these examples demonstrate the way ethnographic questions are famously induced—or emerge—from the ground up, ensuring that anthropological studies are properly grounded in insider thought and practice. Initial non–participating observations, relatively neutral in flavor, facilitate this process of research question development. Non–participating observation is thus easy to do, relatively unobtrusive, and requires little investment of a researcher’s time and resources. Still, despite certain clear advantages to non–participant observation in certain online contexts, even proponents of such techniques commonly recommend it as a beginning rather than an ending approach to ethnography of the Internet, given certain limits and even risks to conducting such observations (Garcia et al. 2009; Kozinets and Handelman 1998). For example, if you begin as a non–participating observer and then introduce yourself later as a participant, especially if you wait too long to introduce yourself, you risk making members feel “spied upon” (Garcia et al. 2009, 59; LeBesco 2004). Researchers might thus consider conducting non–participating observation just enough to determine group norms relevant to such activity (Garcia et al. 2009). Based on what is learned, researchers could then determine whether to continue with non– participant observation, in order to reap the benefits discussed above, or to move into a more participatory phase. Non–participating observation can allow ethnographers to be “participant-experiencers” of certain dimensions of online experience (Walstrom 2004), with access to personal and first-hand knowledge, in that such observations may allow researchers to reproduce experiences similar to members of the group, who might spend most of their time online simply viewing web material. Still, to tap the fuller potential of ethnography in capturing insider experience, researchers often favor a more participatory, interactional, and dialogical approach in their studies of life online, arguing that such practice reveals further complexities and nuances of life online (Boellstorff et al. 2002; Hine 2000; Kozinets 2002; Markham 1998). Participating in life online thus exists along its own continuum. One can simply read emails from group members, occasionally post on a web forum, download newsgroup postings, or sometimes chat up or even play with others in an online gaming reality. But this feels different, even superficial, compared to the way Senft (2008) became an actual camgirl (women who webcast their lives to audiences), how Chen (2009), Golub (2010), and Nardi (2010) “raided” in WoW, or the way Kozinets (2002) rose to be a kind of leader in the Star Trek fan community. This explains why some researchers establish typologies between what “membership” in online communities means: in Lave and Wenger’s (1991) scheme, a continuum from lurker, to novice, to regular, to leader, to elder; or, as Kozinets (2002) sees it, from tourist, to mingler, to devotee, to insider. One risk of researching life online as a participant observer is to assume insiderness before actually achieving that status. It is easy to imagine the Internet as a familiar space open and available to all, when, in fact, it contains cultural micro-niches. Speaking standard English and being fluent with texting language and culture helps research-

14_107-Bernard.indb 472

6/9/14 10:03 AM

14: Ethnography of Online Cultures

473

ers navigate online life. Still, mastering insider slang peculiar to a particular online community—invariably a complex mix of abbreviations and acronyms, emoticons, hidden innuendo and cultural references, and humor—requires time and experience comparable to learning a village dialect or, perhaps more appropriately, a “pidgin” mixing a standard language like English with something else altogether. Ethnographers need only follow Kozinets (2006b) as he decodes a Star Trek fan’s perplexing online posting to convince themselves of this fact. Mastering such lingo will signal various degrees of insiderness—that one is “leet” (elite) not a “newbie” or “noob” (new or amateur)—smoothing interactions with online community members and allowing an ethnographer to eventually merge into Internet settings. Researchers of life online should be aware that some groups or online experiences may remain closed even after months or more of engaged participatory research. For example, multi-player WoW raiding takes an extraordinary time commitment to build up appropriate playing skill and collaborative social smarts. Playerethnographers wishing to be raiders may need to submit formal applications to “guilds,” officially sanctioned persistent social groupings with built-in mechanisms for controlling membership and rank, which help players complete difficult collaborative tasks and endgame content such as those related to raiding (Williams et al. 2006). This may be followed by a lengthy trial period testing a player’s skill and commitment to the group, before eventually landing a spot on a raiding squad. Once accepted, ethnographers might need to commit to play several or more evenings a week, in addition to the time spent in preparation for the raids themselves, poring over sites like elitistjerks.com, mmo-champion.com, or arenajunkies.com, which offer WoW tips and strategies in an environment continually transformed by Blizzard Entertainment’s latest software expansions and patches. Some degree of immersive participant observation is now relatively standard in ethnographic approaches to online worlds. As Williams (2010, 3) tells us: “Scholars skipping this crucial participant observation step will ask the wrong questions in surveys, miss obvious explanations, and make incorrect assumptions about outcomes.” There is little consensus on how to balance more observational as opposed to more deeply participatory approaches. Rather, ethnographers are once again faced with choices, which should be weighed relative to research questions and goals, time and resources, and personal preferences. Researchers may decide it’s important to understand, firsthand, the experiences of an elite WoW raider, perhaps even finding the activity personally enjoyable and rewarding, adding to their desire to make it central to the study. An ethnographer should not enter lightly into such an enterprise. After all, a researcher’s time and energy are limited, and committing to raiding might take away from other data collection strategies, like more neutral observations. To help here, ethnographers might think to the theories one’s study engages and also to the intended eventual audience for one’s research and writing. Audiences vary in their ideas about what counts as legitimate ethnography and indeed legitimate data. Some like Boellstorff et al.’s (2012) authors may prefer even deep first-hand experiences of nearly all situations discussed. But others, like Soukup (1999), may argue for approaches that are more purely observational or only minimally participatory and thus less obtrusive and more objective.

14_107-Bernard.indb 473

6/9/14 10:03 AM

474

Jeffrey G. Snodgrass

ENTREE AND IMPRESSION MANAGEMENT IN AN ONLINE SETTING

Initial observations, then, whether more or less participatory, can effectively prepare the way for an eventual more fully participatory entrée into an Internet-based community setting (see Kozinets 2002, 2006a, 2006b). During early periods of fieldwork, ethnographers begin to learn the language and norms of their online group. They can also hone research questions, assuring they are better able to explain themselves to community members, thus minimizing early stumbles. Still, ethnographers need to keep myriad other factors in mind when entering a community as a more fully fledged participant. For example, when preparing to enter an online environment, researchers should be aware that they will be judged on their interactional style and visual image, as seen, for example, through the text they type and the avatar they manipulate. Online settings offer ethnographers a variety of communications technologies like email, bulletin boards, chat channels and rooms, social networking programs, instant messaging, and VOIP (voice over Internet protocols). One challenge in early Internet encounters is to identify and master the media and communications strategies peculiar to the group that is the focus of study. This entails, in part, technical mastery, but also familiarization with at first invisible rules of etiquette. Complicating the issue, exchanges in digital environments can be shorn of paralinguistic flavor—pauses, stutters, nervous laughter, repetitions, and body language—that might otherwise provide for fuller and more accurate communication. As such, ethnographers entering a new online community will sometimes be called on to quickly read and digest large amounts of text and images, which are at once overwhelming and denuded of meaning. Even before these initial encounters, ethnographers should consider small details like their user or avatar name, as the early choices they make will impact respondents’ impressions of the researcher and thus shape the information informants provide. Ethnographers should choose names both in accordance with the image they wish to project as well as community standards. In WoW, for example, avatar names and designs are often whimsical, though naming and appearance customs vary widely from server to server and guild to guild. Managing image and tone in such contexts is challenging, though researchers can control the situation to a greater extent by, for example, shifting between text- and voice-based chatting according to the situations and their goals. Ethnographers should be aware that shifting into a VOIP system may be shocking to respondents, providing as it does richer data, particularly when complemented by information transmitted through other channels. Such shifts could be especially jarring if timed incorrectly, or if a researcher’s offline persona differs greatly from that signaled by an avatar. Ethnographers will also be judged according to the skill they demonstrate in online worlds. In Second Life, the quality of one’s avatar design shows a user’s competency (Boellstorff 2008; Meadows 2008). In WoW, it is the won gear worn by an avatar that reveals a player’s level of expertise and thus insider status. Researchers should be aware that inhabitants of online worlds will try to infer the identity of the user behind the avatar, reading through online names, appearances, behaviors, and competencies to make judgments about each offline person’s gender, race, age, and level of education (Garcia

14_107-Bernard.indb 474

6/9/14 10:03 AM

14: Ethnography of Online Cultures

475

et al. 2009). For example, members of online videogaming communities associate bad play with being female, assuming that even a female-gendered avatar is actually male if she plays well (Beavis 2005). Of course, ignorance of technical details related to a particular online environment can be charming for newbie participants. And missteps can lead to tutoring from more experienced participants, which can help build social bonds between researchers and their respondents. However, failure to gain competency can also interfere with a successful entrée into a community, excluding researchers from important experiences characteristic of their group. In the long run, a lack of technical and social competence will be more isolating than beneficial. In these early encounters, ethnographers can decide how tactical and even deceitful they want to be. Compared to offline ethnography, it is easy to deceive online inhabitants about one’s aims and purposes as a researcher (more on the ethics of this in a moment). Indeed, one can be a kind of anti-member to a group, intentionally misbehaving in Internet troll fashion to see how others respond, thus potentially revealing how the group functions. Ethnographers should be aware that deception can flow both ways, from ethnographer to subjects and vice versa, which itself can be a source of ethnographic insight (Snodgrass 2006). More commonly, ethnographers will disclose their research aims and agenda (Boellstorff et al. 2012; Garcia et al. 2009). If appropriate, ethnographers may represent themselves as both researchers and also members of the group. Most of my collaborators in my WoW research were, like me, either gamers or at least friendly to gaming as an activity, information we shared with our respondents, which often worked to cement rapport. Initial participant observation may reveal important gatekeeper authorities who manage online environments. Ethnographers should respect these individuals and cultivate good and honest relationships with them. For example, in online gaming communities like WoW, one may wish to speak initially to guild leaders about the research, as I did, thus smoothing all dimensions of a study. Further, the web offers unique opportunities for researchers to contact and communicate with their research subjects. These can be particularly useful in an early exploratory phase just before or after one’s entrée, when researchers wish to communicate their aims or recruit for their project. Online groups often have websites and forums where information relevant to the group is posted. Ethnographers can make use of these, posting explanations of their research, letters of consent, and even online surveys and other instruments for collecting data. (The present WoW guild where I’m now working has a guild website and a Facebook group, where I post information relevant to my study.) Researchers can also strategically place ads in online newsgroups, mailing lists, and bulletin boards, either to explain research aims or to recruit for the study. Likewise, ethnographers can make relatively public announcements in online chat rooms and other Internet spaces. This will minimize their need to repeatedly reintroduce themselves and their projects individually to each and every research subject. It may be useful to communicate using institutional (university) letterhead for some groups but not for those perceived as hostile to institutional authority. Researchers should be aware that Internet group norms vary widely on issues related to how information is shared. Some groups are hostile to postings considered off topic, with such

14_107-Bernard.indb 475

6/9/14 10:03 AM

476

Jeffrey G. Snodgrass

postings interfering with rather than facilitating one’s study (Clegg Smith 2004). Any postings should be done thoughtfully, ideally with permission from the relevant online operators and in accordance with group norms. INTERVIEWING WEBIZENS

Informal conversations and even relatively spontaneous conversational interviews occurring in naturalistic settings are part and parcel of the ethnographer’s toolkit. It is particularly easy to collect and store online exchanges. And ethnographers of the Internet may be tempted to simply eavesdrop, or, minimally, to follow the flow of naturally occurring conversation, without overly influencing, directing, or even participating in such exchanges. Still, ethnographers of life online may complement their collection of such informal talk with more systematic interviews driven by a specific agenda and often based on a protocol of preplanned questions. Typically, such a protocol is built up from knowledge gained through participant observation, assuring that questions are relevant, meaningful, and, comprehensible to respondents. By developing a specific protocol, ethnographers direct respondents to issues of particular interest to a study, even when these are not be what respondents spontaneously want to speak about at that moment. Thus, Magnet (2007) fleshed out her study of Suicide Girls with data from email interviews, and Kozinets (2006b) conducted interviews via email and in person with Star Trek fans. Relying in part on research assistants, Nardi (2010) complemented her in-game ethnography of WoW with face-to-face interviews in Beijing and Shanghai with players who met together to play in Internet cafés. She also worked with undergraduates to conduct in-person interviews with players in the San Diego and Irvine areas. Similarly, Gershon’s (2010) study of real-life relationships that end online drew, in part, on face-to-face interviews conducted primarily with Indiana University undergraduates, many of whom she met through her classes. As these examples show, ethnographers of life online interview respondents via diverse media. They use the Internet itself, exchanging words through email or electronic chat programs, including text-based instant messaging programs, voice-based protocols like Ventrilo, or programs such as Skype, allowing the simultaneous exchange of text, voice, and visual images. Or they can forswear electronic communication altogether and speak face-to-face to research subjects. When choosing how to interview subjects, ethnographers of life online need to consider trade-offs. Interviewing via email is cheap and convenient, as is having interviewees respond to questions posted online, which are forwarded to the researcher or stored online for later retrieval. In the latter case in particular, a researcher can quickly collect information from a large number of easily located interviewees, depending on the websites’ traffic. If a project is presented in an alluring fashion, a researcher might hope that their protocol goes viral, forwarded from one informant to another. Likewise, the burden placed on both interviewers and interviewees is typically low in these contexts. Researchers place questions of interest either online or in subjects’ mailboxes, and interviewees respond at their convenience. Also, interviewees can ponder questions longer, potentially giving even more thoughtful, if more heavily edited and processed

14_107-Bernard.indb 476

6/9/14 10:03 AM

14: Ethnography of Online Cultures

477

responses. As an added advantage, electronic interviews like these will be typed and thus land in researcher hands as naturally occurring transcriptions. Online chat programs provide similar advantages. These provide cheap and convenient conversations. And they typically place a low burden on respondents, as I found when I used WoW’s in-game chat function to interview individuals as they played. Without the pressure to respond immediately to a flesh-and-blood interviewer, respondents take their time, again allowing for potentially more thoughtful answers. Interviewees can use the demands of online multi-tasking as an excuse to delay responses, with researchers never knowing when such delays are real or feigned, which renders interviews more comfortable for certain respondents, if more frustrating for interviewers. Also, a simple mouse-click saves these online conversations as well in ready-made transcribed form. Further, unlike the linear tit-for-tat exchanges characteristic of email and that often dominate social networking programs such as Facebook, chat programs allow for real-time dynamic exchanges between interviewer and interviewee, bringing them closer to more standard offline interviews. Interviewers might even argue that electronic conversations, utilizing as they do technologies of interest to the online studies themselves, yield additional insights into naturally occurring communication (Garcia et al. 2009). Electronic media might be the communication strategy with which online inhabitants are most comfortable, making them even more naturalistic. And the anonymity of an online interview might create a greater sense of safety, leading subjects to reveal more about their lives. This would especially be the case when having discussions about intimate issues, like one’s personal relationship to God, where interviewees may wish to avoid the judgments of researchers who might not share their commitments (Miller and Slater 2000, 183). However, electronic interviews, like all interviews, have their own limits. In face-toface interviews, researchers can more easily ask instantly for clarifications and expansions. This not only enhances a researcher’s depth of understanding but also shows interviewer interest and engagement, building rapport between anthropologists and participants. Such dynamic exchange is difficult, and even impossible, via email, where dialogue is protracted and thus of a different quality. Dialogue is more possible via online chat programs, as they allow for instant responses, but these, too, will lack much of the immediacy of a face-to-face exchange, given the manner that participants can delay responses. Further, there is the added risk of respondent distraction in interviews such as these if respondents are, for example, multitasking in WoW, killing monsters and gathering loot as they chat. Using Internet protocols like Skype, where images accompany text and sound, will help researchers avoid some limitations of email or chat interviews. Still, face-to-face interviews offer even richer paralinguistic cues like crossed arms, eye movements, hesitations and stutters, laughter, cracked voices, and tonal changes, which help ethnographers interpret meaning in and behind respondents’ words. Though reproducing more richness than interviews via email, chat programs, or even the telephone, Skype will not reproduce the full paralinguistic texture of in-person interviews. Indeed, counter to Miller and Slater’s (2000) point about how the anonymity of online interviews promotes intimacy, lacking such richness might hinder the formation of empathy between researchers and their subjects, leading respondents to reveal less about their lives.

14_107-Bernard.indb 477

6/9/14 10:03 AM

478

Jeffrey G. Snodgrass

No single interviewing media or method is appropriate for all contexts and projects. Nor is there any simple prescription about whether ethnographers of life online should turn to more structured interviews in the first place. Rather, researchers need to consider trade-offs between convenience, naturalness, data quality, personal preference, and the nature of the research situation and question. For example, Gershon (2010) was interested in how offline relationships end online. The heart of her research was based on in-person one-on-one interviews, digitally recorded and then transcribed for later analysis. She relied on such interviews because it was difficult for her to ethnographically document such breakups in their naturally occurring contexts as no one wanted her around when they broke up. Instead, she spoke to whoever was willing to discuss this most of intimate of topics. And she preferred the intimacy of face-to-face encounters. Also, it made added sense in her case to conduct interviews in person, given the way her research ultimately concerned actual-world offline relationships, which found their endings, but not necessarily their beginnings or middles, online. Given the complexity of these trade-offs, researchers might mix interview strategies. Some interviews could be conducted via online media and methods. Such a strategy would allow ethnographers to collect a great deal of data in a relatively short amount of time with little expense. Likewise, conducting at least some interviews via media preferred by online respondents could provide ethnographers with a better feel for the communication strategies characteristic of the online environments of interest to their projects. However, researchers could complement online interviews with others conducted offline, potentially enhancing the depth, and also the intimacy, of exchanges with their informants. Offline interviews would also help ethnographers verify the demographics of their interviewees, which is more difficult when relying exclusively on online methods. Researchers should consider similar trade-offs in the use of monitored and moderated group discussions around a particular topic related to life online—that is, when relying on focus groups, a kind of guided group interview where group discussions and interactions produce data. On the one hand, the temporal and spatial flexibility of Internet environments, where it is relatively easy to bring together even spatially distant and temporally separated people from diverse populations, make them particularly attractive places for holding focus groups. As in online interviews, participants in online focus groups, too, can take their time in carefully crafting thoughtful (and oftentimes typed) responses. And digital data resulting from online focus group discussions and interactions are again easily saved and retrieved. On the other hand, computer-mediated focus groups can lack the felt immediacy and physicality of faceto-face exchanges. Also, multiple threads can develop in online focus groups, which, though offering potentially rich data, can confuse participants and researchers, leading both away from topics of more central interest. Getting focus groups to meet synchronously in chat rooms or even via avatars in 3D-graphical environments like Second Life or graphical MUDs—as opposed to through asynchronous bulletin board and newsgroup postings, for example—can alleviate some of these problems, helping to provide immediacy and focus. Again, researchers might consider mixing online with offline focus groups to maximize the different potentials of interactions mediated by computers as opposed to offline

14_107-Bernard.indb 478

6/9/14 10:03 AM

14: Ethnography of Online Cultures

479

social conventions (on online focus groups, see O’Connor and Madge 2003; Stewart and Williams 2005; Sweet 2001). MIXING DATA COLLECTION METHODS: ONLINE SURVEYS AND BEYOND

Yee (2006, 2007) followed initial participant observation with quantitative analyses of large sample surveys, clarifying MMO player demographics and eventually revealing how player motivations and other variables were associated with healthy as opposed to so-called problematic or even addictive patterns of play. Online surveys such as Yee’s can enhance ethnography, illuminating patterns of social experience in even very large data sets (30,000 respondents in this case) quickly and at modest expense. Yee and his collaborators have also been able to sample players from around the world, with current PlayOn 2.0 work (http://blogs.parc.com/playon/) comparing WoW play in the United States, East Asia, and Europe, a global reach that is more difficult and expensive to obtain using either qualitative methods or even traditional offline surveying techniques. The data generated by surveys such as Yee’s are readily exported into SPSS or other quantitative analysis programs, allowing researchers to skip time-consuming and expensive manual data entry steps characteristic of non-electronic offline surveys. Researchers have even argued that online surveys have higher response rates, as it is easy to send respondents email reminders about the survey (Moss and Hendry 2002, 586; Murthy 2008, 842), though this point is contested (Leece et al. 2004). As with online interviews, there is always the possibility that respondents will find the survey of interest and forward it to others, which can expedite the recruitment process. Yee also collects open-ended qualitative interview data in his surveys, which, like online interviews, is retrievable without the need for transcription. And Gunter et al. (2002, 233) note that online surveys are typically completed more quickly and have more detailed responses to such open-ended questions. On the benefits of online surveys, see Brewer and Hunter (2006) and Murthy (2008). In my own WoW research, I complemented participant observation with more structured interviews and conducted small sample surveys utilizing free lists and rating tasks to understand relationships between cultural conceptualizations of online and offline success and well-being (Snodgrass et al. 2011c; Snodgrass et al. 2013; for a comparison of web-based free lists with similar data collected in face-toface interviews and self-administered paper questionnaires, see Gravlee et al. 2012). These were presented through Google’s online survey tool to gamers, for example, via postings in guilds with which I and my research team were associated. My data collection culminated with a formal web questionnaire using the Limeservice software, which I posted on WoW blogs and gamer sites and again circulated through personal play networks. (A survey similar to that employed in the present research is ongoing [http://tinyurl.com/WoWwellness].) In addition to Limeservice (which I like), other online survey instruments include Survey Monkey, Qualtrics, and Zoomerang. These vary in their ease of use, costs (typically either free or minimal), quantitative data storage and export capabilities, compatibility with other software, privacy policies, and quality of their interface. Ethnographers should thus scrutinize each carefully, perhaps taking advantage of some of the paid services’ free trials.

14_107-Bernard.indb 479

6/9/14 10:03 AM

480

Jeffrey G. Snodgrass

Ethnographers of life online wishing to integrate more structured data into their projects should not limit themselves to surveys. Chen et al. (2008) designed a software add-on or “mod” for WoW that allowed them to turn in-game characters into datagathering “bots,” systematically sampling various play dimensions to understand guild dynamics. Kosminsky (2010) analyzed data collected from websites whose creators had themselves created add-ons to track WoW economic activity. Such methods can be readily integrated into qualitative ethnographic projects. For example, a team of researchers at Stanford and the Palo Alto Research Center conducted in-game participant-observation in WoW immediately after the game’s launch in November 2004 (Ducheneaut and Moore 2004; Ducheneaut et al. 2006). They created characters and joined guilds, participating in and observing the WoW community’s regular activities like questing for treasure (alone and in groups), raiding with guild-mates, and competing in player-versus-player combat. Having developed a rich understanding of the game from the inside through qualitative methods, they moved to a complementary quantitative approach. Using in-game features like the “/who” command listing characters currently being played on a given server, Ducheneaut and his team took frequent in-game censuses, which allowed them, for example, to systematically uncover the structure and life cycle of WoW guilds by using social network analysis metrics like centrality and density. Such metrics seem particularly appropriate in online environments, given our previous discussion of the work of Burrell (2009), who conceptualizes online spaces as networks of nested social relations. Techniques like Ducheneaut et al.’s (2006) use of WoW’s “/who” command are constrained according to the rules of particular online worlds (e.g., whether they allow players to expand and customize the world with their tinkering). Creating add-ons might also require technical expertise in software design, perhaps pushing qualitatively minded ethnographers to collaborate with researchers from other fields. Nevertheless, automated analyses are now common on the Internet, for example, in so-called sentiment analyses of digital texts, where researchers use machine learning to detect the moods and opinions of users in online environments (Pang and Lee 2008). These examples only hint at the ways systematic data collection methods can be integrated within an ethnographic study of online environments. Yee and Bailenson (2007) artificially manipulate users’ avatars to point to what they call the “Proteus effect”: the manner in which individuals infer their expected behaviors and attitudes by observing their avatars’ appearance. For example, giving users taller avatars led to more aggressive negotiating behaviors in laboratory experiments. Later, Yee et al. (2009) extended these experiments to an actual online community, finding similar results, with height and attractiveness of avatars predicting players’ performances, and such effects even being carried to subsequent face-to-face offline interactions. This work highlights the relatively untapped potential for using field-based experiments to expand ethnographic knowledge of the social experiences characteristic of online worlds, an experimental potential magnified by the enormous malleability of online environments and avatars. Ethnographers of life online now commonly integrate many other more systematic methods, both qualitative and quantitative, into more standard ethnographic projects of digital life. For example, Horst and Miller (2006) offer a richly textured ethnographic

14_107-Bernard.indb 480

6/9/14 10:03 AM

14: Ethnography of Online Cultures

481

account of cell phone use among low-income Jamaicans, in the process pointing to the methodological potential of such communications technologies for mobile society research. Likewise, Jones and Schieffelin (2009a, 2009b) provide a rich microanalysis of user-created meta-linguistic data related to instant messaging and pore over thousands of comments associated with YouTube videos, pointing to the potential for linguistic anthropological analysis of new media. Indeed, several recent studies investigate the biopsychology of Internet activity—assessing physiological states like blood volume pulse, skin conductance, respiratory response, and EEG patterns associated with Internet surfing and accessing Facebook accounts—leaving it up to creative researchers to decide how even biocultural data might enhance ethnographies of life online (Lu et al. 2010; Mauri et al. 2011). Overall, a variety of systematic methods now offer researchers relatively easy and cheap access to even ephemeral social behavioral and attitudinal data, saved and archived in relatively permanent ways and formats, which can be used to enhance the ethnography of online realities. Though beyond the reach of this chapter, it should be noted that the storage capabilities of the Internet have revolutionized historical inquiry. Materials previously found only in sprawling print archives or on aging magnetic tape are now held in compact digital form—readily retrievable from almost anywhere in the world, backed up multiple times and thus more effectively protected from theft as well as from fire, flood, and other damage. Given the obvious advantages related to data storage, management, access, and protection, most institutes have initiated the digitization of their historical collections, making them available via the Internet. Again, there is no single correct answer as to how to proceed here. Rather, when considering which, if any, additional qualitative and quantitative methods to incorporate into their ethnographies, researchers will need to think about their particular online world and research question, as well as their personal preferences, skills, and eventual audiences. Perhaps the best recommendation in a general sort of way is to mix the methods in an attempt to maximize the ethnographic possibilities for mapping online experience. Qualitatively minded ethnographers might balk at surveys, experiments, or other more structured methods, perhaps daunted by the process of quantitative analysis or simply doubtful of the quality of such data. But they should keep in mind that more systematic field methods can be used to both flesh out descriptions of social contexts— revealing patterns of association not easily visible through other means, and thus potentially generating more precise portraits of online social life—and also to test ideas generated in qualitative phases of research. Likewise, relying at least in part on quantifiable methods allows anthropologists to better communicate with scholars from fields like communications, psychology, and sociology, thus multiplying the potential impact of ethnographic research of online social activity. Indeed, ethnographers can collaborate with scholars in fields like these, where such methods are relatively standard. ANALYSIS OF DATA OBTAINED ONLINE

The Internet provides ethnographers with new opportunities to cheaply and easily access massive amounts of naturally occurring social data. Saving web data is often

14_107-Bernard.indb 481

6/9/14 10:03 AM

482

Jeffrey G. Snodgrass

as easy as the click of a mouse or the tap of the keyboard. Screen-save programs like Camtasia and Hypercam further facilitate archiving web data, from static screenshots to dynamic video. However, much of these data exist as free-floating and decontextualized behavioral traces, which were not directly witnessed or experienced by the ethnographer. Even when the behavior is witnessed first-hand, its meaning is not necessarily transparent, given that it might be informed by other online or offline contexts invisible to the ethnographer. Further, much online data are nonphysical in nature, divorced from the social and paralinguistic cues that might reveal their meaning, adding to the difficulty of ethnographic interpretation. The central challenge of ethnographically analyzing life online, then, is to reestablish the relevant contexts and underlying processes informing online thought and practice (Kozinets 2002, 2006a, 2006b). One way to respond to this challenge is to triangulate across multiple methods and sources of data. Indeed, this is one of the main benefits of using mixed methodological strategies in online contexts. The more disparate sources of information the ethnographer can collect on the online behavior in question, be it qualitative or quantitative, the more likely that researchers will arrive at valid and reliable explanations. Minimally, researchers need to analyze differently online data they have experienced or witnessed themselves first-hand as compared to that they have copied from online computer-mediated interactions. Observing and participating in a conversation unfolding in an online chat room is not the same as stumbling across the conversation’s screen data saved after the fact. The abundance of ready-made streams of Internet data may tempt ethnographers into avoiding the more time-consuming process of producing their own field notes and instead rely on downloads and archived data. However, such a move would interfere with the production of the detailed descriptions of local knowledge and practice—as well as the self-reflexive and personal interpretations of local contexts—that characterize good ethnography. The ethnographer’s first-hand field notes will be particularly useful in allowing researchers to chart their growing understandings of an online community. Researchers should be careful, however, to extrapolate conclusions drawn from their own personal experiences to situations they did not themselves witness. Researchers of online environments, then, would be well-advised to follow the prescriptions of manuals such as Writing Ethnographic Fieldnotes (Emerson 2011) to produce their own notes, from simple jottings and brief methodological observations, to more complex descriptions of recurring cultural scripts and scenes, to more deeply analytical reflections. Indeed, one best-of-both-worlds strategy would be to intersperse downloaded data directly into field notes, thus enhancing the detail of observations. The difficulty of making sense of online data will push many researchers of the Internet to conduct their analyses while continuing to collect new data rather than at the end of a study. Conducting analysis in a continuous way will focus ethnographers on their data collection, assuring that they save that which is necessary to address their most immediate ethnographic understandings, archiving the rest for future consideration. As Glaser and Strauss (1967) point out in their discussion of grounded theory, such an approach allows researchers to ask new questions as they advance in their analysis, a quality particularly valuable when piecing together the contexts relevant to an online reality. Researchers of life online could thus usefully follow Agar’s (1986, 20–32) con-

14_107-Bernard.indb 482

6/9/14 10:03 AM

14: Ethnography of Online Cultures

483

ception of ethnography as not addressing single and static hypotheses, but instead as characterized by a continual cycling deeper into social reality, as researchers develop and answer more focused and relevant questions one after another in ongoing sequential fashion. Indeed, data collection can usefully continue as long as new insights and questions are generated. By contrast, ethnographers who take the strategy of simply downloading everything of seeming interest, rather than more carefully integrating such data into ongoing interpretations, risk information overload. One benefit of online research in this regard is that it is relatively easy to use Internet media and communications technologies to recontact respondents for follow-up data, thus helping researchers clarify and expand on emerging understandings of online worlds. Analytical field notes in particular will help here, ensuring that both analysis and ongoing data collection stay on track and directed (for more on analytical strategies for textual data, see Wutich et al., this volume). More practically, ethnographers of life online should consider organizing field notes with qualitative data management and analysis software like QSR NVivo, Atlas.ti, or MAXQDA. Such software will also help researchers manage the immense quantities of data they extract from the Net, helping them catalogue, organize, and develop these alongside ongoing descriptions and emerging ideas. In mixed qualitative–quantitative research, the trade-down from richness to conceptual clarity provided by such software will help ethnographers identify potential synergies between qualitative and necessarily less rich quantitative data. As their experience grows, participant observer ethnographers will build up contextually sensitive cultural knowledge that renders their own bodies and persons valid research instruments. Slowly, they will learn to follow online data to the places that matter, holistically connecting social dots across unpredictably linked online and offline contexts, a key hallmark of good online ethnography. The self-reflexive knowledge generated by first-hand experiences and observations documented in field notes helps researchers interpret interview transcripts and survey data. Vice versa, more systematic methods like interviews, surveys, and experiments allow researchers to identify patterns of association among variables that might remain invisible to qualitative analysis alone. The potential for respondent deception and misrepresentation also poses challenges for analysis of online data. Ethnographers of online realities must be careful about accepting informant representations of themselves at face value. Researchers should be aware that they may have to limit their interpretations to online avatars rather than to the actual-world people behind the avatars, who often remain anonymous and invisible (Meadows 2008). Triangulating across multiple media and forms of data helps ethnographers link data obtained online to corresponding offline persons and contexts in order to verify that online respondents are who they say they are. For example, a Skype interview with a respondent previously only known through text-based chat— or, alternately, meeting someone known previously only online in their actual-world home—could provide an interesting ethnographic reality check. In these situations, ethnographers should be ready for even shocking changes in impressions of an online person and context as data from new sources is gathered, pointing again to the particularly emergent nature of understandings of life online. Of course, such

14_107-Bernard.indb 483

6/9/14 10:03 AM

484

Jeffrey G. Snodgrass

efforts require more time and thus should be pursued (or not) depending on the nature of one’s research questions and aims. For example, as alluded to earlier, Nardi (2010) is interested to trace WoW play to its offline sources, but Boellstorff (2008) is not. Problems related to identifying the actual-world persons behind the avatars, which theoretically could facilitate analysis, are exacerbated by the fact that profiles of online user communities are notoriously difficult to access. Malaby (2009) had relatively unrestricted access to Linden Lab and Second Life. And Williams and his collaborators collaborated with Sony Entertainment in their study of Everquest to the extent that Sony crafted and made uniquely available to survey respondents a digital sword (Williams et al. 2008). Still, for the sake of privacy and other reasons, companies or managers of online games, websites, and bulletin boards more commonly do not release details related to their clients or users to the general public or to academic researchers. Such lack of accessibility makes it difficult to intelligently sample online populations in ways that might allow researchers to generalize about them, which has been referred to as the “chief problem” for making sense of virtual worlds (Williams 2010, 2). In these situations, ethnographers must improvise, triangulating once again across different sources of online and offline data. For example, they can refer to unofficial baseline demographic surveys of relevant online populations such as Yee’s (2006) and Williams et al.’s (2008), which revealed MMO gamers to be generally older (mid- to late 20s), more educated, wealthier, and more socially connected than expected. Finally, researchers should keep in mind that each form of data collected in online environments requires its own analytical strategies. To micro-analyze online YouTube comments, for example, researchers might turn to linguistic or even literary critical theories. To quantitatively analyze surveys or to tease apart the network properties characteristic of guilds requires others bodies of thought. As in the offline world, there is thus no single analytical strategy appropriate to all Internet contexts. REPRESENTATION OF LIFE ONLINE

In classic ethnographies, the anthropologist’s immersion in a naturalistic setting leads him or her to produce a detailed Geertzian (1973) “thick description” of that social reality, to the extent that readers come to feel like they themselves experience the place. In an ethnography of life online, researchers can act similarly, but first they must decide on an audience, which determines the details presented and issues discussed. For example, Nardi (2010, 7) directed her description to those who had never played WoW but wanted to understand something about the role of online videogaming in U.S. and world culture. This led her to treat her ethnography as a form of “ethnographic reportage.” And she opened her book with detailed vignettes like “A Day in the Life of a Night Elf Priest,” “A Short WoW Primer,” “MMORPGs and Virtual Worlds,” and “Who Plays World of Warcraft,” to give outsiders to the game a better sense of this, from her audience’s point of view, strange and exotic virtual alternate reality. This led Nardi to focus in particular on the topics of gender and addiction, as she found these to be of particular interest to outsiders to the WoW community. Nardi also concentrated on collaborative group raids and guild activities. Here, she hoped to justify the social nature of this online world, pointing to the anthropological relevance of her study and allaying concerns that WoW is just a (silly) game.

14_107-Bernard.indb 484

6/9/14 10:03 AM

14: Ethnography of Online Cultures

485

Each researcher must tailor his or her descriptions to their ethnography’s aims and thus to their audience’s knowledge and concerns. Ethnographers need to think carefully about what will seem strange to this audience as opposed to commonplace and about their audience’s potential biases. Likewise, they will need to weigh carefully the balance they wish to strike between reaching a potentially popular and wider as opposed to narrower academic audience. Of course, the nature of the virtual world will come into play here. Certain online realities will be new and potentially limited to a small niche community of users. Anthropologists may follow Nardi in their presentation of details of such worlds. Others will be integrated more fully into popular culture, leading ethnographers to pursue a different descriptive strategy. Ethnographers have begun to tap into the multimedia potential of the Internet and other digital media—characterized as they are by rich kinetic, aural, and visual data—to give even broader audiences a fuller feel for anthropological research sites and subjects. For example, netnographer Kozinets (2006b) set up his own Star Trek research webpage, where he posted hundreds of digital photographs from his research. Cooper et al.’s (2009) photo-journalistic account of avatars and their creators is a powerful ethnographic document. And Yee has set up a few websites—the Daedalus project (http://www.nickyee.com/daedalus/), and more recently with his research collaborator, Ducheneaut, the previously mentioned PARC PlayOn 2.0—presenting his ongoing research to the public. These digital anthropological innovations point to presentational possibilities for ethnography beyond mere words and texts, and they illuminate the collaborative scholarly possibilities inherent to the Internet. Research blogs and websites like PlayOn 2.0 do not confine themselves to the simple posting and presentation of data. Instead, they are used to engage in consultation and critique, with both other scholars and also members of the communities being represented in a form of collaborative ethnography (Lassiter 2005; Murthy 2008, 847). Indeed, webcams are now commonly available in most households with computers, providing many users the possibility to upload video diary experiences to the web, potentially collaboratively contributing to a YouTube model of the presentation of scholarly research. Such uploads give audiences the chance to access richer and more spontaneous naturalistic data, enhancing ethnography’s presentational potential (Murthy 2008, 843). New web-based technologies, then, blur the line between data collection, analysis, and presentation. Websites like the Daedalus Project and PARC PlayOn 1.0 and 2.0 allow researchers to recruit participants for various online surveys, and comments on this website themselves become a source of data, thus enhancing what we might call data collection. However, dialogue on this site is not just data but also analysis, with web visitors collaboratively working with researchers to understand the nature of, in this case, the psychology and society of WoW and other MMOs. Here, researchers and community members work together to interpret complex and even overwhelming amounts of online data. Further, the website as a whole, as a collaborative project, provides researchers with a way to present their research results to an unusually wide audience. Indeed, ethically, collaborating with online community members, thus allowing them to participate in the research, will help ensure that the research is of practical importance to the communities in question. Getting such a mandate from

14_107-Bernard.indb 485

6/9/14 10:03 AM

486

Jeffrey G. Snodgrass

a community encourages researchers to not just conduct pure inquiries, but rather to pursue issues that advance both scientific questions and the needs and concerns of community members. Kozinets (2002) believes that the new representational and collaborative potential of the Internet enhances ethnographic accountability. Researchers post results for consideration by insiders to the communities in question, who provide a member check. Indeed, scholars are able to post results in a particularly sensorially rich manner (e.g., if they use videologs and relatively unedited transcriptions of member activities), allowing audience members to track more carefully the path from relatively raw data to interpretation. This promotes critique and thus potentially allows ethnographers to correct mistakes, refine arguments, and clarify issues needing further investigation. My own WoW research, for instance, unfolds in the context of a collaborative ethnographic research and teaching laboratory that I lead. I include gamers as members of my research team. Some of these are students, who advance in their understanding of research methods, as they contribute to the study. Others are members and guild leaders of the communities being studied but not students. The 1980s critique of the representational authority of anthropology drew attention to the Greek etymological roots of the word “ethnography”: literally, “writing” (graphos) “culture” (ethnos) (Clifford and Marcus 1986; van Maanen 1988). Taking this critique to heart, many contemporary anthropologists have been interested in representing ethnographic knowledge in ways that are more collaborative, dialogical, and thus ultimately also more valid. The vast proliferation of relatively standard text-based ethnographies is still dominated by the voice of solitary data-collecting anthropologists. However, the possibilities of new digital technologies, with their sensorially rich data and collaborative potential, might succeed where these earlier efforts failed. THE ETHICS OF RESEARCHING LIFE ONLINE

In earlier sections of this chapter, I pointed to the potential benefits for researchers of conducting non–participating observation. In addition to being easy and inexpensive to carry out, non–participating observation is less obtrusive, thus potentially producing particularly naturalistic observations. Incorporating at least an early phase of non–participating observation can also help researchers choose online communities appropriate to their research. Such a research phase can help ethnographers further refine both their research questions and also their behavior in ways that might help them eventually become more readily accepted into particular online communities. Still, it is important for ethnographers of life online to ask not only how such practices benefit themselves as researchers, but also how they benefit—or potentially harm—their research subjects. This issue serves as a starting point for reflecting on the particular ethical challenges of conducting research in online environments. To begin, ethnographers of life online have unprecedented access to information whose creation they themselves did not witness or experience, which creates challenges in and around obtaining informed consent to use that material for research purposes. Much Internet data, then, were not given in confidence to a particular researcher, and its creators might not welcome its use for research purposes. It has been argued that data posted to an Internet forum is technically public, given its ready availability to anyone

14_107-Bernard.indb 486

6/9/14 10:03 AM

14: Ethnography of Online Cultures

487

who might wish to access it, and, as such, informed consent might be considered implicit in the act of posting to online electronic spaces (Rafaeli et al. 1998). My own Institutional Review Board (IRB) suggests in its guidelines on research in so-called virtual worlds that non–participating observations of online activity that do not include active interaction with other avatars are public observations and thus not technically “human subjects research” needing informed consent (Martey and CSU IRB 2012). Nevertheless, ideas about the public nature of the Internet have been contested (e.g., King 1996), with some even comparing non–participating observation to electronic eavesdropping and thus full of moral and ethical pitfalls (Clegg Smith 2004; Sveningsson 2004). Prominent researchers such as Kozinets (2002) recommend a cautionary approach to non–participating observation and also to labeling Internet data public. Given the contested nature of the public or private status of much Internet material, I, too, recommend a cautionary approach aimed to minimize potential harm to research subjects. For example, ethnographers should consider allowing local understandings of what constitutes public as opposed to private online exchange to shape their decisions about how to handle particular forms of data (e.g., see Boellstorff et al. 2012, Chapter 8). Such a cautionary approach would also consider avatar names and emails, even those stumbled across in seemingly public Internet spaces, to be personal identifiers. As such, they should be kept confidential, and researchers should seek permission if they wish to use them in publications. In most cases, researchers should simply anonymize them as they would real participant names, unless there was some pressing reason not to do so, such as respecting respondents’ wishes to attach their names to creative online artifacts like blogs, stories, newsletters, poetry, photos, and gaming accomplishments. For example, researchers can easily track conversations in public Internet spaces, taking notes and even recording them as do I with Ventrilo, through WoW’s “/chatlog” command, or when players broadcast their raids over the Internet so that I might observe them. Here, it is not possible to avoid recording avatar names, but they can be easily deleted, replaced, or anonymized before analysis or publication. Similarly, researchers might consider de-linking particular user names to particular websites or other online spaces. As in the offline world, revealing personal or other sensitive information about a member known to an online group can potentially lead that individual to experience humiliation and ostracism. Properly anonymizing this information in public discussions of it, especially data judged to be potentially compromising, avoids situations such as this. In participatory situations where researchers actively interact with others in online environments, ethical standards approximate more closely their actual-world counterparts (Paccagnella 1997). This helps explain why IRBs will typically allow for verbal consent in online situations involving direct interactions with research respondents, or even waive the necessity for consent in certain situations, given that anonymous and transitory interactions, which characterize many online interactions, are treated similarly in offline contexts. The potential for deception in online environments makes it difficult to exclude interactions with minors if such an action is deemed necessary for ethical and/or

14_107-Bernard.indb 487

6/9/14 10:03 AM

488

Jeffrey G. Snodgrass

research reasons. But ethnographers might consider asking multiple age-related questions—carefully judging speakers’ response time and tone—to better assess if one is interacting with a minor. Boellstorff et al. (2012, 156–58) note that an IRB once requested of one of them to include minors in a study in which they were originally excluded, to improve the quality of the research. This shows that minors should not necessarily be excluded from a study of life online, and in some cases, such exclusion might infringe both on minors’ rights to be heard and on the quality of research in settings where minors can play important roles. In some situations involving direct interactions with research subjects, it may be appropriate to seek permissions to conduct research from online figures of authority like bulletin board or listserv moderators, website creators, or MMO guild leaders. Researchers may even consider informing creators of online worlds about their research, especially given the added potential for collaboration with overseers of online worlds. Still, some companies like Blizzard Entertainment, the creators of WoW, have shown little interest in interacting with the scholarly community, and IRBs have consistently been amenable to not requiring such permissions, given that such a requirement would make certain research projects unfeasible (Boellstorff et al. 2012, Chapter 8). As in offline situations, researchers should ask permission to record conversations with their online respondents. The Internet offers additional recording possibilities, as through computer-to-computer programs (like Ventrilo or Skype) or chat-logging technology. Researchers should inform respondents about the technology being used, noting that collecting video, for example, may require additional permissions. Researchers might also consider investigating the security of their recording technologies. Skype, for example, is very secure in its transmissions, with the main risks to respondents being address lists being maintained on investigators’ unsecured computers. Some online situations will place researchers in positions where they are alternately non–participating observers accessing potentially public data and participants speaking and interacting with online denizens. For example, in some MMOs, there are literally thousands of players engaged in multifold tasks, many of which do not directly involve or even concern the researcher. In these situations, it is not desirable or even possible for researchers to introduce themselves and state their purposes to every individual in that environment. Attempting to explain one’s research aims to all these individuals would, at the least, disrupt the very activities the ethnographer wished to understand, annoying players and potential respondents in the process. Nevertheless, ethnographers can strive in similar online contexts to explain their purposes to individuals when the interaction starts to feel more intimate, as when an online avatar begins to share personal details about their offline lives. This will avoid potential misunderstandings and fallouts with community members, if these members were to find out and disapprove of one’s status as an ethnographer. Researchers can also relieve the felt obligation to continually introduce themselves by posting explanations of their research on relevant online sites, thus making information about their research available to relevant audiences. Finally, active deception, too, has been justified as a valid research strategy in online research. For example, in Yee and Bailenson’s (2007) Proteus effect experiments, some degree of deception is necessary to ensure the integrity of their experiments. Boellstorff

14_107-Bernard.indb 488

6/9/14 10:03 AM

14: Ethnography of Online Cultures

489

et al. (2012) suggest that experiments of any kind, and deception more generally, are contrary to an ethnographic project, which necessarily involves the open and honest cultivation of relationships with study participants (see, e.g., 90–91, 136–42). However, behavioral experiments, and indeed the testing of narrowly defined hypotheses through the controlled manipulation of variables of interest, are increasingly being shown to be useful in illuminating the workings of local culture (and potentially a broader human nature both shaping and shaped by local culture) (e.g., see Henrich et al. 2005; Luhrmann 2012a, 2102b; Ross and Medin 2005). Luhrmann, for instance, uses experiments to show how particular forms of prayer impact Evangelical Christians’ spiritual experience. Her research, in particular, demonstrates how experimental evidence can complement participant observation and other methods, allowing ethnographers to develop more complete understandings of local culture. I suggest that behavioral experiments, if properly contextualized within local thought and practice, can likewise form important pieces of a mixed methodology toolkit for revealing dimensions of online experience and social relations. Still, I once again recommend caution in using such experiments. The risks that such deception might harm respondents should be carefully weighed against anticipated research gains, and all planned deceptions should be carefully vetted with relevant IRBs. Proper post-experimental debriefing—i.e., explaining how the experiment contributes to larger research objectives—would also be necessary to ensure that ongoing relationships with respondents remain good and healthy, that is, not excessively exceeding the little deceptions and tactics that characterize social relations in all times and places. CONCLUSION

Ethnographically studying life online offers many advantages that resemble those of traditional offline ethnography. By immersing themselves in naturalistic community contexts, anthropologists come to appreciate, even deep down in their own psyches and bodies, the meaning and reality of social structures and processes as they actually unfold. Observing what people would be more or less doing even in the absence of the researcher also allows research questions to unfold more naturally from the social contexts in question, freer from researchers’ personal and cultural biases. Such inquiry helps ethnographers appreciate how even unanticipated dimensions of social existence might fit together, each part helping explain the social whole. Ethnographers, too, are able to build intimacy and trust with their respondents, leading to potentially more valid knowledge. Additionally, conducting ethnography of life online offers unique advantages. For example, it can be performed relatively quickly and cheaply compared to offline counterparts. As we see in Nardi’s (2010) study of WoW, just $50 for the game, $14 more for the subscription, a decent computer, and an Internet connection launched her from the comfort of her home into an exotic alternate online reality. And the new presentational possibilities of the Internet, which might potentially blur the boundary between researchers and their subject, point to a potentially even more democratic anthropology. Still, ethnographers of life online face particular challenges. Online ethnographer data collectors are digitally encoded and thus peculiarly disembodied, leading to

14_107-Bernard.indb 489

6/9/14 10:03 AM

490

Jeffrey G. Snodgrass

unique challenges in presenting one’s researcher self to respondents. Also, it is not always clear even where to conduct research, given the way online communities are typically rooted both on the Internet and also in the offline world. This raises analytical challenges as well, as ethnographers struggle to identify the meaningful contexts within which to situate their observations. Here, the abundance of easily downloadable online data poses its own challenges. If netnographers are not careful, they will lose the full richness and detail of online social life, which makes more traditional ethnographies so compelling. Often, ethnographers of life online will need to collect data in an ongoing way from even very faraway locales, some online, others offline. And the lack of informant identifiers, and thus the possibility for deception, makes it hard to generalize outside of one’s own study context. Further, ethnographers of life online can utilize digital media to present their work in new ways that blur the line between researcher and subject. Here, if there are new presentational possibilities, these are met by both technological and also new ethical challenges. Given the challenges of ethnographically documenting life online, this chapter has emphasized the need for a complex mixed methods toolkit. This allows researchers to triangulate across convergent data from multiple sources, be they observations, interviews, surveys, or other forms of online and offline data. Such a toolkit also allows ethnographers to meet the specific challenges of their particular research site and question, disciplinary or subdisciplinary norms, intended audience, available time and resources, and, indeed, their own preferences and passions. The key, then, is adaptability on all levels of a project, from data collection, to analysis, to presentation. Traditional ethnographers must be adaptable, choosing from a rich and varied toolkit. Ethnographers of life online must be even more flexible, adapting an already diverse traditional toolkit to the new demands of dynamic online environments (Kozinets 2002, 2006a, 2006b). Finally, the spread of computer-mediated technologies and interactions points to a future where most social action will be mediated at least in part by computer technologies and the Internet. As such, refining the principles discussed here will ensure the quality of ethnography conducted online and, indeed, almost anywhere else. REFERENCES

Adams, T. L., and S. A. Smith, eds. 2008. Electronic tribes: The virtual worlds of geeks, gamers, shamans, and scammers. Austin: University of Texas Press. Agar, M. 1980. The professional stranger: An informal introduction to ethnography. New York: Elsevier. Agar, M. 1986. Speaking of ethnography. London: Sage. Arnason, L. 2011. New media, new research methods: Current approaches to research in the virtual world. Media International Australia 138: 98–111. Aunger, R. 2004. Reflexive ethnographic science. Lanham, MD: AltaMira. Bainbridge, W. S. 2010. The warcraft civilization: Social science in a virtual world. Boston: MIT Press. Beavis, C. 2005. Pretty good for a girl: Gender, identity and computer games. In Changing views: Worlds in play: Selected papers of the 2005 Digital Games Research Association’s Second International Conference, ed. S. de Castell and J. Jensen, 339–50. Tampere, Finland: DiGRA. Boellstorff, T. 2008. Coming of age in Second Life: An anthropologist explores the virtually human. Princeton, NJ: Princeton University Press.

14_107-Bernard.indb 490

6/9/14 10:03 AM

14: Ethnography of Online Cultures

491

Boellstorff, T., B. Nardi, C. Pearce, and T. L. Taylor. 2012. Ethnography and virtual worlds: A handbook of method. Princeton, NJ: Princeton University Press. Brewer, J., and A. Hunter. 2006. Foundations of multimethod research: Synthesizing styles. Thousand Oaks, CA: Sage. Burrell, J. 2009. The field site as a network: A strategy for locating ethnographic research. Field Methods 21: 181–99. Castells, M. 1996. The rise of the network society. Cambridge, MA: Blackwell. Castronova, E. 2005. Synthetic worlds: The business and culture of online games. Chicago: University of Chicago Press. Chen, C., C. Sun, and J. Hsieh. 2008. Player guild dynamics and evolution in massively multiplayer online games. CyberPsychology & Behavior 11: 293–301. Chen, M. 2009. Communication, coordination, and camaraderie in World of Warcraft. Games and Culture 4: 47–73. Clegg Smith, K. M. 2004. Electronic eavesdropping: The ethical issues involved in conducting a virtual ethnography. In Online social research: Methods, issues, and ethics, ed. M. D. Johns, S. L. S. Chen, and G. J. Hall, 223–38. New York: Peter Lang. Clifford, J., and G. E. Marcus, eds. 1986. Writing culture: The poetics and politics of ethnography. Berkeley: University of California Press. Coleman, E. G. 2009. Code is speech: Legal tinkering, expertise, and protest among free and open source software developers. Cultural Anthropology 24: 420–54. Coleman, E. G. 2010a. The Hacker Conference: A ritual condensation and celebration of a lifeworld. Anthropological Quarterly 83: 99–124. Coleman, E. G. 2010b. Ethnographic approaches to digital media. Annual Reviews in Anthropology 39: 487–505. Cooper, R., J. Dibbell, and T. Spaight. 2009. Alter ego: Avatars and their creators. New York: Chris Boot. Corneliussen, H. G., and J. W. Rettberg, eds. 2008. Digital culture, play, and identity: A World of Warcraft reader. Boston: MIT Press. Doostdar, A. 2004. The vulgar spirit of blogging: On language, culture, and power in Persian Weblogestan. American Anthropologist 106: 651–62. Ducheneaut, N, and R. J. Moore. 2004. Social side of gaming: A study of interaction patterns in a massively multiplayer online game. Proceedings of the 2004 ACM Conference on Computer Supported Cooperative Work (CSCW04). New York: ACM Press. Ducheneaut, N., N. Yee, E. Nickell, and R. J. Moore. 2006. “Alone together?”: Exploring the social dynamics of massively multiplayer online games. In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 407–16. New York: ACM. doi:10.1145/1124772.1124834. Emerson, R. M. 2011. Writing ethnographic fieldnotes, 2nd ed. Chicago: University of Chicago Press. Fischer, M. 2007. Four genealogies for a recombinant anthropology of science and technology. Cultural Anthropology 22: 539–615. Freeman C. 2000. High tech and high heels in the global economy: Women, work and pink-collar identities. Durham, NC: Duke University Press. Garcia, A. C., A. I. Standlee, J. Bechkoff, and Y. Cui. 2009. Ethnographic approaches to the Internet and computer-mediated communication. Journal of Contemporary Ethnography 38: 52–84. Geertz, C. 1973. The interpretation of cultures. New York: Basic Books. Gershon I. 2010. The break-up 2.0: Disconnecting over new media. Ithaca, NY: Cornell University Press.

14_107-Bernard.indb 491

6/9/14 10:03 AM

492

Jeffrey G. Snodgrass

Glaser, B., and A. Strauss. 1967. The discovery of grounded theory: Strategies for qualitative research. Piscataway, NJ: Aldine Transaction. Golub, A. 2010. Being in the World (of Warcraft): Raiding, realism, and knowledge production in a massively multiplayer online game. Anthropological Quarterly 83: 17–45. Gravlee, C. C., H. R. Bernard, C. R. Maxwell, and A. Jacobsohn. 2012. Mode effects in free-list elicitation: Comparing oral, written, and web-based data collection. Social Science Computer Review. doi: 10.1177/0894439312455312. Gunter, B., D. Nicholas, P. Huntington, and P. Williams. 2002. Online versus offline research: Implications for evaluating digital media. Aslib Proceedings 54: 229–39. Gupta, A., and J. Ferguson. 1997. Discipline and practice: “The field” as site, method, and location in anthropology. In Anthropological locations: Boundaries and grounds of a field science, ed. A. Gupta and J. Ferguson, 1–46. Berkeley: University of California Press. Handwerker, W. P. 2001. Quick ethnography. Lanham, MD: AltaMira. Haraway, D. 1991. A cyborg manifesto: Science, technology, and socialist-feminism in the late twentieth century. In Simians, cyborgs and women: The reinvention of nature, 149–81. New York: Routledge. Hayles, N. K. 1999. How we became posthuman: Virtual bodies in cybernetics, literature, and informatics. Chicago: University of Chicago Press. Henrich, J., R. Boyd, S. Bowles et al. 2005. “Economic man” in cross-cultural perspective: Behavioral experiments in 15 small-scale societies. Behavioral and Brain Sciences 28: 795–855. Hindman, M. S. 2009. The myth of digital democracy. Princeton, NJ: Princeton University Press. Hine, C. 2000. Virtual ethnography. London: Sage. Hine, C. 2005. Virtual methods: Issues in social research on the Internet. Oxford: Berg Horst, A. H., and D. Miller. 2006. The cell phone: An anthropology of communication. New York: Berg. Jones, G., and B. Schieffelin. 2009a. Enquoting voices, accomplishing talk: Uses of be + like in instant messaging. Language Communication 29: 77–113. Jones, G., and B. Schieffelin. 2009b. Talking text and talking back: “My BFF Jill” from boob tube to YouTube. Journal of Computer-Mediated Communication 14: 1050–79. Jupp, V. 2006. Netnography. In The Sage dictionary of social research, ed. V. Jupp, 193–95. London: Sage. Juris, J. S. 2012. Reflections on #Occupy everywhere: Social media, public space, and emerging logics of aggregation. American Ethnologist 39: 259–79. King, S. A. 1996. Researching Internet communities: Proposed ethical guidelines for the reporting of results. Information Society 12: 119–27. Kosminsky, E. 2010. World of Warcraft: The viability of massively multiplayer online roleplaying games as platforms for modeling and evaluating perfect competition. Journal of Virtual Worlds Research 2: 4–16. Kozinets, R. V. 2002. The field behind the screen: Using netnography for marketing research in online communities. Journal of Marketing Research 39: 61–72. Kozinets, R. V. 2006a. Netnography 2.0. In Handbook of qualitative research methods in marketing, ed. Russell W. Belk, 129–42. Cheltenham, UK: Edward Elgar Publishing. Kozinets, R. V. 2006b. Click to connect: Netnography and tribal advertising. Journal of Advertising Research 46: 279–88. Kozinets, R. V., and J. Handelman. 1998. Ensouling consumption: A netnographic exploration of the meaning of boycotting behavior. Advances in Consumer Research 25: 475–80. Landzelius K, ed. 2006. Native on the net: Indigenous and diasporic peoples in the virtual age. London: Routledge.

14_107-Bernard.indb 492

6/9/14 10:03 AM

14: Ethnography of Online Cultures

493

Lassiter, L. E. 2005 The Chicago guide to collaborative ethnography. Chicago: University of Chicago Press. Lave, J., and E. Wenger. 1991. Situated learning: Legitimate peripheral participation. Cambridge: Cambridge University Press. LeBesco, K. 2004. Managing visibility, intimacy, and focus in online critical ethnography. In Online social research: Methods, issues, and ethics, ed. M. D. Johns, S.-L. S. Chen, and G. J. Hall, 63–79. New York: Peter Lang. Leece, P., M. Bhandari, S. Sprague, M. Swiontkowski, E. Schemitsch, P. Tornetta, P. Devereaux, and G. Guyatt. 2004. Internet versus mailed questionnaires: A randomized comparison. Journal of Medical Internet Research 6: e30. Lu, D. W., J. W. Wang, and A. C. H. Huang. 2011. Differentiation of Internet addiction risk level based on autonomic nervous responses: The Internet-addiction hypothesis of autonomic activity. Cyberpsychology, Behavior, and Social Networking 13: 371–78. Luhrmann, T. M. 2012a. When God talks back: Understanding the American evangelical relationship with God. New York: Alfred A. Knopf. Luhrmann, T. M. 2012b. Prayer as inner sense cultivation: An attentional learning theory of spiritual experience. Ethos 40: 359–89. Magnet, S. 2007. Feminist sexualities, race, and the Internet: An investigation of suicidegirls. com. New Media & Society 9: 577–602. Malaby, T. 2009. Making virtual worlds: Linden Lab and Second Life. Ithaca, NY: Cornell University Press. Malinowski, B. 1922. Argonauts of the western Pacific. New York: E. P. Dutton. Markham, A. N. 1998. Life online: Researching real experience in virtual space. Lanham, MD: AltaMira. Markham, A. N., and N. K. Baym. 2009. Internet inquiry: Conversations about method. Los Angeles: Sage. Martey, R., and Colorado State University Institutional Review Board. 2012. Research in a virtual world: Guidance for CSU researchers. http://ricro.colostate.edu/IRB/Documents/ VirtualWorld_Guidelines.2.21.11_000.pdf. (accessed November 13, 2012). Mauri, M., P. Cipresso, A. Balgera, M. Villamira, and G. Riva. 2011. Why is Facebook so successful? Psychophysiological measures describe a core flow state while using Facebook. Cyberpsychology, Behavior, and Social Networking. doi: 10.1089/cyber.2010.0377. Meadows, M. S. 2008. I, avatar: The culture and consequences of having a second life. New York: New Riders Press. Miller, D., and D. Slater. 2000. The Internet: An ethnographic approach. Oxford: Berg. Mosco, V. 2004. The digital sublime: Myth, power, and cyberspace. Cambridge, MA: MIT Press Moss, J., and G. Hendry. 2002. Use of electronic surveys in course evaluation. British Journal of Educational Technology 33: 583–92. Murthy, D. 2008. Digital ethnography: An examination of the use of new technologies for social research. Sociology 42: 837–55. Nardi, B. A. 2010. My life as a night elf priest: An anthropological account of World of Warcraft. Ann Arbor: University of Michigan Press. Negroponte, N. 1996. Being digital. New York: Vintage Books O’Connor, H., and C. Madge. 2003. Focus groups in cyberspace: Using the Internet for qualitative research. Qualitative Market Research: An International Journal 6: 133–43. Paccagnella, L. 1997. Getting the seats of your pants dirty: Strategies for ethnographic research on virtual communities. Journal of Computer-Mediated Communication 3. doi:  10.1111/ j.1083-6101.1997.tb00065.x.

14_107-Bernard.indb 493

6/9/14 10:03 AM

494

Jeffrey G. Snodgrass

Pang, B., and L. Lee. 2008. Opinion mining and sentiment analysis. Foundations and Trends in Information Retrieval 2: 1–135. Pearce, C., and Artemesia. 2009. Communities of play: Emergent cultures in multiplayer games and virtual worlds. Boston: MIT Press. Rafaeli, S., F. Sudweeks, J. Konstan, and E. Mabry. 1998. Project!-!: A collaborative quantitative study of computer-mediated communication. In Network and netplay: Virtual groups on the Internet, ed. F. Sudweeks, M. McLaughlin, and S. Rafaeli, 265–81. Cambridge, MA: MIT Press. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–18. Ross, A. 2003. No-collar: The humane workplace and its hidden costs. New York: Basic Books. Ross, N., and D. Medin. 2005. Ethnography and experiments: Cultural models and expertise effects elicited with experimental research techniques. Field Methods 17: 131–49. Senft, T. 2008. Camgirls: Celebrity and community in the age of social networks. New York: Peter Lang. Smith, M. A., and P. Kollock. 1999. Communities in cyberspace. New York: Routledge. Snodgrass, J. G. 2006. Casting kings: Bards and Indian modernity. Oxford: Oxford University Press. Snodgrass, J. G., H. J. F. Dengah, II, M. G. Lacy, J. Fagan, and D. Most. 2011a. Magical flight and monstrous stress: Technologies of absorption and mental wellness in Azeroth. Culture, Medicine, and Psychiatry 35: 26–62. Snodgrass, J. G., M. G. Lacy, H. J. F. Dengah, II, and J. Fagan. 2011b. Enhancing one life rather than living two: Playing MMOs with offline friends. Computers in Human Behavior 27: 1211–22. Snodgrass, J. G., H. J. F. Dengah, II, M. G. Lacy, and J. Fagan, J. 2011c. Cultural consonance and mental wellness in the World of Warcraft: Online games as cognitive technologies of “absorption-immersion.” Cognitive Technology 16: 11–23. Snodgrass, J. G., H. J. F. Dengah II, M. G. Lacy, J. Fagan, D. Most et al. 2012. Restorative magical adventure or warcrack?: Motivated MMO Play and the pleasures and perils of online experience. Games and Culture 7: 3–28 Snodgrass, J. G., H. J. F. Dengah II, M. G. Lacy, and J. Fagan. 2013. An anthropological view of “motivation” models of problematic MMO play: Achievement, social, and immersion factors in the context of culture. Transcultural Psychiatry 50: 235–62. Soukup C. 1999. The gendered interactional patterns of computer-mediated chatrooms: A critical ethnographic study. The Information Society 15: 169–176. Spradley, J. P. 1979. The ethnographic interview. New York: Holt, Rinehart, and Winston. Spradley, J. P . 1980. Participant observation. New York: Holt, Rinehart, and Winston. Sreberny, A., and G. Khiabany. 2010. Blogistan: The Internet and politics in Iran. London: Tauris. Steinkuehler, C. A., and D. Williams. 2006. Where everybody knows your (screen) name: Online games as “third places.” Journal of Computer-Mediated Communication 11: 885–909. Stewart, K., and M. Williams. 2005. Researching online populations: The use of online focus groups for social research. Qualitative Research 5: 395–416. Sveningsson, M. 2004. Ethics in Internet ethnography. In Virtual research ethics: Issues and controversies, ed. E. A. Buchanan, 45–61. Hershey, PA: Information Science Publishing. Sweet, C. 2001. Designing and conducting virtual focus groups. Qualitative Market Research: An International Journal 4: 130–35. Taylor, T. L. 2006. Play between worlds: Exploring online game culture. Cambridge, MA: MIT Press.

14_107-Bernard.indb 494

6/9/14 10:03 AM

14: Ethnography of Online Cultures

495

Turkle S. 1995. Life on the screen: Identity in the age of the Internet. New York: Simon and Schuster. Turkle S. 2011. Alone together: Why we expect more from technology and less from each other. New York: Basic Books. van Maanen, J. 1988. Tales of the field: On writing ethnography. Chicago: University of Chicago Press. Walstrom, M. K. 2004. Ethics and engagement in communication scholarship: Analyzing public, online support groups as researcher/participant-experiencer. In Virtual research ethics: Issues and controversies, ed. E. A. Buchanan, 174–202. Hershey, PA: InformationScience Publishing. Warschauer, M. 2004. Technology and social inclusion: Rethinking the digital divide. Cambridge, MA: MIT Press. Williams, D. 2010. The promises and perils of large-scale data extractions. Chicago: MacArthur Foundation. http://dmitriwilliams.com/research.html (accessed February 19, 2013). Williams, D., N. Ducheneaut, L. Xiong, Y. Zhang, N. Yee, and E. Nickell. 2006. From tree house to barracks: The social life of guilds in World of Warcraft. Games and Culture 1: 338–61. Williams, D., T. Kennedy, and R. Moore. 2011. Behind the avatar: The patterns, practices and functions of role playing in MMOs. Games and Culture 6: 171–200. Williams, D., N. Yee, and S. E. Caplan. 2008. Who plays, how much, and why? Debunking the stereotypical gamer profile. Journal of Computer-Mediated Communication 13: 993–1018. Wilson, S. M., and L. C. Peterson. 2002. The anthropology of online communities. Annual Review of Anthropology 31: 449–67. Wutich, A., G. Ryan, and H. R. Bernard. 2013. Text analysis. In Handbook of methods in cultural anthropology, 2nd ed., ed. C. C. Gravlee and H. R. Bernard. Lanham, MD: AltaMira. Yee, N. 2006. The demographics, motivations and derived experiences of users of massively multiuser online graphical environments. Teleoperators and Virtual Environments 15: 309–29. Yee, N. 2007. Motivations of play in online games. Journal of Cyberpsychology and Behavior 9:772–75. Yee, N., and J. N. Bailenson. 2007. The Proteus effect: The effect of transformed self-representation on behavior. Human Communication Research 33: 271–90. Yee, N., J. N. Bailenson, and N. Ducheneaut. 2009. The Proteus effect: Implications of transformed digital self-representation on online and offline behavior. Communication Research 36: 285–312.

14_107-Bernard.indb 495

6/9/14 10:03 AM

14_107-Bernard.indb 496

6/9/14 10:03 AM

CHAPTER 15

Social Survey Methods William W. Dressler and Kathryn S. Oths

Social surveys are widely employed in the social and public health sciences, being the research method of choice in much of sociology, political science, epidemiology, health education, communications, and public opinion polling, to name a few disciplines. Anthropology is associated primarily with long-term ethnographic research utilizing participant observation and key-informant interviewing as the methods of choice. Nevertheless, there are very good reasons for using social survey methods in anthropological fieldwork, and we want to underscore that phrase as the theme for this chapter. If the reader wants to study social survey methods per se, there is an enormous number of articles and texts from which to learn (e.g., Dillman 2007; Fink 1995, 2006; Nardi 2006; Salant and Dillman 1994). Our focus is somewhat different. We want to address the following question: How can social survey methods help enrich the results of anthropological fieldwork? No doubt, there are many answers to this question, but at least three aims of anthropological research can clearly be enhanced by incorporating social survey methods. First, ever since Sapir’s (1938) reflections on a key informant’s insistence that what the ethnographer heard from other informants was wrong, some anthropologists have argued that culture theory and related research methods must be able to account for and investigate systematic diversity within a society (Pelto and Pelto 1975). Intracultural diversity can be studied in a variety of ways, with one obvious way being to enumerate variability in beliefs and practices within a population using social survey research methods. The data from such efforts enable the ethnographer to specify the confidence with which descriptive statements about those beliefs and practices can be offered. Second, there is a tradition within anthropology of generating hypotheses to explain beliefs and practices and to construct causal models. Inferential statistics provide a logic for evaluating the confidence with which one can assert that two (or more) phenomena are associated and hence may be causally linked in a system. Applying statistics requires collection of systematic quantitative data, and survey methods are one useful way of generating those data. Third, survey research provides a sounder basis for generalizing about a population. Put differently, if a study’s focus is a precise segment of intracultural diversity, survey research enables the ethnographer to locate that segment in the range of diversity. Anthropologists were long ago at the forefront of calling for a qualitative–quantitative mix in ethnographic research (Pelto and Pelto 1975). Since then, there has been 497

14_107-Bernard.indb 497

6/9/14 10:03 AM

498

William W. Dressl er and Kathryn S. Oths

considerable movement within all of the fields earlier cited in the direction of what is now called mixed methods research (Tashakkori and Teddlie 2003), in recognition that the goals of social science research are complex and cannot be achieved by relying solely on one way of knowing. Although there has been progress, this mix, however conceived, has not been as systematic as it might, or should, be. Our aim here is to help anthropologists think about adding social survey methods to the mix in ethnographic fieldwork in a more systematic way. We are not proposing something totally new. As Bennett and Thaiss (1973) noted in their review of social survey methods, Malinowski (1922), in Argonauts of the Western Pacific, urged ethnographers to systematically survey their study communities—to document population size, age and sex distribution, household structure, and other basic sociodemographic variables. Systematic surveys of these and other types were a staple of British social anthropology, especially in Africa in the 1930s (Richards 1935). Fast-forward to the 1960s, with the gradual availability of high-speed computing and standard statistical analysis packages, and anthropologists began to add social survey methods to their fieldwork to test complex hypotheses (Edgerton 1971; Graves 1967). With the more recent advent of portable computing, the sophistication of the application of survey methods has only grown. SOCIAL SURVEY METHODS: AN OVERVIEW

In the following sections, we first present a basic overview of the steps involved in doing social survey research. Second, we present some special issues involved in applying social survey methods in anthropological fieldwork and suggested resolutions to the problems involved.

Formulating Hypotheses In some studies, the ultimate goal may be to carry out a survey that describes the distribution of several variables within a population. More likely, however, the ethnographer wants to test a hypothesis, which is a testable prediction about the relationship between two variables. More often, a set of hypotheses, rather than a single hypothesis, will be proposed describing part of a causal system. Hypotheses are statements that take the form “If A, then B; If not A then not B” at their core. For example, a statement like “As levels of social support increase, expressed depressive symptoms will decrease” is such a hypothesis. Another formula, perhaps more user friendly than the “If A then B” statement, is to name two variables (such as one defining at least two groups as values [e.g., variable = gender; groups: 1=female, 2=male] and the other the primary notion of interest [e.g., the age at which a child begins to help with household tasks]) and state the expected relationship between them (e.g., girls tend to assume duties younger than do boys). The elements of a good hypothesis to include when building it are: (1) the population (e.g., Honduran mestizo peasants); (2) a one-sentence maximum length; (3) use of the plural form (e.g., “girls,” not “a girl”); (4) a completed comparison (e.g., make sure to add “than boys”); and (5) list the variables in the order they occur (e.g., girls → duties → age) (see Pyrczak and Bruce 2011). A directional hypothesis is preferable to one only predicting a difference. These building blocks result in a hypothesis: “Among

14_107-Bernard.indb 498

6/9/14 10:03 AM

15: Social Survey Methods

499

rural Honduran mestizo peasants, girls tend to assume household duties at a younger age than do boys.” From a theoretical and logical point of view, our hypotheses are made up of independent (or predictor) and dependent (or outcome) variables. We speak in terms of theory and logic because, as we will see, actually distinguishing independent and dependent variables empirically can be difficult. We nevertheless conceptualize the relationships among variables in terms of causal antecedence (independent variables) and causal consequence (dependent variables). Traditionally, in the philosophy of science, the logic of investigation is divided into a “context of discovery” and a “context of verification” (Rudner 1966). The former refers to the development of hypotheses and the latter to their testing. Hypotheses are derived from existing theory in part, but hypothesis formulation is also a dynamic and creative act and thus resists being systematized. Anthropology presents a somewhat specialized case, given that we carry out our research in such diverse social and cultural contexts. Hypotheses in anthropological fieldwork must be specified in terms of the particular society or community in which research is to be conducted. Because of this, Dressler (1995) argued that a useful step in hypothesis formulation is “the ethnographic critique of theory.” By this, he meant the careful scrutiny of theories and hypotheses in light of realities observed in the local communities in which the hypotheses are to be evaluated. For example, the hypothesis given above linking social support and depressive symptoms could be derived from any number of mid-range anthropological, sociological, or psychological theories. For anthropologists, the term “social support” immediately becomes problematic. How is social support conceptualized in a particular community? Through what mechanisms are help and support provided (e.g., along lines of kinship, friendship, work relationships, or some combination of all these)? And similar, even thornier, questions could be raised about the formulation and expression of depressive symptoms. Clearly the issues in fieldwork involve more than just the proper specification of a hypothesis, but it starts there. Janes (1990) presents a particularly cogent example of reformulating a social support hypothesis in terms of a Samoan migrant community in northern California. Our major point here is that the integration of social survey methods in anthropological fieldwork is a fundamentally ethnographic enterprise from the start. Of course, writing a proposal to fund your research will not get far if you claim that you can only state your hypotheses after you get to the field! Fortunately, the ethnographic record is large, and the potential to develop working hypotheses through the careful contextualized scrutiny of theory exists for nearly every anthropological question and every part of the globe. Thus, at the outset, the questions to be addressed through the integration of social survey methods into anthropological fieldwork must be carefully framed ethnographically, and be open to amendment through further on-the-ground ethnography.

Research Design Johnson and Hruschka discuss research design in detail in Chapter 3 of this volume. Here, we will outline briefly the various ways in which designs used in social survey research can be applied in fieldwork. We presume at the outset that the ethnographic

14_107-Bernard.indb 499

6/9/14 10:03 AM

500

William W. Dressl er and Kathryn S. Oths

focus of the research has already been determined. Typically this will be a community, but other kinds of foci are common, such as a geographic entity (e.g., an ecological zone), an organization (e.g., a corporation), an institution (e.g., a hospital), a social role (e.g., nursing), the entire globe (e.g., cross-cultural surveys), and others. Research design involves defining what unit of analysis within that ethnographic focus will be employed in the research. A unit of analysis is the entity on which observations are made. In social survey research, the unit involved is conventionally the individual human being, but need not be. There are often good reasons why the household might be the relevant unit of analysis in a particular research design. For example, studies of treatment choice in medical anthropology have taken the household as the unit of analysis, because treatment for one member may be a (joint) decision prioritized according to the health needs of other household members (Oths 1994, 1999). Similarly, the logical unit of analysis might be an organization (Jaskyte and Dressler 2004) or a whole social group or community (Young and Lyson 2006). And, taking whole societies as the unit of analysis, cross-cultural surveys have been a mainstay in anthropological research for decades (Brewis et al. 2011; Ember and Ember 2001; see Ember and Ember, this volume). Survey research design involves how you systematically organize the process of data collection. Different kinds of data can be collected from individuals, such as biological measurements (e.g., blood pressure) and specimens (e.g., cheek cell samples on a swab). Whatever the kinds of data collected, it will almost always involve talking to or interviewing individuals (discussed below). The classic case of survey research is a cross-sectional design, for which a representative sample of individuals from a community is sought. It is cross-sectional because it takes a slice out of the community at one point in time, with all data collected at the same time. Obviously, this research design can limit the kinds of inferences you can make. For example, while some variables are designated as independent and some as dependent, since both are collected at the same time you cannot truly infer that one came before the other. Nevertheless, certain kinds of causal inferences are sometimes reasonable, even in cross-sectional studies. For instance, it is acceptable logic to posit, although the two are being measured simultaneously, that caste membership precedes infant feeding practices (Panter-Brick 1991), not the reverse. A cross-sectional design is almost always the least costly in terms of time and effort because once you have completed it, you are done. It also has the advantage of providing the most comprehensive snapshot of the community at that point in time. Another way of organizing data collection is to collect some data at one point, then collect the same or other data at a later point or points in time. This is referred to as a “longitudinal,” “prospective,” or, sometimes, a “panel” study, depending on the discipline. The major feature of longitudinal research is that it can help reduce temporal ambiguity in causal inference. In principle, a cause can be clearly shown to precede an effect, though in practice it is often not that easy, depending on the variables being studied. If the outcome of interest is some “hard” endpoint—like birth or death—in studies in medical anthropology, or getting married or not in research on family formation, there is little ambiguity. But if the outcome is something that changes over time—like depressive symptoms or fertilizer use—reducing temporal ambiguity can depend on the number of different points in time that data can be collected.

14_107-Bernard.indb 500

6/9/14 10:03 AM

15: Social Survey Methods

501

Most applications of survey research in anthropology will be cross-sectional designs, given its lower cost both in terms of money and human resources, although there are examples of longitudinal designs in the literature. For example, Dressler and colleagues (Dressler et al. 2007) were able to extend what had been designed as a cross-sectional study to a two-year follow-up of over 75% of the originally sampled respondents. Wutich (2009) used multiple interviews across the course of a year in a Bolivian city to gain a more nuanced understanding of how water use varied seasonally, and especially how social resources that help maintain household access to water varied through time. Oths et al. (2001) conducted a longitudinal study of birth outcomes benefiting from the fact that pregnancy lasts nine months. Respondents were interviewed in their first and third trimester of pregnancy to investigate how psychosocial and physical stressors encountered during pregnancy were associated with birth weight. This innovative application of a longitudinal research design did not require the lengthy time periods usually associated with such research. Panel studies are longitudinal studies in which there are successive waves of data collection, following the same respondents through time with multiple data collection points. This approach has been used in medical anthropology to study treatment choice (Oths 1994; Young and Garro 1994). To determine how households respond to illness episodes, these investigators visited a random sample of households in their respective study communities every two weeks to determine who had been sick in the past two weeks and what they did about it. While Young and Garro (1994) pooled their data across six months of data collection, Oths (1994) was able to show that patterns of help-seeking changed between the beginning and the end of the data-collection period in response to an economic crisis. Godoy and colleagues (2009) conducted what is, as far as we know, the first largescale panel study conducted in anthropology. Godoy and his colleagues are working with the Tsimane’, a society in lowland Bolivia with diverse economic practices ranging from classic swidden agriculture to logging and wage labor. Godoy et al. collected data five times from a sample of 888 men and women distributed across 13 villages; the latter were systematically chosen to represent varying degrees of proximity to roads and market towns. With this extensive set of data, these investigators have examined hypotheses derived from ecological, economic, and medical anthropology theory. Clearly, this is a massive undertaking that depends on the coordination of a team of investigators, and it represents a milestone in anthropological research (see Gravlee et al. [2009] for more on longitudinal research in anthropology). In addition to cross-sectional and longitudinal research designs, a specialized design we like to include is the case-control design. While it derives from epidemiology, this design can be very useful in anthropological inference. As its name implies, it involves a comparison of a group of cases and a group of controls. What distinguishes it from merely a two-group comparison is that the groups are defined on the basis of the outcome variable. For example, the epidemiologist who is interested in the factors that contribute to heart attacks identifies a group of people who have suffered heart attacks and compares them on a variety of parameters to people who have not suffered heart attacks. In medical anthropology, this research design has been applied to the study of cultural syndromes such as rlung and susto (Janes 1999; Rubel et al. 1991). This would

14_107-Bernard.indb 501

6/9/14 10:03 AM

502

William W. Dressl er and Kathryn S. Oths

be a useful research design for investigating any relatively rare event (see Cohen’s [1971] study of divorce in Muslim societies of sub-Saharan Africa). The case-control design, though unconventional in social science research, does provide one solution to the dilemma faced by anthropologists in the field: scarce resources (human, temporal, and economic).

Sample Selection Some anthropologists contemplating the incorporation of social survey methods into their work may throw up their hands in despair, thinking that no matter how useful it would be to carry out survey research they will never have enough time, money, or people to do it. As we all know, traditional anthropological fieldwork is hard. Moving into a community, developing rapport, managing fieldwork relationships, improving language facility, refining ideas in light of on-the-ground realities, identifying key informants, conducting open-ended interviews, keeping good field notes, trying not to get sick, and finding some respite from all of the above takes time and energy. How are we supposed to fit a social survey into that agenda? Very carefully. The application of social survey methods must be focused and judicious simply because there is no alternative with respect to testing hypotheses or documenting the range of intracultural diversity. Part of the problem clearly has to do with samples, especially sample size. Adequate sample size is a logistical barrier for the fieldworker—whether working alone or as a member of a small team—that must be confronted. Anthropologists may read studies from other fields that have samples of 1,000 or more respondents and immediately lose heart. This trepidation may be compounded when you bravely seek advice from a statistician who authoritatively asserts that you cannot hope to learn anything of value without a sample that size. We beg to differ. Our position is that sample size is a bit of a red herring when it comes to survey methods in fieldwork. The key here again is the mixed methods, qualitative–quantitative strategy. First, the ethnographer applying social survey methods is not expecting those results to stand alone in the way that a sociological or epidemiologic survey would. The strength of the anthropologist’s survey stems precisely from the way it is carefully embedded in traditional fieldwork and articulates with data collected via participant observation and key-informant interviewing (an excellent example is Snodgrass et al. 2008). Second, the ethnographer can be strategic in focus regarding sampling efforts. For the test of a hypothesis or to document a certain pattern of intracultural diversity, the sample may be drawn from a particular stratum (e.g., age group or neighborhood) within a community that is both ethnographically and theoretically salient. Third, there are modern statistical techniques that enable us to explore the implications of employing a sample that is, by conventional statistical standards, small. For at least these three reasons, sample size should not be seen as a barrier to the use of social survey methods in anthropology. There is a fourth point worth noting. In large studies, survey researchers often have to employ many interviewers who are paid per interview. While there are a number of strategies for checking the accuracy of those interviews, the ethnographer may carry out a small-scale survey him- or herself, perhaps with the aid of a few assistants. The smaller sample size also allows the ethnographer the luxury of longer and more relaxed

14_107-Bernard.indb 502

6/9/14 10:03 AM

15: Social Survey Methods

503

interviews. What may be lost in statistical power may be gained in data quality control. We are not suggesting that issues of statistical power or the need for samples of a particular size to achieve a certain minimum error be ignored, only that there can be advantages to smaller samples (Benfer 1968). The ideal of sample selection in survey research is obtaining a set of respondents whose data can faithfully represent the characteristics of the larger population from which that sample is drawn. A well-selected (and sufficiently large) sample can reproduce population characteristics very well; therefore, it is unnecessary to assess whole populations. The key to representativeness is the unbiased selection of respondents, and the simplest way to achieve this is through random sampling. Many of us have had classroom exercises involving the selection of a “simple random sample” of something (like names from a telephone directory, or items from a top 100 list from the Internet). It requires that we have a complete enumeration of the population; that each member of the population be assigned a number; that we have access to a random numbers table or generator; and that the sample size be determined (ideally by using a standard formula based on confidence level chosen). We then select as respondents those whose number in the list corresponds to that of the number chosen in sequence from our random numbers table. In the technicalities of identifying a starting point in the table, accurately deciding on the number of digits to be used, and then deciding on a direction to move in the table, we sometimes forget the whole point: This way of selecting respondents ensures that no pattern creeps into our decisions (i.e., the likelihood of selecting one respondent is no different from selecting any other respondent). A systematic random sample is an alternative to a simple random sample and is especially useful when no complete enumeration of a population is available. For this, a starting point (e.g., a house on a road) is randomly selected, and then from there every nth house (e.g., every third house) is selected. While this may appear to diverge from the ideal sample, systematic random samples have been shown to perform quite well in practice, provided there is no inherent, undetected pattern in, say, house location. The selection of a random sample of respondents usually follows a number of other decisions that, fundamentally, are sampling decisions. When you decide to study one town rather than another, or one neighborhood rather than another, or one organization rather than another, you are making a sampling decision. In pure survey research, a logically similar kind of process is often systematized under terms like “multi-stage cluster sampling” or “stratified sampling.” For example, imagine a survey researcher wanting to sample respondents from the entire city of Chicago. While certain kinds of enumerations of the population exist, for a variety of reasons the researcher either will not be able to access them (e.g., for reasons of research ethics) or they will be too crude (e.g., a telephone book that only contains landlines). But one can access quality maps that show every block in the city. These blocks can be thought of as clusters of potential respondents. Next, blocks can be randomly sampled, and within each block a systematic random sample of occupied dwellings collected. Thus, the sample is collected at multiple stages, starting with the fact that respondents cluster in blocks. As noted above, anthropologists make similar kinds of decisions, but based on ethnographic criteria that can be built into the sampling design. For example, anthropologists working in an urban area in Latin America may not have the economic and human

14_107-Bernard.indb 503

6/9/14 10:03 AM

504

William W. Dressl er and Kathryn S. Oths

resources to sample the entire population of a city to document beliefs and practices across the full range of socioeconomic differences. But they could identify a transitional neighborhood that contains a substantial mix of working- and middle-class persons. By sampling within this neighborhood (a kind of cluster), they could document the variation in the variables of interest within an ethnographically salient group. Another type of sampling is referred to as “stratified random sampling.” Like cluster sampling, it first identifies relevant groups within the population; but unlike cluster sampling, these groups are usually identified on the basis of variables of interest. For example, an ethnographer may be interested in differences within an immigrant Hispanic community by nationality (i.e., comparing immigrants from Mexico, Guatemala, and Ecuador). One or more of these groups may have a smaller representation within the population, but the researcher wants to be able to compare them. In this situation, he or she may define national origin as sampling strata and sample equally within them. This may lead to an over- or underrepresentation of a group, but this strategy is necessary to ensure enough respondents in each group so that variables measured have stable statistical properties. If the researcher then wants to return to data that reflect the characteristics of the general population, the data can be weighted to do so, but sufficient data will have been obtained within strata to make comparisons meaningful. The ethnographer can exploit the advantages of stratified sampling in unique ways. For example, for his dissertation research Dressler (1982) examined the effects of socioeconomic and cultural change on blood pressure in a developing society. As a doctoral student working alone, he was unable to conduct a large-scale epidemiologic survey, so he used a stratified sampling strategy to focus his efforts. In many (but not all) societies, blood pressure rises with age. Within the 40–49-year-old age group, variation between individuals in blood pressure, and thus the potential for detecting correlates of the differences, is greatest. Therefore, he chose to sample exclusively from that age strata to test hypotheses about the socioeconomic and cultural correlates of blood pressure. By exploiting this empirical characteristic of blood pressure, Dressler was able to reduce the sample size he needed, thus making the collection of survey data practical. In any survey there will be refusals. “Response rate” refers to the number of respondents who participate in the research divided by the total number of potential respondents. Of course, many of these potential respondents have never been contacted. Typically in survey research, some criterion is set, such as failure to contact on three different days at three different times of day, that determines whether a potential respondent will be pursued or not. This can make that actual refusal rate, or the cooperation rate (the proportion of persons actually contacted who participate), somewhat more revealing (see AAPOR [2011] for a thorough discussion of these calculations for nonresponse). Nonresponse is not a problem with respect to the representativeness of a sample, so long as there is no systematic bias in the noncontacts or refusals. This can be difficult to detect, but the in-depth knowledge of a community that an ethnographer has can help in evaluating these potential sources of bias. Thus far, we have discussed only probability samples, but non-probability samples can be useful as well. Public opinion polling in the United States often employs quota samples. These are samples constructed to mirror the characteristics of the population at large (e.g., numbers of men, educational levels, ethnic groups, and so on that cor-

14_107-Bernard.indb 504

6/9/14 10:03 AM

15: Social Survey Methods

505

respond to their proportions in the general population). While unforeseen selection biases may creep into this process, it is a reasonable alternative to probability sampling. The above-mentioned case-control research design often involves non-probability sampling. As we noted, in a case-control design, two groups are formed on the basis of the dependent variable, such as having had the cultural syndrome of susto (a “case”) or not (a “control”). (Keep in mind that this is not a research design peculiar to health research—the outcome variable of interest could be anything.) Forming the group of cases proceeds by case finding. In certain circumstances, such as access to clinical records, it would be possible to sample randomly from a group of cases (see Janes 1999), but in many applications the investigator must simply try to track down as many persons as possible who fit that category. Then, either through population sampling or careful matching of controls to the cases, a sample is constructed in which differences between cases and controls on any variable of interest can be examined. In a case-control design, “matching” refers to selecting controls that correspond to cases on variables that the investigator wishes to control for. As an example, to control for age, you simply select as a control someone who matches the same age or five-year age category (such as 30–34) of your case. Control through matching then requires the use of specific statistical analyses, such as paired-comparison t-tests, but these are straightforward. The case-control design was developed as a means of coping with very small incidence rates in certain diseases and other difficulties in case-finding on a population level. It is a very efficient way of testing hypotheses. The ultimate aims of sampling strategies are both to achieve a reasonable representation of the population of interest and to provide data useful for testing hypotheses. Ethnographers who include social survey methods in their research must remain flexible and inventive to maximize the return on the resources invested in this phase of the research. DATA COLLECTION AND MEASUREMENT

Once a sample has been selected, data collection must proceed. Data can be collected in a variety of ways in survey research. We assume that anthropologists will be most comfortable with the face-to-face interviews collected with an interview schedule (see Weller, this volume, for an in-depth discussion of constructing interview schedules). The difference between an interview schedule and the more familiar kinds of informal interviewing is simply that with an interview schedule, each respondent is asked the exact same questions in the exact same manner with, if necessary, identical probes. Another major approach is the mail-out questionnaire. While this is seldom employed in anthropological research, as far back as the early 1960s social anthropologist John Beattie mailed questionnaires to over 300 village chiefs among the Bunyoro in Africa and had a gratifyingly high response rate (Beattie 1965, 38). Telephone surveys have also been employed by anthropologists. In Chavez et al.’s (2001) study of health care utilization in relation to cervical cancer among Latinas in southern California, the first two phases of the research consisted of intensive ethnographic research and a cultural domain analysis (see below), which they followed up with extensive telephone surveys.

14_107-Bernard.indb 505

6/9/14 10:03 AM

506

William W. Dressl er and Kathryn S. Oths

More recently, the Internet has become a medium for data collection. Snodgrass and associates (2011) have studied persons participating in online computer games by first intensively interviewing gamers at a local level, then carrying out a cultural domain analysis with them. These data were used to formulate a questionnaire posted on the Internet. This generated a convenience sample (i.e., whatever gamers responded) of 262. Our point here is that, while we continue to suspect that anthropologists will prefer the potential for rapport and data quality control afforded by face-to-face interviews, other kinds of data collection can be successfully integrated with ethnography. There are a number of social contexts in which other specialized forms of sampling and data collection can be useful, such as street intercept surveys (i.e., walking up to people on the street to ask for interviews) and others (see Miller et al. 1997). Before proceeding with data collection, it is essential that an interview schedule (or questionnaire) be pretested. Pretesting provides important feedback on how an interview schedule functions, especially in terms of the length of the interview, the ordering of the questions, and any lingering problems with meaning or translation. It is an opportunity to employ what in survey methods has been termed “cognitive interviewing” (Gerber 1999), a kind of debriefing of the pretest respondent to determine if he or she is thinking about the same thing in answering a question as you are in asking it. Questions are formulated in the context of a researcher’s particular conceptual orientation, and unless that orientation provides a common context between researcher and respondent, the data are less likely to be useful for evaluating the hypotheses posed. To give a concrete example, in a study of psychosocial job strain and pregnancy outcomes, Oths and colleagues (2001; Dunn and Oths 2004) discovered through careful ethnographic interviewing, observation, and pretesting that women tended to interpret the phrase “since you’ve been pregnant” to mean since the day their pregnancy test came back positive. This posed all sorts of difficulties for the use of standard scales for measuring smoking, drinking, domestic violence, moving residences, and the like. For instance, instead of asking the industry-standard “Since you’ve been pregnant, how many cigarettes do you smoke a day,” which nearly always resulted in a socially correct response of “none” (since to do otherwise would admit to being a bad mother), a more precise solution was to ask “When was the last time you . . . (smoked a cigarette, drank a beer or glass of wine, drank coffee, etc.),” then eliciting the amount. An accurate estimate of the date of conception (rather than date of pregnancy recognition) was gathered from delivery records, then compared against the “last time” data, to arrive at a better gauge of the behaviors actually engaged in during the critical first weeks of pregnancy. The effects of these variables differed in regression analyses, with the latter a better predictor of birth weight. This ethnographic correction of published scales for certain control variables no doubt accounted for the discovery of the job strain effect on birth weight, which had been previously studied yet undetected. These results underscore the importance of getting the wording right in constructing an interview schedule. Cognitive interviewing can provide an important step in ensuring the quality of data (see Willis and Miller 2011). Once an interview schedule has been carefully constructed, it must be put to use. The interview is a highly specialized form of social interaction between someone who wants information and others who are believed to be the best sources of the desired

14_107-Bernard.indb 506

6/9/14 10:03 AM

15: Social Survey Methods

507

information. While those of us raised in the West take this question-and-answer format for granted, those from other societies lacking such socialization may perceive the engagement as unusual at best, or aggressive, even hostile, at worst (Briggs 1986). An interview is not an everyday conversation, and the respondent who attempts to commandeer the information flow may be viewed as difficult. The feminist critique of survey research takes issue with the extent to which it exercises control over the direction of questioning, deeming the process manipulative (Oakley 1981). We hold that to collect systematic data and to create an affective connection with the respondent that encourages their self-expression need not be mutually exclusive enterprises. At the end of the day, if we go to the field with any intent whatsoever to study a phenomenon, we have no choice but to be directive. There are textbooks full of tips on how to conduct a good interview, so we will add just a few of our own. First and foremost is to understand that nothing can replace practice, practice, and then more practice. Thus, from undergraduate years on up, every opportunity to interview persons, whether formally or informally, should be grasped. One who plans to do social survey research should have (or develop) a high comfort level in talking to others about their lives and opinions. With the main goal of an interview being to achieve an accurate representation of what is in people’s head, the primary technique is to ask a question and then listen. Truly listen. This is nowhere near as easy as it sounds, to be still in our thoughts and not let them race ahead, to drop our opinions in Zen-like fashion, to momentarily suspend our investment in supporting the chosen hypothesis, to be uninfluenced (as much as humanly possible) by motivations or preconceived ideas. What comes out of respondents’ mouths is golden, even if at the moment it appears to contradict our cherished notions. What we learn with experience is that the reality out there is infinitely more interesting and complex than our minds can imagine a priori. That is why hypotheses in anthropology need to be provisional, not clung to when the evidence leads elsewhere. A very hard lesson to learn is to lose the attachment to our ideas while gathering data. Once the data have been collected, they must be coded and combined into measurements. Measurement refers to the principled assignment of numbers to observations. Susan Weller discusses measurement strategies in detail in this volume; here we will highlight a few issues that we believe are especially important in social survey research. We start with the assumption that much (if not all) of the data collected by an ethnographer in survey research should be meaningful to the participants. This may seem at first glance like a trivial statement, but data and measurements must be emically valid. We introduce the criterion of emic validity to stand alongside the well-known notions of face validity, construct validity, predictive validity, and convergent/divergent validity. The validity of a measurement is the degree to which that measurement assesses what it is intended to. An emic approach in ethnography is one in which the description of any phenomenon proceeds in terms that are used by, or meaningful to, the participants themselves. The emic validity of a measurement is evaluated by the demonstration that the measurement orders respondents along some continuum in terms that are meaningful to the respondents themselves. We are not referring to the meaningfulness of terms simply as lexical equivalence in a local language, which can

14_107-Bernard.indb 507

6/9/14 10:03 AM

508

William W. Dressl er and Kathryn S. Oths

be accomplished by the well-known techniques of back-translation, but rather that it is composed of the concepts and terms that culturally structure the domain to which it refers within that particular social setting. This is quite a different notion of measurement than normally encountered in the research methods literature. Typically, when a measurement or scale is constructed, the arbiters of the relevance of particular items (questions) to the domain of interest are other social scientists (DeVellis 2003). Often panels of experts in the field are recruited to supply items that are then winnowed down by the scrutiny of other experts until you arrive at a set of items that “truly” capture the domain of interest. In ethnography, the experts are the participants in a particular social setting. While they rarely have the goal of constructing scales that can be used in social surveys, they do know the terms and concepts by which they communicate with their peers about a particular cultural domain. Therefore, in developing questions for surveys and measurements of important variables, it is of the utmost importance that we demonstrate emic validity in our social survey work. The development of scales and measurements has been an area of obvious convergence in the collection of qualitative and quantitative data. Usually, the ethnographer devotes a considerable amount of time to participant observation and key-informant interviewing to learn about how a particular domain (e.g., family life, religion) is culturally constructed in a particular community. After the ethnographer has gained a certain cultural competence in the domain, with respect to understanding the specific terms used to talk about it and how terms are linked in prototypical processes, he or she is ready to develop a set of questions for use in a survey to measure more systematically some aspect of the domain. In this case, the demonstration of emic validity depends on the skill of the ethnographer in tracing the links between ethnographic data and the questions developed (see Janes’s [1990] study of Samoan migrants to California for a superb example). Cultural domain analysis (Borgatti 1999) refers to a linked set of specialized interviews and analyses to understand the semantic structure of a cultural domain. These interviewing techniques include free listing, pile sorts, and rating and ranking tasks, culminating in a test for cultural consensus in the domain (Romney et al. 1986). Dressler et al. (2005) have demonstrated step-by-step how cultural domain analysis can be used to construct items for survey research, from first generating the terms that people use to talking about a domain to systematically measuring some dimension of that domain. On the technical side, we have incorporated an architecture for Likert scale questioning that results in a more precise answer by breaking the question into two steps to indicate individuals’ intensity of agreement-disagreement. In conventional survey interviewing, respondents are often presented with the entire Likert-response format and encouraged to choose their own response out of the four or more alternatives. We have encountered anthropologists who assert that Likert-response scales cannot be used in the field because it is such an alien task for many people not socialized to standardized testing formats to respond with “3” when asked to evaluate a statement about something. Phrased in that way, it probably is. There are other ways of asking, however. To a survey item such as, “It is difficult to rely on the support of other people,” the respondent can be asked, “Would you agree or

14_107-Bernard.indb 508

6/9/14 10:03 AM

15: Social Survey Methods

509

disagree with this statement?” If the respondent answers “Agree,” it can then be asked, “Do you agree a little or a lot?” (The same format is employed for a disagree answer.) What results is a four-point scale equivalent to strongly agree—agree—disagree— strongly disagree. But note that people are not being asked to respond in terms of the Likert scale. They are simply asked to agree or disagree with an assertion, then the Likert scale is reconstructed from their pattern of agreement or disagreement (Weller and Romney 1988, 42–43). Such a response format helps increase the internal consistency reliability and precision of a scale (see below) because it introduces finer gradations into the measurement than does a simple yes/no or agree/disagree. Thus far, we have encountered no one for whom this is a confusing task. Many orthodox texts on social survey methods will send the investigator to the literature to find existing questions and scales, validated and “normed,” to incorporate into their surveys. This is not inappropriate, especially considering that a theoretical construct may require a very specialized set of questions. In a situation like this, the process of backtranslation usually looms large (see Bernard 2011). But this does not necessarily absolve us of the responsibility of exploring the emic validity of the measurement. Gannotti and Handwerker (2002) provide an example of doing just this by working in reverse from the process we have described above. They were interested in the measurement of disability in Puerto Rico, and, to maximize the comparability of their data to other studies in the area, they selected a widely used measure of disability. They were saved the step of translation and back-translation because a Spanishlanguage version of the scale existed. At the same time, they wanted to confirm that the categories used to identify disability in the existing scale were meaningful to their respondents. So, before using the scale, they had a set of respondents rate the items as more or less useful in describing disability, as the respondents understood that term. When these ratings were analyzed using cultural consensus analysis, the investigators found overall consensus, although there were some items that could be identified as less salient for the specific community being studied. In the end, they concluded that the scale had sufficient emic validity to justify its use in their study (see Ice and Yogo [2005], Kennedy [2005], and Schrauf and Navarro [2005] for additional examples of how to develop and adapt scales). While intensive evaluation of emic validity for each and every measure employed in a mixed methods study would be impractical, we do suggest that investigators invest as much time as it takes to construct, then pretest, survey instruments before actually implementing them. DATA ANALYSIS

Weller (this volume) discusses reliability analyses for evaluating measurements, and Handwerker and Borgatti (this volume) discuss numerical analyses in anthropology. We will build on these chapters to highlight some points of specific relevance to the application of social survey methods in anthropology. As argued above, ethnographers should develop measurements in the field that have high emic validity. An important next step is to evaluate the psychometric characteristics of those measurements using conventional analyses of reliability and validity. As noted, validity refers to the degree to which a scale actually measures

14_107-Bernard.indb 509

6/9/14 10:03 AM

510

William W. Dressl er and Kathryn S. Oths

what it is intended to measure, and reliability refers to the repeatability or reproducibility of the values for the scale. Reliability can be assessed in a variety of ways, the most obvious being that if you give the scale to a set of respondents, and then repeat that within a short enough time span (e.g., two weeks) such that little is likely to have changed substantially, you should get back about the same result. That is, the measurements should be highly correlated. Although theoretically possible, this is not often done, given all fieldworkers’ other tasks. It is more likely that reliability is evaluated using the notion of internal consistency. In conventional scaling theory, it is assumed that we are trying to measure a construct to which we have imperfect empirical access. If we could ask one question (e.g., how persons in a community access social support) that would give us a perfect picture of how they do, we would. But there is no perfectly valid and reliable single question to ask. So, we have to zero in on that construct by asking a number of different questions, all of which pertain to the domain but each of which is imperfect (i.e., can only partially illuminate a pattern or schema we are trying to understand in people’s lived realities). Getting the questions as good as possible is, of course, the whole issue. In theory, if the questions pertain to the domain, respondents should show a measure of consistency in the way they answer the questions. This then can be used to evaluate the reliability of a scale. Cronbach’s alpha is a coefficient of internal consistency (regularity internal to the scale) to assess this that, like a correlation coefficient, varies between 0.0 and 1.0: The closer to 1, the greater the reliability. Our experience has shown that when a researcher takes pains to achieve high emic validity, the reliability of the measure will follow. Assessing the validity of a scale is harder, not just for the ethnographer but for anyone. The best way to assess validity is to have a “gold standard” measurement (such as the measurement of blood pressure with a catheter inserted into a vein) against which you can compare your field measurement (the measurement of blood pressure with a sphygmomanometer). In the social sciences, gold standards are hard to come by; they are even more difficult for the ethnographer in the field to achieve. Evaluation of validity can at least be approximated, however, by inspecting the pattern of associations between the scale of interest and variables that, in theory, should be correlated with that scale. For example, we developed a series of scales of cultural consonance in our Brazilian work. These assess how well individuals are able to approximate, in their own lives, the cultural prototypes for beliefs and behaviors. It seemed reasonable to anticipate that persons with the most economic resources would best be able to achieve these cultural ideals. And, indeed, we found that, overall, cultural consonance was associated with higher socioeconomic status. Similarly, we reasoned that low cultural consonance would be experienced as distressing, and all of the scales of cultural consonance were at least moderately correlated with scales assessing psychological distress. Although we could not evaluate the scales against some gold standard, the pattern of associations lends confidence to our sense that the scales of cultural consonance do measure what they are intended to measure (see Dressler et al. 2005). Once the ethnographer has a set of survey data in hand, the whole range of bivariate and multivariate inferential statistics is available to better understand those data. Like

14_107-Bernard.indb 510

6/9/14 10:03 AM

15: Social Survey Methods

511

Handwerker (2001), we strongly recommend that data be analyzed using multivariate models, and traditional ordinary least squares multiple regression analysis is one of our favorites. As we tell our students, “If you can say it, you can test it,” meaning that any complex hypothesis regarding sociocultural processes can be evaluated through proper specification of a linear model. Or, if you think that there are curvilinear associations (i.e., two variables covary in the same direction to a point, then begin to covary in the opposite direction) moderated by various factors, you can test that, too (see Dressler and Bindon [2000] for an example). A detailed discussion of how to specify linear models is beyond the scope of this discussion and can be found elsewhere (Cohen and Cohen [1975] provides lucid discussions of the specification of complex models). As we have argued, small samples can be desirable in anthropological inference, embedded as they are in a larger data context of ethnography (Benfer 1968). Furthermore, there are some innovations in data analysis to aid in the evaluation of the inferential implications of small samples. We are all familiar with the case of examining the mean of a variable, only to then view a histogram and spot a peculiar respondent with an exceptionally high value on that variable that skews the distribution to the right. Conventionally, this case would be called an “outlier.” The very term implies something at least peculiar, if not downright in error, about that case’s value. If, after double-checking for coding errors, it is the true value for that respondent, what is one to do? Should it be excluded simply because it is extreme? Or would it be better to evaluate the impact that having that respondent in our data has on the analysis of those data? These questions have led some to prefer the idea of “influential case”—literally a respondent who has a particularly strong effect on the outcome of an analysis—to the idea of outlier. There is an entire system for influential case analysis in both ordinary least squares regression analysis and logistic regression analysis (Bollen and Jackman 1985, 1990). It is possible to identify cases that have an extra influence on the results, whether as a result of their values on the dependent variable or on one or more independent variables. We agree with Bollen and Jackman (1985) that the goal of these analyses is not necessarily to eliminate cases from the analysis but rather to understand better what is going on in a dataset. This should be particularly appealing to ethnographers since the respondent who turns out to be an influential case may be known from the data collection process (see Dressler [1991] for an influential case analysis in ethnographic research).

Integrating Qualitative and Quantitative Methods Depicted graphically (see Figure 15.1), the research process appears to begin with a statement of the problem, move to theory to provide an overarching framework (literature review), shift to qualitative discovery (participant observation and ethnographic interviewing), and finally arrive at measurement, that is, quantification (survey interview construction and delivery). However, the arrows in Figure 15.1 remind us that the association between qualitative and quantitative methods is neither a linear nor a static process but rather an iterative one, with each step feeding the others. The potential exists for a seamless movement back and forth across data collection and interpretation. For instance, ethnographic insights might send one back to the literature. Or in the pretesting stage, it may become necessary to once again gather qualitative data to settle unforeseen

14_107-Bernard.indb 511

6/9/14 10:03 AM

512

William W. Dressl er and Kathryn S. Oths

Figure 15.1.  Qualitative-quantitative research process.

ambiguity in emic meaning. This should be the take-home point of our chapter: In survey research it is best to be open to reality and take what the field situation gives us, wellprovisioned with a toolkit from which to select the best instrument for the situation. Our agility in this sense is not unlike that of a football quarterback who changes the play at the line of scrimmage when he reads the formation the defense lines up in. DISCUSSION AND CONCLUSION

Social survey methods provide social scientists with the means to assess the distribution of variables in samples representative of the communities they study. These methods enable investigators to describe that variability and to employ it in the evaluation of hypotheses using the tools of inferential statistics. To fully realize the potential of survey methods in anthropology requires that the ethnographer think in a numerate way. We have often observed anthropologists take the time and effort to construct an impeccable, ground-truthed interview instrument and use it in a survey that is time consuming and generates excellent data. Yet, lacking the ability to imagine the uses to which those data can be put, they leave the full potential of the dataset unrealized. Social surveys are valuable adjuncts to the methods traditionally employed in anthropological fieldwork and they have been used by ethnographers for quite some time. The keys to successful application of social survey methods in anthropology, as we see it, are the careful integration of surveys with other ways of collecting data in the field and the adaptation of the survey method to address the kinds of questions that anthropologists ask. The issues embedded in applying surveys in fieldwork take effort to address and can ultimately be resolved to the benefit of both the investigator and the results obtained. REFERENCES

American Association of Public Opinion Research (AAPOR). 2011. Standard definitions: Final disposition of case codes and outcome rates for surveys, 7th ed. AAPOR. http://www .aapor.org/AM/Template.cfm?Section=Standard_Definitions2&Template=/CM/Content Display.cfm&ContentID=3156 (accessed February 16, 2013). Beattie, J. 1965. Understanding an African kingdom: Bunyoro. New York: Holt, Rinehart, and Winston. Benfer, R. A. 1968. The desirability of small samples for anthropological inference. American Anthropologist 70: 949–51.

14_107-Bernard.indb 512

6/9/14 10:03 AM

15: Social Survey Methods

513

Bennett, J. W., and G. Thaiss. 1973. Survey research in anthropological field work. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 316–37. New York: Columbia University Press. Bernard, H. R. 2011. Research methods in anthropology, 5th ed. Lanham, MD: AltaMira. Bollen, K. A., and R. W. Jackman. 1985. Regression diagnostics. Sociological Methods and Research 13: 510–42. Bollen, K. A., and R. W. Jackman. 1990. Regression diagnostics: An expository treatment of outliers and influential cases. In Modern methods of data analysis, ed. J. Fox and J. S. Long, 257–91. Newbury Park, CA: Sage. Borgatti, S. P. 1999. Elicitation techniques for cultural domain analysis. In Ethnographer‘s toolkit: Enhanced ethnographic methods, vol. 3, ed. J. J. Schensul, M. D. LeCompte, B. K. Nastasi, and S. P. Borgatti, 115–51. Walnut Creek, CA: AltaMira. Brewis, A. A., A. Wutich, A. Falletta-Cowden, and I. Rodriguez-Soto. 2011. Body norms and fat stigma in global perspective. Current Anthropology 52: 269–76. Briggs, C. L. 1986. Learning how to ask: A sociolinguistic appraisal of the role of the interview in social science research. Cambridge: Cambridge University Press. Chavez, L., J. M. McMullin, S. I. Mishra, and F. A. Hubbell. 2001. Beliefs matter: Cultural beliefs and the use of cervical cancer-screening tests. American Anthropologist 103: 1114–29. Cohen, J., and P. Cohen. 1975. Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Lawrence Erlbaum. Cohen, R. 1971. Dominance and defiance: A study of marital instability in an Islamic African society. Washington, DC: American Anthropological Association. DeVellis, R. F. 2003. Scale development: Theory and applications, 2nd ed. Thousand Oaks, CA: Sage. Dillman, D. A. 2007. Mail and Internet surveys: The tailored design method. Hoboken, NJ: Wiley. Dressler, W. W. 1982. Hypertension and culture change: Acculturation and disease in the West Indies. South Salem, NY: Redgrave Publishing. Dressler, W. W. 1991. Stress and adaptation in the context of culture: Depression in a southern black community. Albany: State University of New York Press. Dressler, W. W. 1995. Modeling biocultural interactions in anthropological research: An example from research on stress and cardiovascular disease. Yearbook of Physical Anthropology 38: 27–56. Dressler, W. W., M. C. Balieiro, R. P. Ribeiro, and J. E. dos Santos. 2007. A prospective study of cultural consonance and depressive symptoms in urban Brazil. Social Science and Medicine 65: 2058–69. Dressler, W. W., and J. R. Bindon. 2000. The health consequences of cultural consonance: Cultural dimensions of lifestyle, social support and arterial blood pressure in an African American community. American Anthropologist 102: 244–60. Dressler, W. W., C. D. Borges, M. C. Balieiro, and J. E. Dos Santos. 2005. Measuring cultural consonance: Examples with special reference to measurement theory in anthropology. Field Methods 17: 331–55. Dunn, L. L., and K. S. Oths. 2004. Prenatal predictors of intimate partner abuse. Journal of Obstetric, Gynecologic, and Neonatal Nursing 33: 54–63. Edgerton, R. B. 1971. The individual in cultural adaptation. Berkeley: University of California Press. Ember, C. R., and M. Ember. 2001. Cross-cultural research methods. Walnut Creek, CA: AltaMira. Fink, A., ed. 1995. The survey kit. Thousand Oaks, CA: Sage. Fink, A. 2006. How to conduct surveys: A step-by-step guide. Thousand Oaks, CA: Sage.

14_107-Bernard.indb 513

6/9/14 10:03 AM

514

William W. Dressl er and Kathryn S. Oths

Gannotti, M., and W. P. Handwerker. 2002. Puerto Rican understandings of child disability: Methods for the cultural validation of standardized measures of child health. Social Science and Medicine 55: 11–23. Gerber, E. R. 1999. The view from anthropology: Ethnography and the cognitive interview. In Cognition and survey research, ed. M. G. Sirken, D. J. Herrmann, S. Schechter, N. Schwarz et al., 217–34. New York: Wiley. Godoy, R., V. Reyes-Garcia, C. C. Gravlee, T. Huanaca et al. 2009. Moving beyond a snapshot to understand changes in the well-being of native Amazonians: Panel evidence (2002–2006) from Bolivia. Current Anthropology 50: 563–73. Graves, T. D. 1967. Acculturation, access, and alcohol use in a tri-ethnic community. American Anthropologist 69: 306–21. Gravlee, C. C., D. P. Kennedy, R. Godoy, and W. R. Leonard. 2009. Methods for collecting panel data: What does cultural anthropology have to learn from other disciplines? Journal of Anthropological Research 65: 453–83. Handwerker, W. P. 2001. Quick ethnography. Walnut Creek, CA: AltaMira. Ice, G. H., and J. Yogo. 2005. Measuring stress among Luo elders: Development of the Luo Perceived Stress Scale. Field Methods 17: 394–411. Janes, C. R. 1990. Migration, social change, and health. Stanford, CA: Stanford University Press. Janes, C. R. 1999. Imagined lives, suffering, and the work of culture. Medical Anthropology Quarterly 13: 391–412. Jaskyte, K., and W. W. Dressler. 2004. Studying culture as an integral aggregate variable: Organizational culture and innovation in a group of nonprofit organizations. Field Methods 16: 265–84. Kennedy, D. P. 2005. Scale adaptation and ethnography. Field Methods 17: 412–31. Malinowski, B. 1922. Argonauts of the western Pacific. London: Routledge and Kegan Paul. Miller, K. W., L. B. Wilder, F. A. Stillman, and D. M. Becker. 1997. The feasibility of a streetintercept survey method in an African-American community. American Journal of Public Health 87: 655–58. Nardi, P. M. 2006. Doing survey research. Boston: Pearson Education. Oakley, A. 1981. Interviewing women: A contradiction in terms. In Doing feminist research, ed. H. Roberts, 30–61. London: Routledge. Oths, K. S. 1994. Health care decisions of households in economic crisis. Human Organization 53: 245–54. Oths, K. S. 1999. Debilidad: A reproductive-related illness of the Peruvian Andes. Medical Anthropology Quarterly 13: 386–15. Oths, K. S., L. L. Dunn, and N. S. Palmer. 2001. A prospective study of psychosocial job strain and birth outcomes. Epidemiology 12: 744–46. Panter-Brick, C. 1991. Lactation, birth spacing, and maternal work loads among two castes in rural Nepal. Biosocial Science 23: 137–54. Pelto, P. J., and G. H. Pelto. 1975. Intracultural diversity: Some theoretical issues. American Ethnologist 2: 1–18. Pyrczak, F., and R. Bruce. 2011. Writing empirical research reports, 7th ed. Glendale, CA: Pyrczak Publishing. Richards, A. 1935. The village census in the study of culture contact. Africa 8: 20–33. Romney, A. K., S. C. Weller, and W. Batchelder. 1986. Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–38. Rubel, A. J., C. W. O’Nell, and R. Collado-Ardon. 1991. Susto: A folk illness. Los Angeles and Berkeley: University of California Press. Rudner, R. 1966. Philosophy of social science. Englewood Cliffs, NJ: Prentice Hall

14_107-Bernard.indb 514

6/9/14 10:03 AM

15: Social Survey Methods

515

Salant, P., and D. A. Dillman. 1994. How to conduct your own survey. New York: Wiley. Sapir, E. 1938. Why anthropology needs the psychiatrist. Psychiatry 1: 7–12. Schrauf, R. W., and E. Navarro. 2005. Using existing tests and scales in the field. Field Methods 17: 373–93. Snodgrass, J. G., M. G. Lacy, H. J. F. Dengah, J. Fagan, and D. E. Most. 2011. Magical flight and monstrous stress. Culture, Medicine and Psychiatry 35: 26–62.  Snodgrass, J. G., M. G. Lacy, S. K. Sharma, Y. S. Jhala et al. 2008. Witch hunts, herbal healing, and discourses of indigenous ecodevelopment in North India: Theory and method in the anthropology of environmentality. American Anthropologist 110: 299–312. Tashakkori, A., and C. Teddlie, eds. 2003. Handbook of mixed methods research in social and behavioral sciences. Thousand Oaks, CA: Sage. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Qualitative research methods series, Vol. 10. Newbury Park, CA: Sage. Willis, G. B., and K. Miller. 2011. Cross-cultural cognitive interviewing seeking comparability and enhancing understanding. Field Methods 23: 331–41. Wutich, A. 2009. Water scarcity and the sustainability of a common pool resource institution in the urban Andes. Human Ecology 37: 179–92. Young, F. W., and T. A. Lyson. 2006. Community types and social indicator rates. CrossCultural Research 40: 405–25. Young, J. C., and L. A. Garro. 1994. Medical choice in a Mexican village. Prospect Heights, IL: Waveland.

14_107-Bernard.indb 515

6/9/14 10:03 AM

14_107-Bernard.indb 516

6/9/14 10:03 AM

Part III

INTERPRETING INFORMATION

14_107-Bernard.indb 517

6/9/14 10:03 AM

14_107-Bernard.indb 518

6/9/14 10:03 AM

CHAPTER 16

Reasoning with Numbers W. Penn Handwerker and Stephen P. Borgatti

This chapter explains the reasoning behind numerical methods and illustrates their application to the study of cultures. Numerical methods encompass a wide variety of inter-linked techniques that range from basic mathematical tools (e.g., matrix algebra) and simple descriptive statistics (modes, medians, means, percentages, ratios), to visualization and data reduction techniques (e.g., multidimensional scaling and factor analysis), to classical statistical testing tools (e.g., t-tests and regression). Numerical reasoning consists of answering research questions with numerical methods. ANSWERING QUESTIONS WE ASK IN THE FIELD

In a sense, all anthropological research about cultures begins by collecting data from one person. Then we go to the next person, almost always finding something different, as well as something much the same, and on to the next. We keep track of similarities, note variability, and keep at it until we no longer find much that is new. We then construct a story from the generalizations we arrived at about the people we worked with, their lives, and the circumstances in which they have lived; what those people now think, feel, and do; and who agreed with whom about what and to what extent. Our study has internal validity to the extent that we do a good job of measuring and analyzing. Internal validity can be seen as the object of the question “Did we get it right?” The stylized account of the research process given above highlights eight activities entailed in ordinary fieldwork: 1. We measure variables. Variables refer to phenomena that may vary from one observation to another, such as the amount of land farmed by each household or the number of children a woman has. Measurement transforms sensory information into intelligible mental constructions. Measurement thus refers to what we do when we make and record observations about what we experience in the field. Variables can be used to conceptualize both qualitative and quantitative differences. For example, the variation in the variable kin may consist of qualitative categories such as brother, sister, husband, or wife. The variation in the variable gender (usually) consists of the difference between woman and man, or girl and boy. Variation may also be quantitative. The variable family size may be measured as small, medium, and large (an ordinal form of measurement that only entails relative rank); or be expressed as numbers 5, 10, and 20 (a ratio form of measurement characterized by a true zero, which allows any two values to be expressed as a meaningful ratio of the other). Field study often entails altering the mental constructs we started with

519

14_107-Bernard.indb 519

6/9/14 10:03 AM

520

W. Penn Handwerker and Stephen P. Borgatti in ways that improve their correspondence with those used by the people in our field setting, as when we reconceptualize status based on our informants’ understanding of it. We also may create new ways of thinking about the world to help us better understand what we experience in the field. In either event, we aspire to an understanding of what we see during fieldwork that corresponds with what our informants tried to teach us, and takes that understanding a step further. One of the main threats to the validity of research findings is measurement error. This kind of error can be reduced by intensive and prolonged interaction predicated on personal relationships with informants. Triangulation using a variety of data collection and analysis tools to address any one question or issue helps even more (McNabb 1990).

2. We aim to identify what is typical in one sense or another. For one purpose or another, we may want to know: • The most common (frequent) value of a variable (called the mode), such as how most people conceptualize specific people, events, and processes. • The value that best represents all of the values of a variable in the sense of being the least different from all values. When least different is defined as the sum of squared differences, the value least different from all others is called the mean. For example, we may be interested in the mean age at which children become adults, or people first have sex, marry, give birth, or become grandparents. • The middle of a distribution of values, in the sense that half of the observations have a value greater than the middle and half have a value smaller than the middle. This middle is the median. With the median, we can identify important and unimportant people, events, and processes, or people who move into adulthood, first have sex, marry, give birth, and become grandparents early or late. The median is one of a set of statistics called quantiles that measure relative placement in a distribution. The median, for example, is the 50th percentile. Other quantiles include the 10th percentile (which identifies the location below which lie the earliest or lowest set of cases), the 25th and the 75th percentiles (between which lie the middle 50% of the cases), and the 90th percentile (which identifies the location above which lie the latest or highest set of cases). 3. We evaluate variation from what is typical. For one purpose or another, we may want to know: • How common the mode is. That is, how frequent is the most frequent value? We can express this number as a proportion (the modal frequency/the total count), a percentage (100*a proportion), or a ratio (e.g., number of people who express the most common point of view/the number of people who express any other point of view). For example, if we count how often particular ways of conceptualizing specific people, events, and processes occur, we might be inclined to generalize a consensus if 80% of the people we talk with express the most common (modal) point of view. We probably would infer much cultural diversity if we find many ways of thinking about an issue and only 20% express the most common (modal) view. In the former case, the modal view occurs four times more often than any alternative (80/20); in the latter, it occurs with one-quarter of the frequency of the alternatives (20/80). • How evenly distributed the values of a variable are. For example, if we detect five ways of conceptualizing something and find that all five are equally common, the variable is maximally heterogeneous—thoroughly dispersed. But if most of the

14_107-Bernard.indb 520

6/9/14 10:03 AM

16: Reasoning with Numbers

521

responses fall in just a couple of the categories, the variable is more homogeneous or weakly dispersed. Great variation suggests the operation of contingencies that influence how people conceptualize themselves and the world around them. Little variation suggests the possibility of significant constraints. • What the typical difference is between the value of a randomly chosen observation and the mean value. This is the standard deviation. For example, how much variation on average exists in the age at which children become adults and people first have sex, marry, give birth, and become grandparents. Again, great variation suggests the operation of many contingencies that influence when and, perhaps how, life-course events occur. Little variation suggests the operation of significant constraints bearing on life-course events. • The range of a variable. For example, the earliest point at which an event occurs in people’s lives (the minimum), the latest point that event occurs (the maximum), the difference between these points (the range), and the relative frequency with which an event occurs between these points (a variable’s frequency distribution). The distribution of events may be skewed over a given range of occurrence— most people marry first at relatively young ages, but some marry first late in life, for example. Or a distribution may be approximately symmetric, such as the age at which children undergo one or another puberty rite. Pictures of the distribution of a variable, such as stem-and-leaf plots, box plots, and quantile plots, make it easy to evaluate these salient characteristics of a variable’s distribution. For example, Figure 16.1 shows a box plot that summarizes the number of commercial sex transactions in the past four hours. The line in the middle of the box indicates the median value, the left and right ends of the box indicate the 25th and 75th percentiles, and the ends of the left and right “whiskers” indicate the smallest and largest values in the dataset. 4. We try to identify the variables, people, events, and processes that go together and those that don’t. Similarity coefficients, each in different ways, measure the extent to which two phenomena go together or not. Dissimilarity coefficients measure the extent to which two phenomena don’t go together. Datasets commonly contain thousands of values, making it difficult to directly perceive relationships among variables or cases. Coefficients of similarity and dissimilarity summarize in a single value the relationship between two variables. Common similarity coefficients include Pearson’s r (the correlation coefficient), the simple matching coefficient, and the Jaccard coefficient. Pearson’s r is used with quantitative variables and tells us the extent to which variation in one phenomenon accompanies variation in another. It varies between ±1.00. Positive coefficients tell you that one variable goes up when the other goes up; negative coefficients tell you that one variable goes up when the other goes down; values close to 0.00 tell you that the variation in one variable is independent of the variation in the other—knowing the values of one variable doesn’t

Figure 16.1. Box plot summarizing data on the number of commercial sex transactions in the past four hours.

14_107-Bernard.indb 521

6/9/14 10:03 AM

522

W. Penn Handwerker and Stephen P. Borgatti help you predict the corresponding values of the other variable. For binary data, we have the simple matching coefficient (which counts up the number of times two variables have exactly the same values, expressed as a proportion of the number of times possible) and the Jaccard coefficient (also known as the positive matching coefficient), which counts the number of times that two variables both have values of 1, divided by the number times at least one of them is 1. These coefficients are often used to assess the similarity between objects described by lists of traits that they have or don’t have. Common dissimilarity coefficients include Euclidean distances, the Hamming distance, and one minus any similarity coefficient. Euclidean distance is defined as the square root of the summed squared differences between corresponding values of two variables. The Euclidean distance is zero when the variables are identical and gets larger as the values of the two variables get more and more different. The Hamming distance between any two variables or things is defined as the number of changes that have to be made to the data to make the two variables or things identical. Hence, if a stallion is defined as a horse that is male and adult and a mare is defined as a horse that is female and adult, the Hamming distance between them is just 1—we only need to change the gender to make them the same. Tools like factor analysis, cluster analysis, multidimensional scaling, and correspondence analysis use similarity and dissimilarity coefficients to find patterns in data, typically producing visual displays that help the researcher see the patterns. For example, Figure 16.2 shows a multidimensional scaling of correlations among variables measuring respondents’ liking of different kinds of music. Two variables are near each other in the map if there is a high correlation between them, meaning that people who like one also like the other.

Figure 16.2. Multidimensional scaling of correlations among variables measuring preferences for kinds of music.

14_107-Bernard.indb 522

6/9/14 10:03 AM

16: Reasoning with Numbers

523

5. We try to identify who agrees with whom about what and to what extent, which we can measure explicitly with similarity coefficients. A technique called consensus analysis (Romney et al. 1986) is used to measure the extent to which a sample of respondents exhibits cultural consensus and to assess the extent to which each individual is near the cultural center (meaning that their responses are culturally prototypical). 6. We aspire to an understanding of what we see during fieldwork that goes beyond the few people we had a chance to learn from. A study has external validity to the extent that it is generalizable, meaning that the results obtained on our sample can be safely ascribed to the population from which we drew it. This goal signals the difference between using statistics to describe characteristics of the people we spoke with (descriptive statistics) and using the same statistics to infer something about a larger population (inferential statistics). Research design and sampling procedures address these questions. When it is not feasible to collect data from the entire population that we want our results to apply to, we must take a sample. Unfortunately, gathering information from a different sample, or the same people at different times (the next day) and places (next door), almost always yields different findings even though the population we are studying is the same. The purpose of inferential statistics is to provide a way to estimate or infer the population values (called parameters) from the sample data, and give us a sense of how far off our estimates might be due to the fact that we drew a particular sample of individuals (rather than some other set of individuals that we might have chosen instead). With sampling, we can answer, with some level of accuracy, questions such as these: • What percentage of American voters now believe that we need to prevent global warming? • Have women’s preferences for metrosexual compared to conventionally masculine men changed in the last decade? • Do all forms of traumatic stress experienced in childhood increase the risk of depression in adulthood? To do the kind of analysis of conventional life experience markers like the three listed above, we start by drawing a probability sample such as a simple random sample. In a simple random sample, each person in a given population has an equal of opportunity of appearing in your sample. These are the easiest kinds of samples to analyze, but any sample can be analyzed as long as the probability of each sample member having been chosen is known—that is what’s meant by a probability sample. Once we have the sample and have measured the variables we need, we can compute point estimates of population statistics (e.g., the average age of people’s first brush with cancer). You can also compute confidence intervals for the statistic. A confidence interval is a range of values that are very likely to “cover” the unknown population value. A 95% confidence interval is an interval that has a 95% chance of covering the population parameter (so, if you draw thousands of samples of the same size and computed a 95% confidence interval for each one, only about 5% of them would fail to include the “right” or population answer). In addition, significance tests (e.g., chi-squared, t-tests, F-ratios) yield probabilities that tell you how often to expect a specific result (e.g., a large difference in means) just because of random sampling fluctuations. For example, a significance test will give you the probability—by chance alone—of getting a correlation as large as the one you actually observed. In other words, it tells you the

14_107-Bernard.indb 523

6/9/14 10:03 AM

524

W. Penn Handwerker and Stephen P. Borgatti

probability of pulling a sample in which the correlation happens to be quite large, even if the correlation is zero in the population. Classical inferential statistics make use of the fact that the distributions of certain sample statistics (such as the mean or the Pearson correlation) are known to match certain mathematical distributions when the samples are random and large enough. When the distributions are unknown, other techniques must be used. The bootstrap (Efrom and Tibshirani 1991; Mooney and Duval 1993), jackknife (Tukey 1958), and permutation (Edginton 1980; Fisher 1935) techniques create sampling distributions from the sample data themselves. In a permutation test of, say, a correlation, the observed correlation is compared to the correlations obtained on randomly permuted versions of the data. The proportion of times the randomly permuted data yield a correlation as large as the observed correlation is the p-value of the test, and indicates the probability of getting such a large value even when the variables are known to be independent of each other (because they were constructed by randomly permuting the real data). Most contemporary statistical packages now offer bootstrap and jackknife options, and UCINET (Borgatti et al. 2002) offers permutation tests. 7. We may go further than mere statistical analysis to try to determine why and how the patterns and variability we found might have come into being, why and how they persist over time, and why and under what circumstances they may change. We aim to explain variability in culture and behavior as a function of variability in experience, and search for concrete events and circumstances that shape those experiences. We also understand that what people think and do reflect not only their individual life history, but broader regional and global histories of people, events, and social interaction into which they were born and in which they grew up. So we try to identify events, circumstances, and processes that provide one set of choices to some people and a different set of choices to others. We ask individuals to identify life experiences that were significant to them, and to help us understand why those experiences were significant. Similarity coefficients and statistical tests applied to data such as these can tell us what goes with what and how strongly, and thus provide a warrant for believing that a given relationship is real, not merely a figment of our imagination (McEwan 1963). Handwerker and Crosbie (1982), for example, confirmed (like many earlier studies) that the relationship between gender and social dominance (men tend to be dominant over women) is real. Unfortunately, a statistical analysis of the relationship between only two variables (zero-order analysis) doesn’t allow us to control for additional effects that could be causing the correlation between to the two variables. For example, ice cream sales per day is correlated with the number of drownings, but the relationship is not causal. Controlling for temperature in the analysis completely removes the correlation between ice cream sales and drownings: On hot days, people buy more ice cream, and on hot days more people go swimming. Research that goes beyond cultural description thus raises internal validity issues that go beyond the simple problem of measurement error mentioned earlier. The warrant for inferring that a relationship is determinant as well as real (McEwan 1963) requires us to isolate a suspected relationship so we can tell whether or not it exists, and its strength, when we rule out the other internal validity possibilities. Experimental research designs (e.g., Campbell and Stanley 1966; Cook and Campbell 1979) were created to do just that. Although the study of cultures rarely allows for the use of experimental designs, experimental design principles apply to all research. Multiple regression techniques provide ways to statistically control for variables that we

14_107-Bernard.indb 524

6/9/14 10:03 AM

16: Reasoning with Numbers

525

have no direct control over, as when we do field studies rather than lab experiments. For example, suppose we are studying the effectiveness of a medical treatment but are aware that people’s “will to live” can affect their health outcomes. In an experiment, we can control for people’s desire to get well by randomly assigning who gets the treatment. In a field study, we can only study the difference between those who chose to take the treatment and those who didn’t, but at least we can ask them questions to measure their desire to get well and use multiple regression to statistically control for it. Handwerker and Crosbie (1982) used a form of multiple regression called path analysis to show that, although the relationship between gender and dominance was real, it wasn’t determinant: Social dominance followed from control over access to resources and, independently, from relative size. Larger people tended to dominate smaller people. Men tend to dominate women only because, generally, men are larger than women. But just as larger women tend to dominate smaller women and larger men tend to dominate smaller men, larger women tend to dominate smaller men. 8. And there’s more! Decisions bearing on whether or not a test designed for one population (e.g., a test for disability among middle-class Euro-American children) can be validly applied to another (e.g., children living in Puerto Rico) require a distinctive set of test item analytical procedures (e.g., Gannotti 1998). Evolutionary and developmental change—in which phenomenal properties at one point in time (t1) are contingent on phenomenal properties at an earlier point in time (t0)—can be evaluated with Guttman scaling analysis (e.g., Carneiro 1962; Goodenough 1944, 1963). Linear programming operations (e.g., Chvatal 1983) provide tests for optimization hypotheses. When we change the analytical question from How many to What’s the likelihood, as we do when we analyze binary or qualitative dependent variables, logistic regression (Hosmer and Lemeshow 1989) substitutes for ordinary regression to provide us modeling capabilities sensitive to situational, historical, and regional contingencies. Time series (see, e.g., Gujarati 1995) and event-history analyses (e.g., Yamaguchi 1991) require specialized numerical methods. We need a distinctive set of numerical methods even to conceptualize the population processes and demographic characteristics so central to an understanding of human–environmental relationships and much cultural evolution and social change (e.g., Handwerker 1990). And there’s still more! Quadratic assignment procedure (QAP), k-means cluster analysis, 3-D multidimensional scaling plots, Fourier blob icon analysis of time-series, perceptual mapping analysis, and network analysis are only a few of the numerical reasoning methods that answer important analytical questions. The development of Kullbak-Liebler information measures (see Burnham and Anderson 2001; Richard 2005) makes possible comparative tests of alternative models that allow quantitative judgments about model adequacy that cross method boundaries (e.g., logistic regression and ordinary regression). WHAT NUMERICAL ANALYSIS DOES

Numerical methods thus constitute nothing more than explicit tools of data collection and analysis that address core research questions: “Did we get it right?” and “To whom, if anyone, can we validly generalize?” Different numerical techniques answer different questions. Numerical reasoning consists of answering important questions about cultures with the appropriate numerical method. The processes by which we construct our understanding of the world integrate sensory information with information from various forms of memory and, necessarily, create

14_107-Bernard.indb 525

6/9/14 10:03 AM

526

W. Penn Handwerker and Stephen P. Borgatti

new things (Handwerker 2009). Because the process means, at the level of consciousness, making guesses, we can’t help but make mistakes. Because each of us can draw only on our own limited prior experience, we don’t see the world perfectly; we rely heavily on others to find our errors. Even when we have excellent advice and all the information necessary to see clearly, we make silly decisions, as Barbara Tuchman documented in her 1987 book The March of Folly, beginning with the prototype decision by the Trojans to accept a gift horse filled with Greek warriors. Even when we don’t make silly decisions, we consistently fail to make rational choices, as real-world choices depend heavily on availability heuristics, confirmation biases, evaluations based on likeness rather than likelihood, overestimates of the likelihood of rare events, and many forms of “irrelevant” information. Why, then, should we assume a particular way of thinking about the world of experience is not merely a figment of one’s imagination, that a label defined in a particular way, that means used to distinguish one phenomenon from another (i.e., measurement), and that claimed relationships between one phenomenon and any other(s) correspond with the world of experience in some meaningful way? As Campbell (1970) observed, research consists of a search for ways to distinguish mental constructions that are largely fantasy from constructions that consist of a bit less fantasy. Data collection boils down to an active search for the errors that must pervade the constructs you use to formulate questions and make observations, the observations you make, and the error or bias attached to the times and places and people from whom you collected those data. Iterative data collection distinguishes good research from bad. Design each observation and question to test at least one part of your growing theoretical understanding. Note errors. Ask for clarification. Rethink the theory. Link micro-level observations and interviews with historical records and macro-level trends that only time-series data can reveal. Try again. Explicit numerical reasoning thus greatly facilitates finding judgment errors. Better, it allows us to see things we’d otherwise miss (see Table 16.1). NUMERICAL ANALYSIS FOR THE STUDY OF CULTURES

Conventional research methods or statistics courses will introduce you to the questions we summarized above. However, such an introduction is not enough for studying cultures. Why? Because to analyze the mental constructions and behavior shared among people—culture—we often have to turn those techniques on their side. Cultures consist most usefully of the shared norms that tell us what we should and should not do, the shared assumptions that generate those norms, and the patterns of behavior that may or may not correspond to those norms (Handwerker 2009). Cultures thus inescapably embody spatial and temporal autocorrelation, meaning that what one cultural participant does or tells you will correspond closely to what another participant in the same culture does or tells you. The errors you make in predicting what one cultural participant will do or say will correspond closely to the errors you make predicting what another cultural participant will do or say. This means that a random sample of people does not constitute a random sample of culture. The culture of an individual consists of configurations of cognition, emotion, and behavior that intersect in multiple ways with the culture of other individuals.

14_107-Bernard.indb 526

6/9/14 10:03 AM

Table 16.1. Correspondence between Selected Numerical Analysis Tools and Questions Anthropologists Try to Answer RESEARCH QUESTIONS

NUMERICAL ANALYSIS TOOLS

Who agrees with whom about what and to what degree?

What’s the agreement about?

What’s there to be explained?

Who (what) acts (looks) like whom (what) and to what degree?

What goes with what and to what degree?

Can we see a suspected relationship even after we control for everything else we can think of?

Consensus analysis Multidimensional scaling (MDS) Quadratic assignment (QAP) Similarity and dissimilarity coefficients Cluster analysis Correspondence analysis Perceptual mapping/ PROFIT analysis Factor analysis Simple summary statistics Numerical transformations Graphical analysis tools for single variables Graphical analysis tools for two or more variables Survival analysis Ordinary Least Squares (OLS) multiple regression Logistic multiple regression Probabilities

14_107-Bernard.indb 527

6/9/14 10:03 AM

528

W. Penn Handwerker and Stephen P. Borgatti

Hence, random samples of individuals will yield a random sample of the intersecting configurations of cognition, emotion, and behavior (i.e., the cultures) in a population. But random samples (defined by case-independence) of cultural phenomena (which necessarily contain case-dependence) cannot exist: They constitute mutually exclusive alternatives. To identify and describe cultures, we must accurately characterize spatial, temporal, network, and social autocorrelation. Accurate and precise characterizations depend on samples designed to actively search for cultural variation that comes from specific forms of variation in life experiences. Sample size depends on the degree of similarity among cases. Sample designs for the study of cultures (Handwerker 2005) thus don’t require large sample sizes and don’t depend on random selection. Useful sample designs stratify the population by contrasting life experiences that may produce cultural differences, employ judgmental selection of key informants and critical cases, and select other cases based on their availability, either out of availability (convenience) or through a snowball procedure. Sample size for specific strata is set by quota, depending on the average level of agreement. Efficient sample designs track levels of agreement and expand sample sizes and change stratification criteria consistent with levels of agreement and identified cultural boundaries. Because the study of cultures depends on the analysis of many individuals simultaneously, analysis depends heavily on techniques that clarify the structure embedded in huge amounts of information that would otherwise just confuse us. Factor analysis, for example, constructs a small set of variables (factors or principal components) from additive combinations of existing similarities among variables. Each factor represents an inference of the existence of one or more underlying, unmeasured variables that might explain the observed pattern of similarities. In other words, a set of variables (say, measuring attitudes toward government policies) may be highly correlated because all are caused by an underlying factor (such as an underlying liberal or conservative orientation). In a factor analysis, the factors are sorted so that the first factor accounts for the maximum amount of correlations in the original data matrix. The second factor accounts for the maximum amount of the remaining variance not explained by the first factor, and so on. Factor analysis output includes correlations between each factor and variables, called loadings. The square of a loading tells you how much variability a case or variable shares with the unobserved variable represented by each factor. The sum of squared loadings for a factor (its eigenvalue or latent root) tells you how much variation all your cases or variables share with a factor. A factor’s eigenvalue divided by the sum of eigenvalues for all factors expresses, as a proportion, the factor’s explanatory power. Consensus analysis (e.g., Borgatti and Halgin 2011; see Boster 1981, 1987, 1988; Romney et al. 1986; Weller and Romney 1988) uses factor analysis to answer what may be the single most important question in the study of culture: Who agrees with whom about what and to what degree? However, the factor analysis done in consensus analysis turns the data on their side and regards the cases as variables and vice versa. Consensus analysis analyzes the pattern of agreements between persons and searches for underlying factors, representing cultures, that explain their responses. A very large first factor signals the existence of a single culture underlying the responses of each individual. This doesn’t mean the individuals necessarily agree with each other, but rather that the pattern is such that disagreement is explained by the

14_107-Bernard.indb 528

6/9/14 10:03 AM

16: Reasoning with Numbers

529

distance each person is from the culture center. Two people at the center agree on virtually everything. Two people at the periphery often disagree, but not systematically, and both will tend to agree more with people at the center than with each other. When several factors are found that explain large amounts of variance, it identifies the existence of several different cultures within our sample. It means that people are agreeing and disagreeing with each other in systematic ways that reflect their allegiance to different underlying factors (cultures). UCINET consensus analysis procedures produce output that identifies the relative importance and explanatory power of factors and an answer key based on factor loadings corrected for guessing, which tells us what the agreement is all about. Low or negative factor loadings signal cultural differences arising from life experiences that differ in significant ways. Diagnostic analysis can tell us whether low or negative factor loadings mean a weak consensus with generally low levels of agreement, subpopulation differences in the level of agreement (e.g., Kennedy et al. 2012), or the existence of alternative or opposed subpopulation agreements about the domain or some domain items expressed, possibly, as a cultural cline (e.g., Fuentes 2008). Correspondence analysis (Greenacre 1984; Weller and Romney 1990) tells us not only who agrees with whom but also the extent to which who agrees with whom about what. William Dressler has pioneered the application of consensus analysis to important questions in applied anthropology. In particular, see his construction of regionally and historically specific measures of poverty (see Dressler 1996, 1997; Dressler et al. 1996). The most common measures of the poverty construct define it by reference to an income level presumed to constitute the minimum necessary for food, clothing, and shelter and, thus, presume a narrow biological definition of need. A measure of poverty more germane to understanding culture and behavior might consist of a measure of relative deprivation in lived experiences (e.g., Aberle 1966). The critical question is to what is poverty/deprivation relative? In both Brazil and the rural South, Dressler used consensus analysis to identify a widespread cultural agreement about what constituted success in life. The resulting measure of cultural consonance encompasses the lived experience of poverty with its multiple dimensions—income, education, age, marital status, occupation, and social support and integration. This approach thus offers a new, explicit tool for understanding the sources, patterning, and implications of disadvantage in human communities. CONCLUSION

Unaided, our senses tell us about surface phenomena—the net effects of chance and complexly textured multidimensional influences on what we think, feel, and do that operate all the time. Built-in perceptual biases, moreover, make it difficult to distinguish real structure from spurious patterns. For example, suppose that the length of time it takes for a child to get to school is a function of many factors along the route, including traffic, red lights, road construction, and so on. By chance, everything could go wrong and the child arrives very late; most of the time, though, some things go well and others don’t, so the child arrives on time. Now suppose the teacher scolds children when they are egregiously late. Because extreme lateness requires a number of things to go wrong at the same time, it is unlikely to happen again the next day. But since the

14_107-Bernard.indb 529

6/9/14 10:03 AM

530

W. Penn Handwerker and Stephen P. Borgatti

teacher is probably not conducting a statistical analysis, he or she is very likely to think that the reason was the scolding, which was evidently efficacious. In ordinary life, such misattribution of chance events may be of little consequence. But when we are doing research to find out how the world works, we must use numeric methods to shield us from spurious correlations. Numerical methods help us identify chance fluctuations in what we see, detect the subtle configurations of meaning that remain unspoken, and, perhaps, unrealized, by our informants, both identify and assess the relative importance of the multidimensional historical, regional, situational, and individual influences on meaning and behavior. Explicit reasoning with numbers thus lets us see things in and about people’s lives we’d otherwise miss and answers questions in ways that help us pinpoint those errors in judgment that make us quintessentially human. REFERENCES

Aberle, D. F. 1966. The peyote religion among the Navaho. Chicago: Aldine. Borgatti, S. P., M. G. Everett, and L. C. Freeman. 2002. UCINET for Windows: Software for social network analysis. Harvard, MA: Analytic Technologies. Borgatti, S. P., and D. S. Halgin. 2011. Consensus analysis. In A companion to cognitive anthropology, ed. D. B. Kronenfeld, G. Bennardo, V. C. de Munk, and M. D. Fischer, 171–90. Malden, MA: Blackwells. Boster, J. S. 1981. How the exceptions prove the rule. Ph.D. diss., University of California, Berkeley. Boster, J. S. 1987. Introduction. American Behavioral Scientist 31: 150–62. Boster, J. S. 1988. “Requiem for the omniscient informant”: There’s life in the old girl yet. In Directions in cognitive anthropology, ed. J. Dougherty, 177–97. Urbana: University of Illinois Press. Burnham, K. P., and D. R. Anderson. 2001. Kullbak-Liebler information as a basis for strong inference in ecological studies. Wildlife Research 28: 111–19. Campbell, D. T. 1970. Natural selection as an epistemological model. In A handbook of methods in cultural anthropology, ed. R. Naroll and R. Cohen, 51–85. Garden City, NY: Natural History Press. Campbell, D. T., and J. C. Stanley. 1966. Experimental and quasi-experimental designs for research. Chicago: Rand, McNally. Carneiro, R. L. 1962. Scale analysis as an instrument for the study of cultural evolution. Southwestern Journal of Anthropology 18: 149–69. Chvatal, V. 1983. Linear programming. San Francisco: Freeman. Cook, T. D., and D. T. Campbell. 1979. Quasi-experimentation: Design & analysis issues for field settings. Boston: Houghton Mifflin Co. Dressler, W. W. 1996. Culture and blood pressure: Using consensus analysis to create a measurement. Cultural Anthropology Methods (CAM) 8(3): 6–8. Dressler, W. W. 1997. Culture and patterns of poverty. Paper delivered at the 71st annual meeting of the Society for Applied Anthropology, Seattle. Dressler, W. W., J. E. Dos Santos, and M. Campos Balierio. 1996. Studying diversity and sharing in culture: An example of lifestyle in Brazil. Journal of Anthropological Research 52: 331–53. Edginton, E. S. 1980. Randomization tests. New York: Marcel Dekker. Efrom, B., and R. Tibshirani. 1991. Statistical data analysis in the computer age. Science 253: 390–95. Fisher, R. A. 1935. The design of experiments. London: Oliver & Boyd.

14_107-Bernard.indb 530

6/9/14 10:03 AM

16: Reasoning with Numbers

531

Fuentes, C. 2008. Pathways from interpersonal violence to sexually transmitted infections: A mixed-method study of diverse women. Journal of Women’s Health 17: 1591–603. Gannotti, M. E. 1998. The reliability and validity of the PEDI in Puerto Rico. Ph.D. diss., University of Connecticut. Goodenough, W. H. 1944. A technique for scale analysis. Educational and Psychological Measurement 4: 179–90. Goodenough, W. H. 1963. Some applications of Guttman scale analysis to ethnography and culture theory. Southwestern Journal of Anthropology 19: 235–50. Greenacre, M. J. 1984. Theory and applications of correspondence analysis. New York: Academic Press. Gujarati, D. N. 1995. Basic econometrics. New York: McGraw-Hill. Handwerker, W. P. 1990. Demography. In Medical anthropology, ed. T. M. Johnson and C. F. Sargent, 319–47. New York: Praeger. Handwerker, W. P. 2005. Sample design. In Encyclopedia of social measurement, ed. K. KempfLeonard, 117–30. New York: Elsevier. Handwerker, W. P. 2009. The origin of cultures. Walnut Creek, CA: Left Coast Press, Inc. Handwerker, W. P., and P. V. Crosbie. 1982. Sex and dominance. American Anthropologist 84: 97–104. Hosmer, D. W., and S. Lemeshow. 1989. Applied logistic regression. New York: John Wiley. Kennedy, D. P., R. A. Brown, D. Golinelli et al. 2012. Masculinity and HIV risk among homeless men in Los Angeles. Psychology of Men & Masculinity. Advance online publication. doi: 10.1037/a0027570. McEwan, Peter J. M. 1963. Forms and problems of validation in anthropological research. Current Anthropology 4: 155–83. McNabb, S. 1990. The uses of “inaccurate” data. American Anthropologist 92: 116–29. Mooney, C. Z., and R. D. Duval. 1993. Bootstrapping. Quantitative Applications in the Social Sciences Series, Vol. 95. Newbury Park, CA: Sage. Richard, S. A. 2005. Testing ecological theory using the information-theoretic approach. Ecology 86: 2805–14. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus. American Anthropologist 88: 313–38. Tuchman, B. W. 1987. The march of folly. New York: Random House. Tukey, J. W. 1958. Bias and confidence in not quite large samples. Annals of Mathematical Statistics 29: 614. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Qualitative Research Methods Series, Vol. 10. Newbury Park, CA: Sage. Weller, S. C., and A. K. Romney. 1990. Metric scaling: Correspondence analysis. Quantitative Applications in the Social Science Series, Vol. 75. Thousand Oaks, CA: Sage. Yamaguchi, K. 1991. Event history analysis. Applied Social Research Methods, Vol. 28. Newbury Park, CA: Sage.

14_107-Bernard.indb 531

6/9/14 10:03 AM

14_107-Bernard.indb 532

6/9/14 10:03 AM

CHAPTER 17

Text Analysis Amber Wutich, Gery Ryan, and H. Russell Bernard

There is enormous interest across the social sciences in the systematic analysis of text. Little wonder: Most of the recoverable data about human thought and human behavior are texts of one kind or another. In this chapter, we survey methods of text analysis in the social sciences and discuss how anthropologists in particular have used those methods to look for meaning and pattern in written text. We cover two broad types of text analysis: the linguistic tradition, which treats text as an object of analysis itself, and the sociological tradition, which treats text as a window into human experience. For the linguistic tradition, we review how anthropologists have collected and produced texts, analyzed indigenous literatures, discovered patterns and structures in performance styles, and compared the production of narratives within and across cultures. (Another major tradition grounded in linguistics is the analysis of discourse, or language in interaction. See the chapter by Graham and Farnell, this volume.) For the sociological tradition, we review the methods of schema analysis, grounded theory, classical content analysis, and analytic induction. We also discuss the ways in which computer programs facilitate these analyses and make light work of labor-intensive methods like word-based analysis and semantic network analysis. Throughout, we focus on methods for collecting and analyzing written texts such as political speeches, song lyrics, personal diaries, transcriptions of interviews, newspaper editorials, and so on. Many of these methods, however, serve just as well in dealing with images, such as photographs, video, commercial movies, kinescopes of old television shows, and so on. And the methods we review span the inductive-to-deductive continuum. Inductive approaches, like analytic induction, are used to identify and develop theories from on-the-ground observations. Deductive approaches, like classical content analysis, typically are used to test hypotheses. Both approaches have much to add to the advancement of theory. THE LINGUISTIC TRADITION

Anthropology: A Passion for Collecting Texts Debates about the value of structuralism, functionalism, historical particularism, materialism, and postmodernism come and go, but the value of faithfully produced texts is undisputed. Among Franz Boas’s lasting contributions is the corpus of texts he collected, translated, and published (or deposited in archives) from speakers of Bella Bella (Boas 1928b), Sahaptin and Salishan (Boas 1917), Keresan (Boas 1928a) and, with George Hunt, Kwakiutl (Boas and Hunt 1902–1905, 1906; see also Boas 1910, 533

14_107-Bernard.indb 533

6/9/14 10:03 AM

534

Amber Wutich et al.

1935–1943). In 1893, Boas taught Hunt to write Kwakiutl, Hunt’s native language. By the time Hunt died in 1933, he had produced 5,650 pages of text—a corpus from which Boas produced most of his reports about Kwakiutl life (Rohner 1966). To the extent possible before the invention of voice-recording devices, Boas trained his students and collaborators to collect verbatim texts. Following Boas’s example with Hunt, Paul Radin worked with Sam Blowsnake, a Winnebago. Blowsnake wrote the original manuscript (in Winnebago) that became (in translation) Crashing Thunder: An Autobiography of a Winnebago Indian (Blowsnake and Radin 1983 [1920, 1926]). Among Boas’s other students, Swanton (1909), Goddard (1911), Kroeber (1907), Lowie (1930), and Sapir (Sapir and Curtain 1974 [1909]; Sapir and Dixon 1910; Sapir and Swadesh 1978 [1939]) collected and analyzed indigenous language text, and Margaret Mead and Gregory Bateson produced hours and hours of cinema verité about Bali dance (see the South Pacific Ethnographic Archives, U.S. Library of Congress)—a rich, textual record that can be turned to again and again as new insights and new methods of analysis become available. The concern for the collection and archiving of text remains undiminished in linguistic anthropology. In the 1970s, Eric Hamp edited the Native American Tests Series of the International Journal of American Linguistics (see, e.g., Bernard and Salinas Pedraza 1976; Furbee-Losee 1976). As the literary language of a pre–colonial-era civilization, Maya has attracted particular attention. Munro Edmonson’s translations of the books of Chilam Balam (1982, 1986) and Dennis Tedlock’s translation of the Popol Vuh (1985) are outstanding examples. Following the example of Boas with Hunt and Radin with Blowsnake, Bernard (Bernard and Salinas Pedraza 1989; Salinas Pedraza and Bernard 1978), El Guindi (El Guindi and Hernández Jiménez 1986), Lurie (1961), and Sexton (Bizarro Ujpán and Sexton 1981, 1985, 2001), among others, have helped indigenous people create narratives themselves in indigenous languages. Bernard (1997) provided indigenous people with the computer technology and training to produce books in their own languages. Here again, the emphasis is on the production of texts, not on their analysis (see Salinas Pedraza 1997). Computing resources facilitate the production of texts by native speakers of previously nonliterary languages. González Ventura (1997), the first contemporary Mixtec to produce a book in Mixtec, discussed the challenges and possibilities presented by the production of indigenous-language text by computer. Numerous language revitalization programs have undertaken the task of cataloguing and making nativelanguage texts available online (Eisenlohr 2004). Massachusett was no longer spoken by the mid-1800s, but members of the Wampanoag nation have reconstructed the language through the analysis of historical documents. They have also begun to translate English-language Wampanoag texts such as the Wampanoag Prayer into Massachusett (Jennings and O’Brien 1998). The Archive of the Indigenous Languages of Latin America, developed by Joel Sherzer and colleagues at the University of Texas (see AILLA 2012), has brought together texts, images, and recordings of hundreds of languages. Together, these texts constitute a widening and increasingly available corpus of new data ready to be analyzed.

14_107-Bernard.indb 534

6/9/14 10:03 AM

17: Text Analysis

535

Analysis of Indigenous Literature Some 389 languages in the world are spoken by a million people or more (Lewis 2009). Some, like Uyghur (about 8 million speakers), Kurdish (about 9 million speakers), and Yoruba (about 19 million speakers), have active literary traditions. Others, like Quechua (about 10 million speakers), have very few literary works. Almost all the known literary works in Somali (14 million speakers) are from the last 50 years (Andrzejewski 2011). Indigenous literary traditions, however, continue to emerge. Māori, for example, has perhaps 60,000 native speakers, but was named a national language in Aotearoa/New Zealand in 1987 and is developing quickly as a literary language. These developments, too, produce text for anthropological analysis. Postel-Coster (1977), for example, drew on an indigenous tradition of novels in Indonesia that began with western Sumatran, or Minangkabau, writers in the 1920s. He saw novels as the modern continuation of the myth in nonliterary societies, rather than a source of factual information. Novels dealing with young couples’ struggles to marry, he said, reveal “problems of Minangkabau culture” such as “matrilineal succession, polygynous marriage, the enormous impact of the extended family on an individual’s life, and the question of merantau—the traditional emigration of young people” (1977, 137). Besnier (1995) studied an indigenous literacy tradition on the Polynesian atoll of Nukulaelae. The texts produced by the people of Nukulaelae include letters, sermons, and announcements of events. Besnier analyzes correspondence by identifying and presenting exemplars of structural regularities. For example, the main body of letters on Nukulaelae: usually begins with a greeting identical to the greeting used in face-to-face interactions (taalofa “hello,” a Samoan borrowing). This is followed by references to the health of everyone at the writer’s and recipient’s ends, and sometimes by a very long series of invocations to God’s grace and kindness. (Besnier 1995, 86)

Besnier follows this with a series of exemplars. Besnier noticed that letter writers: always adhere to a religious reference scheme in opening themes, Christian for the majority, or Baha’i, etc., for the handful of religious converts. In letters written by younger people, the introduction is usually much shorter and more predictable in content than that of letters written by older individuals. These introductions bear many similarities to the beginning of formal speeches. (1995, 87)

Using literary methods of analysis, then, Besnier identifies covariation between certain elements of style in text and independent variables like religious affiliation and age.

Patterns in Performance The discovery of regularities in narrative performance is achieved mostly through the analysis of written text. The work of Dell Hymes is of singular importance. Based on his analysis of texts from speakers of Shoalwater, Kathlaket, and Clackamas Chinook (collected by Boas, Sapir, and others over 80 years), Hymes (1977, 431) found

14_107-Bernard.indb 535

6/9/14 10:03 AM

536

Amber Wutich et al.

that the narratives are “organized in terms of lines, verses, stanzas, scenes, and what many call acts.” He argued further that the three languages share a common “fabric of performance style” and “form of poetic organization” (Hymes 1977, 431). This was a truly significant discovery, for it made clear that Native North American texts have something to contribute to a general theory of poetics and literature. In a series of publications (1976, 1977, 1980a, 1980b, 1981), Hymes showed that historical and contemporary texts of Native American narrative performances are organized into verses and stanzas and form groups of either fives and threes or fours and twos. Virginia Hymes’s analysis (1987, 67–68) of Dell Hymes’s comparative techniques makes clear the importance of “working back and forth between content and form, between organization at the level of the whole narrative . . . lines within a single verse . . . even words within a line.” Hymes’s approach revealed that the method of Boas and others, which organized Native American narratives into lines of text, hid from view “a vast world of poetry waiting to be released by those of us with some knowledge of the languages” (V. Hymes 1987, 65). Boas and his students did not document narrative organization, but they did innovate conventions for marking paralinguistic features of performance such as nonphonemic vowel length and stress—an uncommon practice among linguists of the time (Hymes 1977, 452–53). In his study of Zuni narratives, Tedlock (1972, 221) found that such paralinguistic features as voice quality, loudness, and pausing are key indicators of performance. Later, using a sixteenth-century Quiché Maya manuscript called the Popol Vuh, Tedlock (1987) demonstrated that an analysis of paralinguistic features among modern performers can reveal how ancient texts were meant to be narrated. As was true of many medieval manuscripts in Europe, the Popol Vuh had been written as an undifferentiated mass of text with almost no punctuation—that is, there are no clues about how a performer might have used techniques such as intonation, timing, or pauses. Tedlock’s solution was to capture oral narratives—speeches, prayers, songs, stories—from modern speakers of Quiché and look for patterning that hinted at how the Popol Vuh may have been enunciated. Additionally, Tedlock relied on Andrés Xiloj, a native speaker who learned to read Quiché, to narrate the Popol Vuh following the dictates of modern Quiché performance. Tedlock devised conventions for marking pauses, accelerations, verse endings, and so on, and used this new text to determine that Quiché verse has the same structure as ancient Middle Eastern texts—texts that predate Homer. Indeed, he concluded that the same structure is found in all living oral traditions that have not yet been influenced by writing. This is a contribution to a general theory of poetics and literature of the sort that Hymes had envisioned a decade earlier for ethnopoetics. A key method of text analysis in ethnopoetics is text presentation. In his presentation of The Hot Pepper Story as told in Kuna by Chief Mastayans, Sherzer (1990, 178) used a highly literal translation. The text repeats a small number of words and themes, and Sherzer felt that a more liberal translation would fail to capture the poetics of performance. So, Sherzer describes the thematic elements he sees in the text but uses the device of literalness in the translation to draw the reader’s attention to those elements. In later work on a Kuna gathering-house chant, Sherzer (1994) worked with

14_107-Bernard.indb 536

6/9/14 10:03 AM

17: Text Analysis

537

his assistant, Alberto Campos, to produce a Kuna transcription and Spanish translation. By studying Campos’s translation against the original Kuna, Sherzer found that verse breaks are determined by a regular melodic shape and by turn-taking between the narrator, Chief Olopinikwa, and the responding chief. In his presentation of the performance, Sherzer uses the convention of beginning verses and lines flush on the left of each page and indenting the lines of the responding chief. Anthropologists are constantly experimenting with methods for presenting text of indigenous performance that capture the subtleties of performance. In translating Ñähñu (Otomí) parables, folktales, and jokes, Bernard and Salinas (1976) presented a fully literal translation and a fully liberal translation, in addition to a transcription of the Ñähñu. At the time, Bernard felt that there was no way to mediate between the characteristics of the original, free Ñähñu and a free English translation. Later, in translating Salinas’s four-volume ethnography of the Ñähñu, Bernard tried a middle course—one in which the English is grammatical but also one that makes clear from the style that it is a translation (see Bernard and Salinas 1989). Text analysis produces new text, which can, in turn, be analyzed.

Intra- and Intercultural Comparisons of Narratives Understanding differences in how members of a culture construct and relate narratives has long been of interest to anthropologists. Quinn’s (2005) edited volume Finding Culture in Talk explores different methods for cultural analysis of discourse. For example, Mathews (1992, 2005) collected 60 tellings of La Llorona (Weeping Woman), a morality tale told across Mexico. Mathews focuses on two versions of the tale: one (typically told by men) in which an irresponsible wife commits suicide, freeing her husband from a bad marriage, and a second (typically told by women) in which a mistreated wife commits suicide and her ghost goes on to lead bad men to their deaths (thus freeing their wives and children). Mathews developed a “story grammar” to examine how “the motives of the main characters draw upon culturally shared schemas about gendered human nature” (1992, 129). Her analysis explores the ways in which men and women use “the tales to communicate with each other about contested views of male dominance” (2005, 151). Herzfeld (1977) studied variations in renditions of the khelidonisma, or swallow song, sung in Greece to welcome spring. He collected song texts from ancient, medieval, modern historical, and contemporary sources across Greece. His purpose was to show that inconsistencies in the texts emerge from the structural principles that underlie the rite de passage for welcoming spring in rural Greece. To make his point, Herzfeld looked for anomalies—like “March, my good March” in one song compared to “March, terrible March” in another. Herzfeld claims that good refers ironically to a source of anxiety. Herzfeld argues there is strong evidence that March is a source of anxiety for Greek villagers, such as the avoidance of certain activities during the drimata (the first three days of March, a transition from winter to summer) and the association of the drimes (the first three days of August, a transition from summer to winter) with malevolent spirits. Herzfeld concludes that there is symbolic danger associated with these transition months. His approach to symbolic analysis requires deep involvement with the culture, including an intimate familiarity with the language, so

14_107-Bernard.indb 537

6/9/14 10:03 AM

538

Amber Wutich et al.

that the symbolic referents emerge during the study of those expressions. You can’t see the connections among symbols if you don’t recognize the symbols or what they mean. In addition to intracultural variants in the composition of narratives, analyses may examine variation in the structuring of narratives, such as temporal sequence in conversational narratives (Hill 2005). For example, Ochs and Capps (2001), in their book Living Narrative, examine linearity as a key dimension in the construction and analysis of narratives. Drawing on an analysis of quotidian narratives, they illustrate how linearity can range from closed to open temporal and causal order. This approach has been used to examine linearity in adolescent life histories (Habermas et al. 2009) and schizophrenia patients’ narratives of emotional events (Gruber and Kring 2008). Furbee (1996, 2000) studied a new cult originating in Lomantán, a Tojolab’al Maya village in Chiapas, Mexico. According to the local story, Dominga Hernández was cutting wood in 1994 when God appeared and gave her religious images, including the Christ Child, the Virgin Mary, Saint Joseph, and animals of the crèche, to care for and charged her with building a new church. Within 43 days, Hernández had mobilized the financial support and donated labor needed to build the church that now houses the images. Furbee collected 26 Tojolab’al language accounts of the Lomantán miracle from 18 villages, including 8 that were loyal to the PRI (then the ruling political party) and 10 that were loyal to the PDR (an opposition party). Furbee argues that the use of Spanish terms is associated with the authoritativeness of the speaker—given that the dominant Spanish language signals power—and the speaker’s political alignment with the ruling party. The relative absence of Spanish loan words among speakers from the PDR villages, which were sympathetic to Zapatistas, is “just what one might predict from those who oppose the prevailing hegemony and who are engaged in a revitalization movement” (Furbee 1996, 13). Bletzer and Koss (2006) compared the narratives of 62 impoverished women in the United States who had survived rape, including 25 Cheyenne women, 24 Anglo women, and 13 Mexican American women. Bletzer and Koss analyzed the narratives for similarities and differences in themes and in narrative structures. For example, Anglo and Mexican American women expressed thoughts of revenge against their assailants, but none of the Cheyenne women did (Bletzer and Koss 2006, 17). Chafe (1980) and five colleagues hired a professional filmmaker and produced a simple, wordless film called The Pear Story. They showed it to speakers of English, Chinese, Japanese, Malay, Greek, Thai, German, Haitian Creole, Persian, and Sacapultec Maya. Participants were asked to tell the story to a third person within 30 minutes of viewing the film. Those stories were transcribed verbatim, including pauses, pause filters, stutterings, and such. Based on that research, Chafe (1980) identified the existence of idea units (about six seconds long and containing about six words) commonly marked by intonation that involves a rise in pitch or a fall. Chafe suggested “that these idea units, these spurts of language, are linguistic expressions of focuses of consciousness” (1980, 15) that are packaged into sentences. The Pear Story method has been used to generate narratives for comparison in linguistics (Chappell 1988; Chui and Lai 2008; Maroney 2004; McGregor 2006) and in a range of cultural contexts (e.g., Erbaugh 1990; Orero 2008).

14_107-Bernard.indb 538

6/9/14 10:03 AM

17: Text Analysis

539

The Pear Story method was inspired by the pioneering work by F. C. Bartlett (1967 [1932]) on how people remembered and retold folktales, and that work was the basis for Bartlett’s development of schema analysis. THE SOCIOLOGICAL TRADITION

Schemas, Metaphors, and Mental Models Schema analysis is based on the idea that people use shortcuts or simplifications in their thinking to help them deal the complexities of reality (Casson 1983, 430). Bartlett’s (1967 [1932]) work on how people remember things suggested the existence of such simplifications or “building blocks of cognition” (Rumelhart 1980). In an influential book, Schank and Abelson (1977) postulated that schemas—or scripts, as they called them—enable culturally skilled people to make assumptions that fill in missing details of a story. We often say things like “Fred lost his data because he forgot to save his work.” We know that Fred’s forgetting to save his work didn’t actually cause him to lose his data. A whole set of links are left out, but they are easily filled in by any listeners who have the background to do so. Some schemas may be universal (e.g., Piaget’s [1970] findings on cognitive schemas in children), and some are surely personal and idiosyncratic. Somewhere between universal and idiosyncratic schemas are cultural schemas: They are developed through experience but are held by a population (D’Andrade 1995, 130; Quinn 2005, 38; Rice 1980, 154). Cultural schemas comprise shared rules for interpersonal behavior—rules that provide the kind of flexibility that allows for immediate changes of response as people react to one another’s actions (Ridgeway 2006). In this way, cultural schemas are analogous to generative grammars, where a finite and shared set of rules, operating on a finite and shared lexicon produces an infinite number of acceptable utterances (Chomsky 1957, 13). Analyzing cultural schemas requires close attention to linguistic and paralinguistic features such as metaphors, proverbs, repetitions, pauses, speaker transitions, turn taking, and interruptions (D’Andrade 1984, 1987, 1995). In her study of American marriage, Quinn (1982, 1987, 1992, 1996, 1997) exemplifies schema analysis in anthropology. Working with 11 married couples, she produced over 300 hours of recorded interview narratives. Quinn and her students searched this corpus of text for repeated words, phrases, and patterns of speech. They paid particular attention to metaphors and the underlying principles that might produce patterns in those metaphors. For instance, Quinn found that people talk about their surprise at the breakup of a marriage by saying that they thought the couple’s marriage was “like the Rock of Gibraltar” or had been “nailed in cement.” People use these metaphors because they assume that their listeners know that cement and the Rock of Gibraltar are things that last forever. Quinn’s emphasis on metaphor owes much to the work by Lakoff and Johnson (2003 [1980]). Using this approach, Quinn found that the hundreds of metaphors in her corpus of texts fit into just eight linked concepts (lastingness, sharedness, compatibility, mutual benefit, difficulty, effort, success [or failure], and risk of failure)—and that these concepts are linked together in a schema that guides the discourse of ordinary Americans about marriage.

14_107-Bernard.indb 539

6/9/14 10:03 AM

540

Amber Wutich et al.

Metaphors are not the only linguistic features used to infer meaning from text. Price (1987) observes that when people tell stories, they leave out information that everyone knows. Thus, in her study of 14 narratives of illness and misfortune in a Mestizo community in Ecuador, Price looks for what is not said in narratives (but is observable in ethnography) to identify underlying cultural assumptions (1987, 314). D’Andrade notes that the “repetition of associative links” may be “the simplest and most direct” sign of cultural schemas (1991, 294). For instance, in C. Strauss’s (1997) research into whether late capitalism has produced a fragmented, postmodern consciousness, she examines the associative links in one former factory worker’s views on being a workingman, having to pay a disproportionate share of taxes and struggling to get by. Other examples of inductive research engaged in the search for cultural schemas in texts include Holland’s (1985) study of Americans’ reasoning about interpersonal problems, Swartz’s (1997) study of Mombasa Swahili understanding of the role the four humors play in body function, and Izquierdo and Johnson’s (2007) study of envy and sorcery in Matsigenka illness narratives in the Peruvian Amazon. Garro combined schema analysis with cultural consensus analysis (Romney et al. 1986) in her studies of high blood pressure (1988) and diabetes (1995, 2000) within and across Anishinaabe communities in Canada. Garro (2000) argues that schema analysis “provides more insight into the nature of individual and cultural knowledge,” but “no explicit grounds for determining if something is shared” (p. 285). In contrast, consensus analysis “helps to corroborate claims about shared cultural understanding, while pointing out areas where there is less agreement” (p. 312). A similar approach was adopted by Handwerker and Wolfe (2010) in their study of cultural understandings of “bad teeth,” by Worthman et al. (2002) in their research on American family life, by Kempton et al. (1995) in their study of American environmental values, and by Stone-Jovicich et al. (2011) in their study of two groups in South Africa on opposing sides of intensive water use. Some cognitive scientists—including anthropologists—test schemas in experiments. Rice (1980) hypothesized that “comprehension of a story” would depend on its “assimilation to the schemata of the hearer” (p. 156). Using an experimental design, she presented Americans with two Eskimo stories, “Kayatuq and the Red Fox” and “Nakkayuq and His Sister,” in a range of formats that conformed more closely to either the American or Eskimo storytelling schema. Then she asked participants to recall certain passages from the stories. When the passages fit the American story schema, Rice found—as she predicted—that participants agreed about the events they remembered and recalled more exactly worded phrases than when the stories fit the Eskimo story schema. Schema analysis provides a powerful approach to determine how people think, where this thinking comes from, and what happens when someone thinks differently. In research on imprisoned sex offenders, Waldram (2008) observed inmates whose therapists required them to tell their life stories in a way that was consistent with a treatment schema focused on why people commit sex offenses. A prisoner who failed to construct a life history that conformed to this schema (e.g., by omitting drunkenness) became ineligible for early parole. Related methods, such as voice-centered analysis (Brown and Gilligan 1993) and personal semantic network analysis (C. Strauss 1992, 2005), help researchers compare respondents’ views to idealized (or schematic) understandings of how one should

14_107-Bernard.indb 540

6/9/14 10:03 AM

17: Text Analysis

541

experience the world. Examples of such research include Anderson-Fye’s (2003) exploration of how Belizean girls resist globalizing thin-body norms and Beauboeuf-Lafontant’s (2008) examination of how black American women’s experiences of depression interact with the dominant discourse about “strong black women.” Similar to schema analysis, a thread of research in the linguistics tradition of text analysis deals with the discovery of mental models through systematic examination of text (Carley and Palmquist 1992; Johnson-Laird 1983, 2010; Jones et al. 2011). Anthropologists participated early on in the development of what is today the crossdisciplinary field of cognitive science (for a review, see Tyler 1969) but interest in cognitive studies waned in anthropology after about 1980 (see Bender et al. 2010; Boster 2011). There has been a revival of interest among anthropologists in recent years in mental models and related topics as anthropologists work in teams with psychologists and other scholars (e.g., Atran and Medin 2008; Bang et al. 2007).

Grounded Theory Grounded theory is a set of techniques for inducing meaning—concepts, connections, models—from people’s own words. The goal is to create new theories that predict people’s beliefs or behaviors (Glaser and Strauss 1967). Since its inception, grounded theory has flourished across the social sciences, but particularly in disciplines in which inductive studies are common, such as education (e.g., Blase 1982), nursing (e.g., Beck 1993), and management (e.g., Maznevski and Chudoba 2000). Inductive approaches have a long history in ethnographic case studies (Agar 1979, 1980, 1983; Becker et al. 1961). In fact, we find that grounded theory formalizes many of the techniques that ethnographers use—in-depth interviewing, purposive sampling, negative case analysis, constant comparison, identifying commonalities and differences across texts, and developing theories from on-the-ground experience—though ethnographers seldom discuss these methods explicitly. There are some controversies about the extent to which grounded theory should be radically inductive (Glaser 1992, 1999), allow some deduction (Corbin and Strauss 2007; Strauss 1987; Strauss and Corbin 1990), or treat elicited narratives as partial, co-constructed, and subjective (Charmaz 1990, 2006; see Boychuk Duchscher and Morgan [2004] for a comparison). Nevertheless, all these approaches share a commitment to understanding people’s views of themselves and the meaning they ascribe to their experiences and to using this knowledge to generate theory. As a method, grounded theory involves a process by which the analyst becomes increasingly “grounded” in the data, developing ever-richer concepts and models of how the phenomenon being studied really works. Typically, grounded theory studies begin with one or a few respondents in a target population. Markovic (2006), for example, studied women with gynecological cancer in Australia. She interviewed one woman she called Tipani and, reading the transcript line by line, identified in-vivo codes—that is, codes drawn from the respondent’s own words (Bernard and Ryan 2009; Bogdan and Biklen 1992; Lincoln and Guba 1985; Lofland and Lofland 1995; Sandelowski 1995a; Strauss and Corbin 1990; Taylor and Bogdan 1984). Markovic concentrated on repetitions of ideas in Tipani’s narrative, things like “I’ve had blood now and again” and “I started bleeding” to develop themes such as “physical symptoms” (2006, 416–17). Throughout this process, researchers keep detailed memos

14_107-Bernard.indb 541

6/9/14 10:03 AM

542

Amber Wutich et al.

that document their questions about the data, methodological decisions, and hunches about how things are connected (Montgomery and Bailey 2007; Strauss and Corbin 1990). For example, Markovic mentions making notes to document “the similarities and dissimilarities in the way women expressed each theme” (2006, 417). Once some emerging themes are identified, the next step is to select new respondents who can add novel ideas or connections to the data—a technique called theoretical sampling (Coyne 1997; Drauker et al. 2007). Markovic explains how her first interview with Tipani yielded basic themes and a set of hypotheses about the conditions under which immigrant and native-born women in Australia decide to seek professional medical help for symptoms of gynecological cancer. Then, with the specific hypotheses in mind, Markovic analyzed the transcript of her second interview, with Betty. That interview yielded new themes, as did her third interview with Tulip, about the conditions under which women immediately sought care, and so on. The process—coding, memoing (writing field notes about codes), constant comparison, testing hunches about how themes predict outcomes, and developing impressions about themes connect to one another (Boeije 2002)—continues until there is theoretical saturation (Fontanella et al. 2008, 2011), and new narratives yield no new categories or insights. It took Markovic (2006) 30 interviews (with 13 Australians and 17 immigrants) to finalize her model of the conditions under which women in Australia seek care for gynecological cancer. Once researchers have a model, they check it using negative case analysis (Becker 1961; Miles and Huberman 1994; Strauss and Corbin 1990) to ensure that there are no cases that disprove the model or disconfirm parts of it. Models can also be checked through consultation with ethnographic experts (e.g., participants in the study) or expert scholars (for an example, see Kearney et al. 1994). The overarching goal of grounded theory is to induce a testable theory, but scholars present their findings in different ways. Some concentrate on inducing categories, or a typology, from narrative data. For example, Boeri’s (2004) study of aging heroin users yielded nine ethnographically derived categories, such as “weekend warriors,” who only occasionally lose control while binging, and “junkies” whose lives revolve completely around drug use. Another approach is to build and present a model detailing the connections among a set of concepts. For instance, Eyre et al.’s (1998, 467) study of adolescent sexuality in a California high school yielded “a model of sex-related behavior as a set of interrelated games,” including courtship, duplicity, disclosure, and prestige. In addition to identifying concepts and connections among them, some grounded theorists also conduct comparative analyses that seek to identify or explain differences in models across groups. Markovic and colleagues, for example, found no salient differences between Australian and immigrant women’s models of the causes of gynecological cancer (Manderson et al. 2005). Based on their analysis, the researchers were able to suggest some conditions (e.g., recent immigrants or those of non-European origin) under which differences might emerge. This provided solid hypotheses for future researchers to test.

Classical Content Analysis While grounded theory is concerned with inducing hypotheses from data, content analysis typically comprises techniques for deductive theory-building. The object is to

14_107-Bernard.indb 542

6/9/14 10:03 AM

17: Text Analysis

543

test hypotheses about differences across the contexts from which the texts arise, across the people who produced them, or across changes in meaning over time or space. In addition to written texts, the raw data in content analysis can come from sources like clay pots, song lyrics, advertisements, political cartoons, or television shows. Netzley (2010), for example, examined the depiction of gay and straight characters in 98 episodes of primetime television programming in the United States. Two observers coded for traits like the character’s gender, sexual identity, and involvement in sexual situations. Netzley found that, contrary to her hypothesis, gay characters were more likely than straight ones to be depicted in sexual situations. The methodological issues associated with content analysis are all evident here. Does Netzley’s sample of 98 episodes justify generalizing to all primetime programming? Did the coders make valid and reliable judgments about when sexual situations were depicted (or not)? These two issues in particular, sampling and coding, are at the heart of content analysis. Sampling in content analysis involves two steps. The first step is the identification of the corpus of texts. If a small number of texts is collected (e.g., 40 life histories), then they all can be analyzed. When hundreds or thousands of texts are involved, a representative sample of records is required. For instance, Waitzkin and Britt (1993) randomly selected 50 transcripts from 336 audio recordings for an analysis of doctor– patient interactions about patients’ self-destructive behaviors. Sometimes purposive sampling is appropriate; it is particularly common in non-statistical content analyses. For example, Trost (1986) selected five cases to represent each of the 32 possible combinations of five binary variables in his study of teenagers’ relationships with their parents, yielding a dataset of 160 interviews. Patton (1990, 169–86) and Sandelowski (1995b) provide useful reviews of nonrandom strategies for sampling texts, and see Guest’s chapter, this volume. The second step in sampling is called “segmenting” (Tesch 1990) or “unitizing” (Krippendorff 2013) and is designed to identify the basic, nonoverlapping units of analysis within the texts. The units may be the entire texts (books, interviews, responses to an open-ended survey question) or segments (words, word-senses, sentences, themes, paragraphs). Where the object is to compare across texts—to see whether or not certain themes occur—the whole text (representing an informant or an organization) is the appropriate unit of analysis. When the object is to compare the number of times a theme occurs across a set of texts, then what Kortendick (1996) calls a “context unit”—a chunk of text that reflects a theme—is likely to be the appropriate unit of analysis. Coding is the heart and soul of sociological text analysis, whether it is classical content analysis, grounded theory, or schema analysis (Berelson 1952, 147–68; Holsti 1969, 95–126). In classical content analysis, codes are typically developed from definitions given in the literature. In exploratory content analysis, inductive coding (open coding in grounded theory terms) is appropriate (see Ryan and Bernard [2003] for a review of theme identification techniques). In either case, after an initial period of scrutinizing key resources (academic publications, the dataset itself, or both), researchers must make some tough decisions to define each code: What are the inclusion and exclusion criteria for this code? Is a particular instance of meaning “in” or “out”? What values can the code take (e.g., present/absent or high/medium/low) (Bernard 2011, 429–32;

14_107-Bernard.indb 543

6/9/14 10:03 AM

544

Amber Wutich et al.

Bernard and Ryan 2009, Chapter 4; Seidel and Kelle 1995)? This process involves a lot of forethought to determine what the research questions are asking exactly and how they will be tested at the end of the analysis. The codes are then assembled into a codebook. While the temptation to construct a huge, exhaustive codebook (that captures every possible code of interest) can be great, this is not a good idea. On a practical level, coders can only apply a limited number of codes without getting overwhelmed or forgetting important information. As a rule of thumb, limiting codebooks to 30–60 themes keeps them manageable (Bogdan and Biklen 1992, 166; Miles and Huberman 1994, 58). In addition to containing lists of codes and their definitions, the codebook contains important macro-level information about when and how the codes can be applied to text. For instance, the codebook should make clear how texts should be segmented for coding. It specifies which codes are allowed to co-occur, which must be exclusive, and which should be hierarchical (Richards and Richards 1991). Codebook development is covered in detail by Dey (1993, 95–151), Crabtree and Miller (1992), Miles and Huberman (1994, 55–72), and Bernard and Ryan (2009). Once the coding instructions—those contained in individual codes and in the overall codebook—are finalized, they can be tested. The standard for doing this is to have multiple coders apply the codes independently to set of texts and then measure the amount of intercoder agreement (Ryan 1999). Typical measures of intercoder agreement include Cohen’s kappa (Cohen 1960), Krippendorff’s alpha (Krippendorff 2004), and simple percent agreement (see Lombard et al. [2002] for a comparison of measures). By analyzing misunderstandings and disagreements, researchers can identify and correct omissions and fuzzy definitions. Through iterative work with codes, codebooks, and coders, these codebook definitions become sharper and more robust—and ready to be applied in a real analysis (Carey et al. 1996; Krippendorff 2004). In classical content analysis, researchers typically test a hypothesis derived from the literature. Margolis (1984), for example, tested the hypothesis that changes in mothers’ social roles in the United States were driven by economic, technological, and demographic changes from colonial times to the present. She tested the hypothesis by examining advice manuals and columns aimed at mothers over the last 200-plus years on deductively identified topics such as breastfeeding, giving affection, and schedule-setting. Gravlee and Sweet (2008) used classical content analysis in their longitudinal study of the concepts of race, racism, and ethnicity in medical anthropology journals, and Cruz (2002) examined the treatment of Latinos and Latin Americans in U.S. history textbooks. Because anthropologists are often concerned with inducing meaning from texts, they may use exploratory approaches to content analysis. Nichter and colleagues (2009) identified 12 key themes—such as “control of emotions” and “expressing critical political views”—in advertisements for clove cigarettes in Indonesia. They determined whether each theme occurred in advertisements targeting four predetermined age groups in the country. Anthropologists and other scholars who work with the Human Relations Area Files (HRAF) typically use a mixed deductive-inductive approach to content analysis. The Outline of Cultural Materials (OCM) provides a list of topics that are tagged (i.e.,

14_107-Bernard.indb 544

6/9/14 10:03 AM

17: Text Analysis

545

identified) in the HRAF texts (Ember and Ember 2009). From there, researchers may refine the tags inductively to develop new codes for hypothesis testing. For example, in his cross-cultural study of friendship, Hruschka (2010) began with codes, such as “gift-giving” and “tit-for-tat,” which he had identified deductively from the literature. After reading the HRAF texts tagged with the OCM codes for “friendship” and “artificial kinship,” he inductively developed additional codes, such as “ritual initiation” and “touching,” based on ethnographic depictions of friendship.

Analytic Induction Analytic induction is a formal method to build up causal explanations of phenomena from a small number of cases by applying some rules of logic formalized by John Stuart Mill (1898, 259). Analytic induction was proposed as an alternative to statistical analysis by Znaniecki (1934, 249–331), and Robinson (1951) laid out the steps in the method clearly: 1. Study a single case and develop a theory to account for it. 2. Test the theory on a second case. If the theory fits, test it on a third and a fourth, and so on until you find a case that doesn’t fit. 3. Modify the theory to accommodate the case that doesn’t fit or redefine the phenomenon you’re trying to explain so that it includes the rogue case. 4. Repeat until the theory explains all new cases.

Usually, between 18 and 50 cases are needed to establish a stable theory (Caren and Panofsky 2005). Of course, no matter how many cases you explain, you never know if some case will turn up that falsifies your theory. However, a theory built on, say, 20 cases that explains another 20 independent cases, would be strong in any field of scholarship. Classic examples of analytic induction include Lindesmith’s (1968 [1947]) study of drug addiction, Cressy’s study of embezzling (1953), and McCleary’s (1978) study of parole violation. More recently, ethnographers have used the method to examine issues ranging from dowry practices (Lang 1993), to risk management among pastoralists (Moritz et al. 2011), to the success of the Brazilian landless movement (Kröger 2011). Ragin (1987, 1994, 2000; Rihoux and Ragin 2009) formalized the logic of analytic induction. His method, called qualitative comparative analysis (QCA), allows both “crisp set” (i.e., binary) and “fuzzy set” (i.e., continuous) conditions. Other recent innovations include the inclusion of temporal reasoning in QCA (Caren and Panofsky 2005). Computer programs like FS/QCA (Ragin et al. 2006), TOSMANA (Cronqvist 2011), and ANTHROPAC’s truth tables function (Borgatti 1992) perform versions of analytic induction analysis. Schweizer (1991, 1996) used analytic induction in his analysis of social status in Chen Village, China. Starting with qualitative data—Chan et al.’s (1984) history of Chen Village—Schweizer coded to determine whether each of 13 actors experienced an increase or a decrease in status after each of 14 events (such as the Great Leap Forward, land reform and collectivization, and an event known locally as “the great betrothal dispute”). Over time, 9 people consistently won or lost, while 4 lost sometimes and won other times. This produced 17 unique combinations of actors and

14_107-Bernard.indb 545

6/9/14 10:03 AM

546

Amber Wutich et al.

outcomes. Schweizer coded the 17 cases according to three binary conditions (urban vs. rural origin, proletarian vs. nonproletarian background, presence vs. absence of external ties) and one dependent variable (whether the actor was a success or failure overall). In 88% of cases, successful people either came from a city OR had a proletarian background AND had good external ties. If an actor had failed in the Chen Village disputes, then, in all cases, he or she came from a village AND came from a nonproletarian family OR had no ties to authorities beyond the village. In short, in a communist revolutionary environment, it pays to have friends in high places; people from urban areas are more likely to have those ties; and it helps to have been born into a politically correct (i.e., proletarian) family.

Computer-Assisted Text Analysis One problem with text as data, as all fieldworkers know, is that it piles up quickly. The sheer volume and problems of handling and sorting through so much information made text analysis less popular in the past than it is now. What has changed is the development of software for managing, coding, and analyzing text. Programs for text analysis tend to come from two traditions: Programs like Atlas. ti and Nvivo came from the grounded theory tradition, while programs like WordStat and Crawdad came from the data-mining tradition. Programs like MAXQDA and Dedoose sought to bridge these two traditions. Today, text processing programs are converging in the tasks they perform. The CAQDAS (Computer Assisted Qualitative Data Analysis Software) Network Project (http://www.surrey.ac.uk/sociology/research/ researchcentres/caqdas) and Lewins and Silver’s guides to using software (2007) and choosing software (2009) provide information on the features of popular programs. Some scholars prefer to conduct their analyses on paper, and for many methods (e.g., schema analysis, grounded theory) there are few differences between rigorous paper-based analyses and those conducted using software. Software, however, facilitates clear record-keeping. This allows us to document the micro-level decisions we make during the course of a study—how we built codebooks, why we decided to collapse two concepts together or split them apart, how our understanding of a model emerged over months of analysis. This, in turn, lets our colleagues restudy our data or reproduce our work—activities that are crucial to building cross-cultural theories in anthropology. These programs also make it possible to do some analyses that would be otherwise impossible (or, at least, very unpleasant). To illustrate the possibilities, we briefly discuss three methods below. WORD-BASED ANALYSIS: COUNTING AND KWIC STUDIES

Counting the words people use can reveal much about how they think and feel, but this deceptively simple technique is daunting to do by hand. Ryan and Weisner (1996) told fathers and mothers of adolescents: “Describe your children. In your own words, just tell us about them.” Mothers were more likely than fathers to use the words “friends” and “creative,” while fathers were more likely than mothers to use the words “school” and “student.” These word counts became clues to the themes that were used in coding the texts. Pennebaker et al. (1997) examined narratives collected longitudinally from HIV-negative men whose romantic partners had recently died of AIDS

14_107-Bernard.indb 546

6/9/14 10:03 AM

17: Text Analysis

547

complications. They found that men who initially used more words associated with insight (e.g., “realize”) and causation (e.g., “effect”)—and fewer of these words as time went on—tended to have a more “positive state of mind” (i.e., more pleasure, productivity, etc.) one year after their partner’s death. The KWIC (key word in context) technique is another simple but revealing technique: Every use of a substantive word in a text or set of texts is displayed in a context of some specified length—like the sentence or paragraph in which each word occurs, or ten words on either side of each key word, and so on. This technique was used for centuries to produce concordances of literary works, like the Bible (Merbecke 1550) or the works of Shakespeare (Clarke 1881). As with word counts, software now makes light work of the KWIC technique. C. Strauss (2005) argued that a KWIC-like analysis could help identify shared but unspoken cultural assumptions, such as the way the word “work” is used in the context of “work ethic” or “hard work” in discussions of welfare in the United States. Wegerif et al. (1999) used the KWIC technique in their study of why students in one Mexican preschool developed more problem-solving skills than students in another, similar preschool. Teachers in the problem-solving preschool used phrases like vamos a ver (let’s look) more often and to convey “the shared nature of exploration and discovery of knowledge”; in the other class, teachers used vamos a ver infrequently and in ways designed “to control (wait until I tell you what to do) and direct (look at the blackboard)” students’ thinking (Wegerif et al. 1999, 140). CONTENT DICTIONARIES

Content dictionaries combine word-based analysis and classical content analysis. Using software, a researcher builds a code definition that contains an unlimited number of words, word stems, or word combinations. Even simple words, like “cold” or “horse” or “think” require many definitions and contexts, but once a computer-based dictionary is built, the software can automatically code a set of texts—like transcribed interviews or downloaded web sites. In pioneering work that demonstrated the potential for computer-based coding of text, Stone and colleagues (Ogilvie et al. 1966; Stone et al. 1966) showed that a computer program was better than human coders at differentiating real suicide notes from fakes designed to imitate them. In anthropology, Colby (1966) showed that the Navajo regarded their homes as havens and places of relaxation, whereas the Zuni depicted their homes as places of discord and tension. Much of the work with content dictionaries, however, is in psychology (e.g., Gottschalk and Bechtel 2002; Rosenberg et al. 1990) and criminology (Dixon et al. 2008; Maruna 2010). Recent studies by Gottschalk and colleagues have demonstrated that software can be trained to employ DSM-IV categories to make neuropsychiatric evaluations from writing samples. In a study of Napoleon Bonaparte’s adolescent writings, for example, the computer program detected that “Napoleon felt depressed, isolated, self-critical, and alienated” (Gottschalk et al. 2002, 546). SEMANTIC NETWORK ANALYSIS

Semantic network analysis treats “words as actors” (Schnegg and Bernard 1996) and involves measuring and plotting relationships among words across one or more texts. As early as 1959, Charles Osgood (1959) created word co-occurrence matrices

14_107-Bernard.indb 547

6/9/14 10:03 AM

548

Amber Wutich et al.

and applied factor analysis and dimensional plotting to describe the relation of major factors to one another. Semantic network analysis is also used to examine the relationship among codes applied to texts. García-Álvarez and López-Sintas (2002) analyzed transcripts of lengthy interviews conducted with retiring founders of family businesses in Spain. They coded the transcripts for values like “hard work,” “honesty,” and “risk-taking” and analyzed the co-occurrence matrix of codes with multi-dimensional scaling (MDS). The researchers interpreted the two dimensions of the MDS graph as self-fulfillment (e.g., determination) versus group orientation (e.g., constancy) and business-as-an-end (e.g., growth) versus business-as-a-means (e.g., family orientation). The development of software has made the construction and analysis of co-occurrence matrices much easier and has stimulated the development of this field (Barnett and Danowski 1992; Danowski 1982, 1993).

Teamwork, Collaboration, and Interdisciplinary Research Any project that involves coding also requires the development of codes and a codebook—and this is an exercise that is best done through teamwork. At a minimum, a code-based text analysis might require a few days of work with a colleague who is willing to help perform intercoder agreement checks for a small subsample of texts. A large text analysis project might require a team of assistants—interviewers, transcriptionists, translators, and coders. To design and manage such a project requires skills beyond the basics of text analysis, including hiring, training, and data quality checks. Guest and MacQueen’s (2008) Handbook for Team-based Qualitative Research lays out the challenges of teamwork in text analysis and provides suggestions to how those challenges can be met. Researchers who develop teamwork skills have new opportunities for collaboration and interdisciplinary work. Since Boas’s work with George Hunt on the Kwakiutl texts, there has been a tradition of collaboration in anthropological text analysis. Much early collaboration involved native and nonnative ethnographers working side by side to produce transcriptions, translations, and interpretations of texts recorded in written form for the first time. This tradition has continued, particularly in the realm of linguistic anthropology, as in Bernard and Salinas’s (1976) work on Ñähñu parables and humor. In addition, anthropologists share a rich history of inter-subdisciplinary collaboration. For example, Jane Hill (1992) proposed the existence of a Uto-Aztecan Flower World complex, in which flowers and related images were used to symbolize the Spirit Land, vital life forces, and masculine strength in Uto-Aztecan song and oratory. In collaboration with archaeologist Kelley Hays-Gilpin, Hill established the presence of Flower World imagery in a wide range of Mesoamerican and southwestern contexts, including ceramics, ritual regalia, and murals (Hays-Gilpin and Hill 1999, 2000). A recent collaborative trend involves interdisciplinary research by anthropologists working with colleagues from many disciplines, including medicine, ecology, and law. On large interdisciplinary teams, anthropologists often take the lead on ethnographic research, including participant observation, interviewing, and text analysis. For example, anthropologists in the Long-term Ecological Research Network have led text analyses on Arizona water decision-makers’ understandings of climate-related risk (Wutich et al. 2010) and on locals’ role in constructing scientific knowledge about

14_107-Bernard.indb 548

6/9/14 10:03 AM

17: Text Analysis

549

the Florida Everglades (Ogden 2008). These collaborations help us expand both the breadth of the questions we ask as anthropologists and the potential audience and impact of our findings. Yet, as Sillitoe (2007) argues, interdisciplinary research can be challenging because it requires that we explain our methods to colleagues who may not share our assumptions—it forces us to be explicit about how we know what we know. Skills in systematic methods of collecting and analyzing qualitative data are essential to interdisciplinary teamwork for anthropologists. SOME GENERAL OBSERVATIONS

One of our goals for this chapter has been to show the common foundations and complementary contributions of different methods for text analysis. Scholars in the linguistic anthropology tradition rarely cite the work of colleagues in the sociological tradition, and vice versa. Even within the sociological tradition, the literatures on schema analysis, grounded theory, and content analysis are essentially distinct, though they all involve identifying and refining concepts, marking (coding, tagging) concepts in text, and linking concepts into theoretical models. Anthropologists have much to gain by reading, working, and innovating across these methodological boundaries. There is increasing interest among anthropologists, as there is across the social sciences, in mixing methods for text analysis (Creswell and Clark 2007; Denzin 2010; Teddlie and Tashakkori 2003). From the time of Boas and Kroeber, there has been a strong mixed methods tradition in anthropology, and today, anthropologists may combine open coding (a qualitative technique) with meta-coding or semantic network analysis (quantitative techniques) to identify key themes early in an analysis (Ryan and Bernard 2003). Recent examples of the mixed methods approach include Hruschka’s (2010) cross-cultural study of friendship, Gravlee and Sweet’s (2008) research on the concept of race in medical anthropology, and Wutich’s (2009) research on gender and emotional distress in water-insecure Cochabamba, Bolivia. CONCLUSION

Text analysis as a research strategy permeates the social sciences, and the range of methods for conducting text analysis is remarkable. Investigators examine words, sentences, paragraphs, pages, documents, ideas, meanings, paralinguistic features, and even what is missing from the text. They interpret, mark, retrieve, and count. By turns, they apply interpretive analysis and numerical analysis; they use text analysis to explore for themes and to confirm hypotheses. Improvements in methods for capturing and mining qualitative data are making these tasks easier. Voice recognition software, for example, is lessening the burden of transcription. Such developments will make it easier for scholars to explore for and explain patterns in texts and will, we believe, support methodological innovation in a new generation of anthropologists. REFERENCES

Agar, M. 1979. Themes revisited: Some problems in cognitive anthropology. Discourse Processes 2: 11–31. Agar, M. 1980. Getting better quality stuff: Methodological competition in an interdisciplinary niche. Urban Life 9: 34–50. Agar, M. 1983. Microcomputers as field tools. Computers and the Humanities 17: 19–26.

14_107-Bernard.indb 549

6/9/14 10:03 AM

550

Amber Wutich et al.

Anderson-Fye, E. 2003. Never leave yourself: Ethnopsychology as mediator of psychological globalization among Belizean schoolgirls. Ethos 31: 59–94. Andrzejewski, B. W. 2011. The rise of written Somali literature. Journal of African Cultural Studies 23: 73–80. Archive of the Indigenous Languages of Latin America (AILLA). University of Texas Libraries. Austin, Texas. http://www.ailla.utexas.org (accessed February 12, 2013). Atran, S., and D. Medin. 2008. The native mind and the cultural construction of nature. Cambridge, MA: MIT Press. Bang, M., D. L. Medin, and S. Atran. 2007. Cultural mosaics and mental models of nature. Proceedings of the National Academy of Sciences 104: 13868–74. Barnett, G., and J. Danowski. 1992. The structure of communication: A network analysis of the International Communication Association. Human Communication Resources 19: 164–285. Bartlett, F. C. 1967 [1932]. Remembering: A study in experimental and social psychology. Cambridge: Cambridge University Press. Beauboeuf-Lafontant, T. 2008. Listening past the lies that make us sick: A voice-centered analysis of strength and depression among black women. Qualitative Sociology 31: 391–406. Beck, C. T. 1993. Teetering on the edge: A substantive theory of postpartum depression. Nursing Research 42: 42–48. Becker, H., B. Geer, E. C. Hughes, and A. L. Strauss. 1961. Boys in white: Student culture in medical school. Chicago: University of Chicago Press. Bender, A., E. Hutchins, and D. L. Medin. 2010. Anthropology in cognitive science. Topics in Cognitive Science 2: 374–85. Berelson, B. 1952. Content analysis in communication research. Glencoe, IL: Free Press. Bernard, H. R . 1997. Language preservation and publishing. In Indigenous literacy in the Americas, ed. N. H. Hornberger, 139–56. Berlin: Mouton de Gruyter. Bernard, H. R. 2011. Research methods in anthropology: Qualitative and quantitative approaches, 5th ed. Walnut Creek, CA: AltaMira. Bernard, H. R., and G. Ryan. 2009. Analyzing qualitative data: Systematic approaches. Thousand Oaks, CA: Sage. Bernard, H. R., and J. Salinas Pedraza. 1976. Otomí parables, folktales, and jokes. International Journal of American Linguistics, Native American Texts Series 1(2). Chicago: University of Chicago Press. Bernard, H. R., and J. Salinas Pedraza. 1989. Native ethnography: A Mexican Indian describes his culture. Thousand Oaks, CA: Sage. Besnier, N. 1995. Literacy, emotion, and authority: Reading and writing on a Polynesian atoll. Cambridge: Cambridge University Press. Bizarro Ujpán, I., and J. Sexton. 1981. Son of Tecun Uman: A Maya Indian tells his life story. Tucson: University of Arizona Press. Bizarro Ujpán, I., and J. Sexton. 1985. Campesino: The diary of a Guatemalan Indian. Tucson: University of Arizona Press. Bizarro Ujpán, I., and J. Sexton. 2001. Joseño: Another Mayan voice speaks from Guatemala. Albuquerque: University of New Mexico Press. Blase, J. J. 1982. A social-psychological grounded theory of teacher stress and burnout. Educational Administration Quarterly 18: 93–113. Bletzer, K., and M. P.Koss. 2006. After-rape among three populations in the Southwest. Violence against Women 12: 5–29. Blowsnake, S., and P. Radin, ed. and tr. 1983. Crashing Thunder: Autobiography of a Winnebago Indian. Lincoln: University of Nebraska Press. Orig. pub. in 1920 by the University of California Press and again in 1926 by Appleton.

14_107-Bernard.indb 550

6/9/14 10:03 AM

17: Text Analysis

551

Boas, F. 1910. Kwakiutl tales. New York: Columbia University Press. Boas, F . 1917. Folk-tales of Salishan and Sahaptin tribes. Lancaster, PA: American Folk-Lore Society. Boas, F. 1928a. Kerasan texts. New York: The American Ethnological Society. Boas, F. 1928b. Bella Bella texts. New York: Columbia University Press. Boas, F. 1935–43. Kwakiutl tales, New Series. New York: Columbia University Press. Boas, F., and G. Hunt. 1902–1905. Kwakiutl texts. Leiden, Netherlands: E. J. Brill. Boas, F., and G. Hunt. 1906. Kwakiutl texts, Second Series. Memoirs of the American Museum of Natural History, 14. New York: G. E. Stechert. Boeije, H. 2002. A purposeful approach to the constant comparative method in the analysis of qualitative interviews. Quality & Quantity 36: 391–409. Boeri, M. W. 2004. “Hell, I’m an addict, but I ain’t no junkie”: An ethnographic analysis of aging heroin users. Human Organization 63: 236–45. Bogdan, R., and S. K. Biklen. 1992. Qualitative research for education: An introduction to theory and methods, 2nd ed. Boston: Allyn and Bacon. Borgatti, S. 1992. ANTHROPAC 4.0. Columbia, SC: Analytic Technologies. www.analytictech .com (accessed February 12, 2013). Boster, J. S. 2011. Data, method, and interpretation in cognitive anthropology. In The Blackwell companion to cognitive anthropology, ed. D. Kronenfeld, G. Bennardo, V. C. de Munck, and M. Fischer, 131–52. Cambridge/Oxford: Blackwell. Boychuk Duchscher, J., and D. Morgan. 2004. Grounded theory: Reflections on the emergence vs. forcing debate. Journal of Advanced Nursing 48: 605–12. Brown, L. M., and C. Gilligan. 1993. Meeting at the crossroads: Women’s psychology and girls’ development. Feminism & Psychology 3: 11–35. Caren, N., and A. Panofsky. 2005. TQCA: A technique for adding temporality to qualitative comparative. Sociological Methods & Research 34: 147–72. Carey, J. W., M. Morgan, and M. J. Oxtoby. 1996. Intercoder agreement in analysis of responses to open-ended interview questions: Examples from tuberculosis research. Cultural Anthropology Methods Journal 8: 1–5. Carley, K., and M. Palmquist. 1992. Extracting, representing, and analyzing mental models. Social Forces 70: 601–36. Casson, R. 1983. Schemata in cultural anthropology. Annual Review of Anthropology 12: 429–62. Chafe, W. 1980. The deployment of consciousness in the production of narrative. In The pear stories. Cognitive, cultural, and linguistic aspects of narrative production, ed. W. Chafe, 9–50. Norwood, NJ: Ablex. Chan, A., R. Madsen, and J. Unger. 1984. Chen Village. The recent history of a peasant community in Mao’s China. Berkeley: University of California Press. Chappell, H. 1988. Aspectually-bounded structures in the Chinese pear stories: The two le’s in Chinese: A case of homonymy, polysemy or neither? La Trobe working papers in linguistics, Vol. 1. http://arrow.latrobe.edu.au:8080/vital/access/HandleResolver/1959.9/146190 (accessed January 21, 2012). Charmaz, K. 1990. “Discovering” chronic illness: Using grounded theory. Social Science and Medicine 30: 1161–72. Charmaz, K. 2006. Constructing grounded theory: A practical guide through qualitative analysis (Introducing Qualitative Methods series). Thousand Oaks, CA: Sage. Chomsky, N. 1957. Syntactic structures. The Hague: Mouton. Chui, K., and H.-L. Lai. 2008. The NCUU corpus of spoken Chinese: Mandarin, Hakka, and southern Min. Taiwan Journal of Linguistics 6: 119–44.

14_107-Bernard.indb 551

6/9/14 10:03 AM

552

Amber Wutich et al.

Clarke, M. C. 1881. The complete concordance to Shakespeare: Being a verbal index to all the passages in the dramatic works of the poet. London: Bickers and Son. Cohen, J. 1960. A coefficient of agreement for nominal scales. Educational and Psychological Measurement 20: 37–48. Colby, B. N. 1966. The analysis of culture content and the patterning of narrative concern in texts. American Anthropologist 68: 374–88. Corbin, J., and A. Strauss. 2007. Basics of qualitative research: Techniques and procedures for developing grounded theory, 3rd ed. Thousand Oaks, CA: Sage. Coyne, I. 1997. Sampling in qualitative research. Purposeful and theoretical sampling: Merging or clear boundaries? Journal of Advanced Nursing 26: 623–30. Crabtree, B. F., and W. L. Miller. 1992. A template approach to text analysis: Developing and using codebooks. In Doing qualitative research, ed. W. L. Miller and B. F. Crabtree, 93–109. Thousand Oaks, CA: Sage. Cressy, D. R. 1973 [1953]. Other people’s money: A study in the social psychology of embezzlement. Montclair, NJ: Patterson Smith. Creswell, J. W., and V. L. P. Clark. 2007. Designing and conducting mixed methods research. Thousand Oaks, CA: Sage. Cronqvist, L. 2011. Tosmana—Tool for small-N analysis [Version 1.3.2]. Trier, Germany. http://www.tosmana.net (accessed February 12, 2013). Cruz, B. C. 2002. Don Juan and rebels under palm trees: Depictions of Latin Americans in U.S. history textbooks. Critique of Anthropology 22: 323–42. D’Andrade, R. 1984. Cultural meaning systems. In Culture theory: Essays on mind, self, and emotion, ed. R. A. Shweder and R. A. Levine, 88–119. Cambridge: Cambridge University Press. D’Andrade, R. 1987. A folk model of the mind. In Cultural models in language and thought, ed. N. Quinn and D. Holland, 112–48. Cambridge: Cambridge University Press. D’Andrade, R. 1991. The identification of schemas in naturalistic data. In Person schemas and maladaptive interpersonal patterns, ed. M. J. Horowitz, 279–301. Chicago: University of Chicago Press. D’Andrade, R. 1995. The development of cognitive anthropology. Cambridge: Cambridge University Press. Danowski, J. 1982. A network-based content analysis methodology for computer-mediated communication: An illustration with a computer bulletin board. In Communication yearbook, ed. R. Bostrom, 904–25. New Brunswick, NJ: Transaction Books. Danowski, J. 1993. Network analysis of message content. In Progress in communication science, XII, ed. W. D. Richards and G. A. Barnett, 197–221. Norwood, NJ: Ablex Publishing. Denzin, N. 2010. Moments, mixed methods, and paradigm dialogs. Qualitative Inquiry 16: 419–27. Dey, I. 1993. Qualitative data analysis: A user-friendly guide for social scientists. London: Routledge and Kegan Paul. Dixon, L., C. Hamilton-Giachritsis, and K. Browne. 2008. Classifying partner femicide. Journal of Interpersonal Violence 23: 74–93. Drauker, C., D. S. Martsolf, R. Ross, and T. B. Rusk. 2007. Theoretical sampling and category development in grounded theory. Qualitative Health Research 17: 1137–48. Edmonson, M. S. 1982. The ancient future of the Itza: The book of Chilam Balam of Tizimin. Austin: University of Texas Press. Edmonson, M. S., tr. and ed. 1986. Heaven born Mérida and its destiny: The book of Chilam Balam of Chumayel. Austin: University of Texas Press. Eisenlohr, P. 2004. Language revitalization and new technologies: Cultures of electronic mediation and the refiguring of communities. Annual Review Anthropology 33: 21–45.

14_107-Bernard.indb 552

6/9/14 10:03 AM

17: Text Analysis

553

El Guindi, F., and A. Hernández Jiménez. 1986. The myth of ritual. A native’s ethnography of Zapotec life-crisis rituals. Tucson: University of Arizona Press. Ember, C. R., and M. Ember. 2009. Cross-cultural research methods. Walnut Creek, CA: AltaMira. Erbaugh, M. S. 1990. Mandarin oral narratives compared with English: The pear/guava stories. Journal of the Chinese Language Teachers Association 25: 21–42. Eyre, S. L., V. Hoffman, S. G. Millstein. 1998. The gamesmanship of sex: A model based on African American adolescent accounts. Medical Anthropology Quarterly 12: 467–89. Fontanella, B. J. B., B. M. Luchesi, M. G. B. Saidel, J. Ricas, E. R. Turato, and D. G. Melo. 2011. Sampling in qualitative research: A proposal for procedures to detect theoretical saturation. Cadernos de Saude Publica 27: 388–94. Fontanella, B. J. B., J. Ricas, E. R. Turato. 2008. Saturation sampling in qualitative health research: Theoretical contributions. Cadernos de Saude Publica 24: 17–27. Furbee, L. 1996. The religion of politics in Chiapas: Founding a cult of communicating saints. Presented at the 96th meeting of the American Anthropological Association, San Francisco, November 20–24. Furbee, L. 2000. Prestige, power, and potential for language shift: The intrusion of Spanish on Tojolab’al Maya, ed. D. Gilbers, J. A. Nerbonne, and J. Schaeken, 99–103. Studies in Slavic and General Linguistics Series, Vol. 28. Amsterdam-Atlanta: Rodopi. Furbee-Losee, L., ed. 1976. Mayan texts I. International Journal of American Linguistics, Native American Texts Series, 1(1). Chicago: University of Chicago Press. García-Álvarez, E., and J. López-Sintas. 2002. Contingency table: A two-way bridge between qualitative and quantitative methods. Field Methods 14: 270–87. Garro, L. C. 1988. Explaining high blood pressure: Variation in knowledge about illness. American Ethnologist 15: 98–119. Garro, L. C. 1995. Intracultural variation in causal accounts of diabetes: A comparison of three Canadian Anishinaabe (Ojibway) communities. Culture, Medicine and Psychiatry 20: 381–420. Garro, L. C. 2000. Remembering what one knows and the construction of the past: A comparison of cultural consensus theory and cultural schema theory. Ethos 3: 275–319. Glaser, B. G. 1992. Emergence vs. forcing: Basics of grounded theory analysis. Mill Valley, CA: Sociology Press. Glaser, B. G. 1999. The future of grounded theory. Qualitative Health Research 9: 836–45. Glaser, B. G., and A. Strauss. 1967. The discovery of grounded theory: Strategies for qualitative research. New York: Aldine. Goddard, P. E. 1911. Jicarilla Apache texts. Anthropological papers of the American Museum of Natural History 8. New York: The Trustees. González Ventura, J. 1997. Experiences in the development of a writing system for Ñuu Savi. In Indigenous literacy in the Americas, ed. N. H. Hornberger, 157–71. Berlin: Mouton de Gruyter. Gottschalk, L., and R. Bechtel. 2002. Computerized content analysis of writings of Mahatma Gandhi. Journal of Nervous and Mental Disease 193: 210–16. Gottschalk, L., D. DeFrancisco, and R. Bechtel. 2002. Computerized content analysis of some adolescent writings of Napoleon Bonaparte: A test of the validity of the method. Journal of Nervous and Mental Disease 190: 542–48. Gravlee, C., and E. Sweet. 2008. Race, ethnicity, and racism in medical anthropology, 1977– 2002. Medical Anthropology Quarterly 22: 27–51. Gruber, J., and A. Kring. 2008. Narrating emotional events in schizophrenia. Journal of Abnormal Psychology 117: 520–33.

14_107-Bernard.indb 553

6/9/14 10:03 AM

554

Amber Wutich et al.

Guest, G., and K. MacQueen, eds. 2008. Handbook for team-based qualitative research. Lanham, MD: AltaMira. Habermas, T., S. Ehlert-Lerche, and C. de Silveira Cybele. 2009. The development of the temporal macrostructure of life narratives across adolescence: Beginnings, linear narrative form, and endings. Journal of Personality 77: 527–60. Handwerker, W. P., and S. H. Wolfe. 2010. Where bad teeth come from: Culture and causal force. Human Organization 69: 398–406. Hays-Gilpin, K. A., and J. H. Hill. 1999. The flower world in material culture: An iconographic complex in the Southwest and Mesoamerica. Journal of Anthropological Research 55: 1–37 Hays-Gilpin, K. A., and J. H. Hill. 2000. The flower world in prehistoric Southwest material culture. In The archeology of regional interaction: Religion, warfare, and exchange across the American Southwest and beyond, ed. M. Hegmon, 411–28. Boulder: University of Colorado Press. Herzfeld, M. 1977. Ritual and textual structures: The advent of spring in rural Greece. In Text and context, ed. R. K. Jain, 29–45. Philadelphia: Institute for the Study of Human Issues. Hill, J. H. 1992. The flower world of Old Uto-Aztecan. Journal of Anthropological Research 48: 117–44. Hill, J. H. 2005. Finding culture in narrative. In Finding culture in talk: A collection of methods, ed. N. Quinn, 157–202. New York: Palgrave Macmillan. Holland, D. 1985. From situation to impression: How Americans get to know themselves and one another. In Directions in cognitive anthropology, ed. J. W. D. Dougherty, 389–412. Urbana: University of Illinois Press. Holsti, O. R. 1969. Content analysis for the social sciences and humanities. Reading, MA: Addison-Wesley. Hruschka, D. 2010. Friendship: Development, ecology, and evolution of a relationship. Berkeley: University of California Press. Hymes, D. 1976. Louis Simpson’s “The Deserted Boy.” Poetics 5: 119–55. Hymes, D. 1977. Discovering oral performance and measured verse in American Indian narrative. New Literary History 8: 431–57. Hymes, D. 1980a. Verse analysis of a Wasco text: Hiram Smith’s “At’unaqa.” International Journal of American Linguistics 46: 65–77. Hymes, D. 1980b. Particle, pause, and pattern in American Indian narrative verse. American Indian Culture and Research Journal 4: 7–51. Hymes, D. 1981. In vain I tried to tell you: Essays in Native American ethnopoetics. Philadelphia: University of Pennsylvania Press. Hymes, V. 1987. Warm Springs Sahaptin narrative analysis. In Native American discourse: Poetics and rhetoric, ed. J. Sherzer and A. Woodbury, 62–102. Cambridge: Cambridge University Press. Izquierdo, C., and A. Johnson. 2007. Desire, envy and punishment: A Matsigenka emotion schema in illness narratives and folk stories. Culture, Medicine and Psychiatry 31: 419–44. Jennings, J. (Strong Woman), and F. J. O’Brien (Moondancer). 1998. Bringing back our lost language. American Indian Culture and Research Journal 22: 215–22. Johnson-Laird, P. N. 1983. Mental models: Towards a cognitive science of language, inference, and consciousness. Cambridge, MA: Harvard University Press. Johnson-Laird, P. N. 2010. Mental models and human reasoning. Proceedings of the National Academy of Science 107: 18243–50. Jones, N. A., H. Ross, T. Lynam, P. Perez, and A. Leitch. 2011. Mental models: An interdisciplinary synthesis of theory and methods. Ecology and Society 16: 46. http://www.ecologyandsociety.org/ vol16/iss1/art46/ (accessed February 12, 2013). Kearney, M. H., S. Murphy, and M. Rosenbaum. 1994. Mothering on crack cocaine— A grounded theory analysis. Social Science and Medicine 38: 351–61.

14_107-Bernard.indb 554

6/9/14 10:03 AM

17: Text Analysis

555

Kempton,W., J. Boster, and J. Hartley. 1995. Environmental values in American culture. Cambridge, MA: MIT Press. Kortendick, O. 1996. Drei Schwestern un ihre Kinder. Rekonstruktion von Familiengeschichte und Identitätstransmission bei Indischen Nederlanders mit Hilfe computerunterstützer Inhaltsanlyze. Canterbury, UK: University of Kent, Center for Anthropology and Computing. Krippendorff, K. 2013. Content analysis: An introduction to its methodology, 3rd ed. Beverly Hills, CA: Sage. Krippendorff, K. 2004. Reliability in content analysis: Some common misconceptions and recommendations. Human Communication Research 30: 411–33. Kroeber, A. 1907. Gros Ventre myths and tales. Anthropological papers of the American Museum of Natural History 1 (Part 3). New York: The Trustees. Kröger, M. 2011. Promotion of contentious agency as a rewarding movement strategy: Evidence from the MST-paper industry conflicts in Brazil. The Journal of Peasant Studies 38: 435–58. Lakoff, G., and M. Johnson. 2003 [1980]. Metaphors we live by. Chicago: University of Chicago Press. Lang, H. 1993. Dowry and female competition: A Boolean reanalysis. Current Anthropology 34: 775–78. Lewins, A., and C. Silver. 2007. Using software in qualitative research: A step-by-step guide. Los Angeles: Sage. Lewins, A., and C. Silver. 2009. Choosing a CAQDAS package. 6th ed. http://www.surrey.ac.uk/ sociology/research/researchcentres/caqdas/files/2009ChoosingaCAQDASPackage.pdf (accessed October 12, 2011). Lewis, M. P., ed. 2009. Ethnologue: Languages of the world, 16th ed. Dallas: SIL International. http://www.ethnologue.com/ (accessed January 12, 2012). Bottom of Form. Lincoln, Y. S., and E. G. Guba. 1985. Naturalistic inquiry. Beverly Hills, CA: Sage. Lindesmith, A. R. 1968 [1947]. Addiction and opiates. Chicago: Aldine. Lofland, J., and L. H. Lofland. 1995. Analyzing social settings, 3rd ed. Belmont, CA: Wadsworth. Lombard, M., J. Snyder-Duch, and C. Campanella Bracken. 2002. Content analysis in mass communication research: An assessment and reporting of intercoder reliability. Human Communication Research 28: 587–604. Lowie, R. H. 1930. A Crow text, with grammatical notes. Berkeley: University of California Press. Lurie, N. 1961. Mountain Lone Wolf woman, sister of Crashing Thunder: The autobiography of a Winnebago woman. Ann Arbor: University of Michigan Press. Manderson, L., M. Markovic, and M. Quinn. 2005. “Like roulette”: Australian women’s explanations of gynecological cancers. Social Science & Medicine 61: 323–32. Margolis, M. 1984. Mothers and such: Views of American women and why they changed. Berkeley: University of California Press. Markovic, M. 2006. Analyzing qualitative data: Health care experiences of women with gynecological cancer. Field Methods 18: 413–29. Maroney, E. M. 2004. Aspect in ASL: A typological-functional analysis. Proceedings of the Annual Meeting of the Berkeley Linguistics Society, Vol. 30, No. 1. Maruna, S. 2010. Mixed method research in criminology: Why not go both ways? In Handbook of quantitative criminology, ed. A. R. Piquero and D. Weisburd, 123–40. London: Springer. Mathews, H. F. 1992. The directive force of morality tales in a Mexican community. In Human motives and cultural models, ed. R. D’Andrade and C. Strauss, 127–62. New York: Cambridge University Press. Mathews, H. F. 2005. Uncovering cultural models of gender from accounts in folktales. In Finding culture in talk: A collection of methods, ed. N. Quinn, 105–56. New York: Palgrave Macmillan.

14_107-Bernard.indb 555

6/9/14 10:03 AM

556

Amber Wutich et al.

Maznevski, M. L., and K. M. Chudoba. 2000. Bridging space over time: Global virtual team dynamics and effectiveness. Organization Science 11: 473–92. McCleary, R. 1978. Dangerous men: The sociology of parole. Beverly Hills, CA: Sage. McGregor, W. B. 2006. Focal and optional ergative marking in Warrwa (Kimberley, Western Australia). Lingua 116: 393–423. Merbecke, J. 1550. A concorda[n]ce: That is to saie, a worke wherein by the ordre of the letters of the A.B.C. ye maie redely finde any worde conteigned in the whole Bible, so often as it is there expressed or mencioned. London: Richardus Grafton. Miles, M., and A. M. Huberman. 1994. Qualitative data analysis, 2nd ed. Thousand Oaks, CA: Sage. Mill, J. S. 1898. A system of logic, ratiocinative and inductive: Being a connected view of the principles of evidence and the methods of scientific investigation. People’s edition. London: Longmans, Green, and Co. Montgomery, P., and P. H. Bailey. 2007. Field notes and theoretical memos in grounded theory. Western Journal of Nursing Research 29: 65–79. Moritz, M., J. Giblin, M. Ciccone, A. Davis, J. Fuhrman, M. Kimiaie et al. 2011. Social riskmanagement strategies in pastoral systems: A qualitative comparative analysis. CrossCultural Research 45: 286–317. Netzley, S. B. 2010. Visibility that demystifies: Gays, gender, and sex on television. Journal of Homosexuality 57: 968–86. Nichter, M. S. Padmawati, M. Danardono, N. Ng, Y. Prabandari, and Mark Nichter. 2009. Reading culture from tobacco advertisements in Indonesia. Tobacco Control 18: 98–107. Ochs, E., and L. Capps. 2001. Living narrative: Creating everyday lives in storytelling. Cambridge, MA: Harvard University Press. Ogden, L. A. 2008. Searching for paradise in the Florida Everglades. Cultural Geographies 15: 207–29. Ogilvie, D. M., P. J. Stone, and E. S. Schneidman. 1966. Some characteristics of genuine versus simulated suicide notes. In The general inquirer: A computer approach to content analysis, ed. P. J. Stone, D. C. Dunphy, M. S. Smith, and D. M. Ogilvie, 527–35. Cambridge: MIT Press. Orero, P. 2008. Three different receptions of the same film. European Journal of English Studies 12: 179–93. Osgood, C. 1959. The representational model and relevant research methods. In Trends in content analysis, ed. I. de Sola Pool, 33–88. Urbana: University of Illinois Press. Patton, M. Q. 1990. Qualitative evaluation and research methods. Thousand Oaks, CA: Sage. Pennebaker, J. W., T. J. Mayne, and M. E. Francis. 1997. Linguistic predictors of adaptive bereavement. Journal of Personality and Social Psychology 72: 863–71. Piaget, J. 1970. Piaget’s theory. In Carmichael’s manual of child psychology, 3rd ed., Vol. 1, ed. P. Mussen, 703–32. New York: Wiley. Postel-Coster, E. 1977. The Indonesian novel as a source of anthropological data. In Text and context, ed. R. K. Jain, 135–50. Philadelphia: Institute for the Study of Human Issues. Price, L. 1987. Ecuadorian illness stories. In Cultural models in language and thought, ed. D. Holland and N. Quinn, 313–42. Cambridge: Cambridge University Press. Quinn, N. 1982. “Commitment” in American marriage: A cultural analysis. American Ethnologist 9: 755–98. Quinn, N. 1987. Convergent evidence for a cultural model of American marriage. In Cultural models in language and thought, ed. N. Quinn and D. Holland, 173–92. Cambridge: Cambridge University Press.

14_107-Bernard.indb 556

6/9/14 10:03 AM

17: Text Analysis

557

Quinn, N. 1992. The motivational force of self-understanding: Evidence from wives’ inner conflicts. In Human motives and cultural models, ed. R. D’Andrade and C. Strauss, 90–126. New York: Cambridge University Press. Quinn, N. 1996. Culture and contradiction: The case of Americans reasoning about marriage. Ethos 24: 391–425. Quinn, N. 1997. Research on the psychodynamics of shared task solutions. In A cognitive theory of cultural meaning, ed. C. Strauss and N. Quinn, 189–209. Cambridge: Cambridge University Press. Quinn, N. 2005. How to reconstruct schemas people share. In Finding culture in talk: A collection of methods, ed. N. Quinn, 33–81. New York: Palgrave Macmillan. Ragin, C. C. 1987. The comparative method: Moving beyond qualitative and quantitative strategies. Berkeley: University of California Press. Ragin, C. C. 1994. Introduction to qualitative comparative analysis. In The comparative poliical economy of the welfare state, ed. T. Janowski and A. M. Hicks, 299–317. Cambridge: Cambridge University Press. Ragin, C. C. 2000. Fuzzy-set social science. Chicago: University of Chicago Press. Ragin, C. C., K. A. Drass, and S. Davey. 2006. Fuzzy-set/qualitative comparative analysis 2.0. Tucson: Department of Sociology, University of Arizona. Rice, G. E. 1980. On cultural schemata. American Ethnologist 7: 152–71. Richards, T., and L. Richards. 1991. The NUDIST qualitative data analysis system. Qualitative Sociology 14: 307–25. Ridgeway, C. L. 2006. Linking social structure and interpersonal behavior: A theoretical perspective n cultural schemas and social relations. Social Psychology Quarterly 69: 5–16. Rihoux, B., and C. Ragin. 2009. Configurational comparative methods: Qualitative comparative analysis (QCA) and related techniques. Thousand Oaks, CA: Sage. Robinson, W. S. 1951. The logical structure of analytic induction. American Sociological Review 16: 812–18. Rohner, R. P. 1966. Franz Boas among the Kwakiutl. In Pioneers of American anthropology, ed. J. Helm, 213–22. Seattle: University of Washington Press. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–38. Rosenberg, S. D., P. P. Schnurr, and T. E. Oxman 1990. Content analysis: A comparison of manual and computerized systems. Journal of Personality Assessment 54: 298–310. Rumelhart, D. E. 1980. Schemata: The building blocks of cognition. In Theoretical issues in reading comprehension: Perspectives from cognitive psychology, linguistics, artificial intelligence, and education, ed. R. J. Spiro, B. C. Bruce, and W. B. Brewer, 33–58. Hillsdale, NJ: Lawrence Erlbaum. Ryan, G. 1999. Measuring the typicality of text: Using multiple coders for more than just reliability and validity checks. Human Organization 58: 313–22. Ryan, G., and H. R. Bernard. 2003. Techniques to identify themes. Field Methods 15: 85–109. Ryan, G., and T. Weisner. 1996. Analyzing words in brief descriptions: Fathers and mothers describe their children. Cultural Anthropology Methods Journal 8: 13–16. Salinas Pedraza, J. 1997. Saving and strengthening Indigenous Mexican languages: The CELIAC experience. In Indigenous literacy in the Americas, ed. N. H. Hornberger, 171–88. Berlin: Mouton de Gruyter. Salinas Pedraza, J., and H. R. Bernard. 1978. Rc Hnychnyu. The Otomí. Albuquerque: University of New Mexico Press. Sandelowski, M. 1995a. Qualitative analysis: What it is and how to begin. Research in Nursing and Health 18: 371–75.

14_107-Bernard.indb 557

6/9/14 10:03 AM

558

Amber Wutich et al.

Sandelowski, M. 1995b. Sample size in qualitative research. Research in Nursing and Health 18: 179–83. Sapir, E., and J. Curtain. 1974 [1909]. Wishram texts. New York: AMS Press. Sapir, E., and R. B. Dixon. 1910. Yana texts. University of California Publications in American Archeology and Ethnology 9(1). Berkeley: University of California Press. Sapir, E., and M. Swadesh. 1978 [1939]. Nootka texts: Tales and ethnological narratives. New York: AMS Press. Schank, R. C., and R. P. Abelson, 1977. Scripts, plans, goals, and understanding: An enquiry into human knowledge structures. Hillsdale, NJ: Lawrence Erlbaum. Schnegg, M., and H. R. Bernard. 1996. Words as actors: A method for doing semantic network analysis. Cultural Anthropology Methods Journal 8(2): 7–10. Schweizer, T. 1991. The power struggle in a Chinese community, 1950–1980: A social network analysis of the duality of actors and events. Journal of Quantitative Anthropology 3: 19–44. Schweizer, T. 1996. Actor and event orderings across time: Lattice representation and Boolean analysis of the political disputes in Chen Village, China. Social Networks 18: 247–66. Seidel, J., and U. Kelle. 1995. Different functions of coding in the analysis of textual data. In Computer-aided qualitative data analysis: Theory, methods and practice, ed. U. Kelle, 52–61. London: Sage. Sherzer, J. 1990. Verbal art in San Blas. Cambridge: Cambridge University Press. Sherzer, J. 1994. The Kuna and Columbus: Encounters and confrontations of discourse. American Anthropologist 96: 902–25. Sillitoe, P. 2007. Anthropologists only need apply: Challenges of applied anthropology. Journal of the Royal Anthropological Institute 13: 147–65. Stone, P. J., D. C. Dunphy, M. S. Smith, and D. M. Ogilvie, eds. 1966. The general inquirer: A computer approach to content analysis. Cambridge, MA: MIT Press. Stone-Jovicich, S. S., T. Lynam, A. Leitch, and N. A. Jones. 2011. Using consensus analysis to assess mental models about water use and management in the Crocodile River Catchment, South Africa. Ecology and Society 16: 45. http://www.ecologyandsociety.org/vol16/iss1/art45 (accessed February 12, 2013). Strauss, A. 1987. Qualitative analysis for social scientists. Cambridge: Cambridge University Press. Strauss, A., and J. Corbin. 1990. Basics of qualitative research. Grounded theory procedures and techniques. Thousand Oaks, CA: Sage. Strauss, C. 1992. What makes Tony run? Schemas as motives reconsidered. In Human motives and cultural models, ed. R. D’Andrade and C. Strauss, 191–224. Cambridge: Press Syndicate of the University of Cambridge. Strauss, C. 1997. Partly fragmented, partly integrated: An anthropological examination of “postmodern fragmented subjects.” Cultural Anthropology 12: 362–404. Strauss, C. 2005. Analyzing discourse for cultural complexity. In Finding culture in talk: A collection of methods, ed. N. Quinn, 157–202. New York: Palgrave Macmillan. Swanton, J. R. 1909. Tlingit myths and texts. Washington, DC: Government Printing Office. Swartz, M. C. 1997. Illness and mortality in the Mombasa Swahili community: A metaphorical model in an Islamic culture. Culture, Medicine, and Psychiatry 21: 89–114. Taylor, S. J., and R. Bogdan. 1984. Introduction to qualitative research methods, 2nd ed. New York: Wiley. Teddlie, C., and A. Tashakkori. 2003. Preface. Handbook of mixed methods in social and behavioral research. Thousand Oaks, CA: Sage. Tedlock, D. 1972. Finding the center: Narrative poetry of the Zuni Indians. New York: Dial.

14_107-Bernard.indb 558

6/9/14 10:03 AM

17: Text Analysis

559

Tedlock, D. 1985. Popol Vuh: The Mayan Book of the Dawn of Life and the Glories of Gods and Kings. New York: Simon and Schuster. Tedlock, D. 1987. Hearing a voice in an ancient text: Quiché Maya poetics in performance. In Native American discourse: Poetics and rhetoric, ed. J. Sherzer and A. Woodbury, 140–75. Cambridge: Cambridge University Press. Tesch, R. 1990. Qualitative research: Analysis types and software tools. New York: Falmer Press. Trost, J. E. 1986. Statistically nonrepresentative stratified sampling: A sampling technique for qualitative studies. Qualitative Sociology 9: 54–57. Tyler, S. A., ed. 1969. Cognitive anthropology. New York: Holt, Rinehart and Winston. Waitzkin, H., and T. Britt. 1993. Processing narratives of self-destructive behavior in routine medical encounters: Health promotion, disease prevention, and the discourse of health care. Social Science and Medicine 36: 1121–36. Waldram, J. B. 2008. The narrative challenge to cognitive behavioral treatment of sexual offenders. Culture, Medicine, and Psychiatry 32: 421–39. Wegerif, R., N. Mercer, and S. Rojas-Drummond. 1999. Language for the social construction of knowledge: Comparing classroom talk in Mexican preschools. Language and Education 13: 133–50. Worthman, C. M., J. DeCaro, and R. Brown. 2002. Cultural consensus approaches to the study of American family life. MARIAL Working Paper 13, Department of Anthropology, Emory University, Atlanta. Wutich, A. 2009. Intrahousehold disparities in women and men’s experiences of water insecurity and emotional distress in urban Bolivia. Medical Anthropology Quarterly 23: 436–54. Wutich, A., T. Lant, D. White, K. Larson, and M. Gartin. 2010. Comparing focus group and individual responses on sensitive topics: A study of water decision-makers in a desert city. Field Methods 22: 88–110. Znaniecki, F. 1934. The method of sociology. New York: Farrar and Rinehart.

14_107-Bernard.indb 559

6/9/14 10:03 AM

14_107-Bernard.indb 560

6/9/14 10:03 AM

CHAPTER 18

Cross-Cultural Research Carol R. Ember, Melvin Ember, and Peter N. Peregrine

Cultural anthropology, among other things, is a comparative discipline. We who call ourselves cultural anthropologists like to emphasize how customs vary from place to place and how they may change over time. Indeed, we delight in the diversity of human cultures. Yet many cultural anthropologists are uncomfortable with the idea of explicit, systematic, cross-cultural comparison—the subject of this chapter. One reason for the discomfort may be our emphasis on fieldwork. We train several years for our fieldwork and spend a lot of time in the field. Fieldwork is central to our professional lives as well as to the discipline. The usual objective of fieldwork is to discover the details and particulars of a single community or culture. Those details remind us that each culture is unique, its combination of patterns of behavior and belief like no other. To compare cultures is not to deny their individual uniqueness. Ethnography tells us what is distinctive about a particular culture; cross-cultural comparison tells us about what is generally true for some, many, or even all human cultures. To generalize across cultures, we build on the particulars of ethnographies to formulate statements about the similarities and differences of cultures and what they may be related to. The serious epistemological issue is whether it is possible to formulate such general statements in the first place. Cross-culturalists argue that it is. We focus in this chapter on methods for systematic comparisons across cultures— comparisons that, we expect, will answer questions about the incidence, distribution, and causes of cultural variation (see C. R. Ember and M. Ember 2009 and Special Issue 1991). The methods are familiar—unbiased sampling, repeatable measurements, statistical evaluation of results, and the like. The relationship between cross-cultural research and ethnography is analogous to that between epidemiology and clinical practice in medicine. In ethnographic research and in clinical practice, the focus is on the individual case, while in cross-cultural research, as in epidemiology, the focus is on populations. Epidemiologists look at the incidence and distribution of diseases across populations and try to understand the causes of those diseases, primarily through correlational analyses of presumed causes and effects. Similarly, cross-cultural researchers are interested in causes and effects of cultural variation across the world or across regions of the world. UNIQUENESS AND COMPARABILITY

To illustrate how things can be unique and comparable at the same time, consider the following ethnographic statements about sexuality in three different cultures: 561

14_107-Bernard.indb 561

6/9/14 10:03 AM

562

Carol R. Ember et al.

1. The Mae Enga in the Western Highlands of Papua New Guinea believe that “copulation is in itself detrimental to male well-being. Men believe that the vital fluid residing in a man’s skin makes it sound and handsome, a condition that determines and reflects his mental vigor and self-confidence. This fluid also manifests itself as his semen. Hence, every ejaculation depletes his vitality, and over-indulgence must dull his mind and leave his body permanently exhausted and withered” (Meggitt 1964, 210). 2. “The Nupe men [of Nigeria], certainly, make much of the physically weakening effects of sexual intercourse, and teach the younger generation to husband their strength” (Nadel 1954, 179). 3. “[T]he milk avoidances of the Nilotes [Shilluk of the Sudan] are dependent on fear of contamination associated with the sexual act. . . . Only small boys herd the cattle and milk them, for once a boy has reached maturity there is the danger that he may have had sexual contact, when if he milked, or handled manure, or even walked among the cattle in their pens, he would cause them to become sterile. . . . If a man has had sexual relations with his wife or another he is considered unclean and does not drink milk until the sun has set the following day” (Seligman and Seligman 1932, 73).

Each statement about male sexuality is unique, but there are also similarities in these statements that suggest a continuum—a variation in the degree to which males in a society believe that heterosexual sex is harmful to their health. Enga and Nupe males apparently think that heterosexual sex is harmful to them. Shilluk males think that heterosexual sex would cause harm to their cattle and seemingly to cows’ milk. The Shilluk statements are not clearly about harm to men’s own health. But if we ask “Do people believe that male heterosexuality (even with legitimate partners) brings some harm or danger?” we would have to say that all three of the cultures mentioned had such a belief. The important point here is that similarities cannot be seen or recognized until we think in terms of variables, qualities or quantities that vary along a specified dimension. There is no right or wrong conceptualization of variables; researchers may choose to focus on any aspect of variation. But once researchers perceive and specify similarity, they can perceive and recognize difference. Measurement—deciding how one case differs from another in terms of some scale—is but a short conceptual step away. Consider now the following ethnographic statements: 4. For the Cuna of Panama, “the sexual impulse is regarded as needing relief, particularly for males, and as an expression of one’s niga, a supernatural attribute manifested in potency and strength. On the other hand it is considered debilitating to have sexual relations too often, for this will weaken one’s niga” (Stout 1947, 39). 5. And in regard to the Bedouin of Kuwait, “It [sexual intercourse] is the one great pleasure common to rich and poor alike, and the one moment of forgetfulness in his daily round of troubles and hardships that Badawin [Bedouin] or townsmen can enjoy. Men and women equally love the act, which is said to keep a man young, “just like riding a mare” (Dickson 1951, 162).

The Bedouin beliefs contrast most sharply with the beliefs in the other cultures, because heterosexual intercourse appears to be viewed by them as purely pleasurable, with no negative associations. The Cuna seem to be somewhere in the middle. While they view sex as important, they appear to believe that too much is not good.

14_107-Bernard.indb 562

6/9/14 10:03 AM

18: Cross-Cultural Research

563

The variable “degree of men’s fear of sex with women” can be conceptualized as a continuum with gradations. In a cross-cultural study of this variable, the first author identified four scale points: Societies with mostly negative statements (in the ethnography) about heterosexuality were considered high on men’s fear of sex with women; societies with more or less an equal number of negative and positive statements were considered ambivalent; those with mostly positive statements were considered relatively low on men’s fear of sex with women; and those with only positive statements were considered as lacking men’s fear of sex with women. While the variable as operationally defined does not capture everything in a culture’s beliefs about heterosexuality, it does capture some distinguishable similarities and differences across cultures (C. R. Ember 1978a). These examples show that if we focus on a specified aspect or dimension of the variation, similarities and differences become apparent. Framing meaningful and answerable questions, and identifying useful dimensions to do so, is an art. To be meaningful, a question should have some theoretical (generally explanatory) importance. To be answerable, the question should be phrased in a way that allows us to get to an answer on the basis of research. In cross-cultural research, as in most social science research, framing the question (often in a single sentence) is half the battle. Once we frame a question in answerable terms, it is not hard to decide how to go about seeking an answer. And we do not have to limit ourselves to one possible answer; often cross-cultural research involves testing several, not necessarily alternative, answers to the question at issue. FRAMING THE RESEARCH QUESTION: THE KINDS OF QUESTIONS ASKED

There are at least four kinds of questions we can ask in cross-cultural research: 1. Descriptive/statistical questions. These deal with the prevalence or frequency of a trait. How common is the belief that sex is dangerous to one’s health? What proportion of societies have it? How common is polygyny in the world’s societies? 2. Questions about causes of a trait or custom. Examples are: Why do some societies have the belief that heterosexual sex is harmful? Why do some societies insist on monogamous marriage, whereas most allow polygyny (multiple wives)? Why is war very frequent in some societies and less frequent in others? 3. Questions about the consequences or effects of a particular trait or custom. This kind of question may be phrased broadly: What are the effects of growing up in a society with a great deal of war? Or a consequence question may be phrased much more specifically: What is the effect of polygyny on fertility? 4. Questions that are nondirective and relational. Rather than theorizing about causes or consequences, a researcher may simply ask if a particular aspect of culture is associated with some other aspects. Is there a relationship between type of marriage and level of fertility? Is more war associated with more socialization for aggression in children? No causal direction is specified beforehand with a nondirective relational question.

Of the four types of questions, the descriptive/statistical type is the easiest to address because it tells the researcher what to count in a representative sample of societies. To estimate the frequency of monogamy versus polygyny, we need to establish what each society allows and make a count of each kind of society. The consequence and

14_107-Bernard.indb 563

6/9/14 10:03 AM

564

Carol R. Ember et al.

relational questions usually specify a set of concrete things to look at. If you want to know whether type of marriage has an effect on or is related to fertility, then you know you need to measure both variables (type of marriage and fertility). Open-ended causal (and consequential) questions are the most challenging. The only thing specified in the in the open-ended causal question is the dependent variable (the variable to be explained); the only thing specified in the open-ended consequential question is the independent variable (the variable that may have effects). Exactly which variables may be causes or effects is something the investigator has to decide on, often as suggested by some theory. Like a detective who theorizes about suspects and their motives and opportunities, the pursuit of causes involves testing alternative explanations or theories that specify why something is the way it is or how it came to be that way. In science, theories in the literature are usually starting points, but new theories can also be tested. The basic assumption of cross-cultural research is that comparison is possible because repeated patterns can be identified. Cross-culturalists believe that all generalizations about culture require testing, no matter how plausible we may think they are. This applies to descriptive generalizations presumed to be true (e.g., the presumption that hunter–gatherers are typically peaceful) as well as to presumed relationships or associations (e.g., the presumption that hunting is more likely to be associated with patrilocality). As it turns out, neither presumption is generally true of hunter–gatherers (C. R. Ember 1975, 1978b). It is necessary to test all presumed generalizations or relationships because they may be wrong, and we are entitled (even obliged) to be skeptical about any one that has not been tested and supported by an appropriate statistical test. Cross-culturalists do not believe that cross-cultural research is the only way to test theory about cultural variation. However, such research is viewed as one of the important ways to test theory, if only because cross-cultural research (if it involves a worldwide sample of cases) provides the most generalizable results of any kind of social scientific research. Most cross-culturalists believe in the multi-method approach to testing theory; that is, they believe that worldwide cross-cultural tests should be supplemented (when possible and not impossibly costly) by studying variation in one or more particular field sites (comparisons of individuals, households, communities) as well as within particular regions (e.g., North America, Southeast Asia); theories may also be testable using historical data, experiments, and computer simulations. Before we discuss the advantages and disadvantages of the various kinds of comparative and cross-cultural research, here is a little historical background. HISTORY OF CROSS-CULTURAL RESEARCH

The first cross-cultural study was published in 1889 by Edward B. Tylor. In that study, Tylor attempted to relate marital residence and the reckoning of kinship to other customs, such as joking and avoidance relationships. But perhaps because of Francis Galton’s objection to Tylor’s presentation—see the “Discussion” section at the end of Tylor’s paper—little cross-cultural research was done for the next 40 years. (We’ll discuss what has come to be called “Galton’s Problem” later.) Cross-cultural research started to become more popular in the 1930s and 1940s, at the Institute of Human Re-

14_107-Bernard.indb 564

6/9/14 10:03 AM

18: Cross-Cultural Research

565

lations at Yale. The person who led this rebirth was anthropologist George Peter Murdock (1897–1985). Murdock had obtained his Ph.D. at Yale in a combined sociology/ anthropology department that called itself the “Science of Society” (its comparative perspective was established by its founder, William Graham Sumner). Perhaps the major boost to cross-cultural studies was the Yale group’s development of an organized collection of ethnographic information (first called the “CrossCultural Survey,” the precursor of the Human Relations Area Files) that scholars could use to compare the cultures of the world. In the early part of the twentieth century, Sumner had compiled voluminous materials on peoples throughout the world, but his compilation was limited to subjects in which he was personally interested. Later, at the Institute of Human Relations, the group of social and behavioral scientists led by Murdock (including psychologists, physiologists, sociologists, and anthropologists) set out to improve on Sumner’s work by developing the Cross-Cultural Survey. The aim was to foster comparative research on humans in all their variety so that explanations of human behavior would not be culture bound. The first step was to develop a classification system that would organize the descriptive information on different cultures. This category system became the Outline of Cultural Materials (Murdock et al. 2008). The next step was to select well-described cases and to index the ethnography on them, paragraph by paragraph, sometimes even sentence by sentence, for the subject matters covered on a page. As the aim was to file information by subject category to facilitate comparison, and since a page usually contained more than one type of information, the extracts of ethnographic information were typed using carbon paper to make the number of copies needed (corresponding to the number of topics covered); carbon paper was used because the Cross-Cultural Survey antedated photocopying. In 1946, the Cross-Cultural Survey directorship was turned over to Clellan S. Ford, who undertook to transform it into a consortium of universities. In 1949, first five then eight universities joined together to sponsor a not-for-profit consortium called the Human Relations Area Files, Incorporated (HRAF), with headquarters in New Haven, Connecticut. Over the years, the HRAF consortium added member institutions, and more and more cultures were added to the HRAF collection. In the early days, member institutions received the annual installments of information on xeroxed sheets of paper. Later, the preferred media became microfiche. Since 1994, the only media have been electronic, first CD-ROM and now online (titled eHRAF World Cultures [http:// ehrafworldcultures.yale.edu]). Technological changes have allowed the full-text HRAF Collection of Ethnography to become more and more accessible and more and more efficiently searchable. New cases are added each year. Today, when old cases are added to the electronic HRAF, they are updated if possible. (For more information on the evolution of HRAF, see M. Ember 1997.) The accessibility that HRAF provides to indexed ethnographic information has undoubtedly increased the number of cross-cultural studies. Considering just worldwide comparisons (the most common type of cross-cultural study), in the years between 1889 and 1947 there were only 10 worldwide cross-cultural studies. In the following 20, years there were 127. And in the next 20 years (ending in 1987), there were 440 (C. R. Ember and Levinson 1991, 138). At present, there are probably 1,000 worldwide

14_107-Bernard.indb 565

6/9/14 10:03 AM

566

Carol R. Ember et al.

cross-cultural studies in the published literature. (HRAF is compiling a bibliography of cross-cultural studies; this number is just an estimate.) TYPES OF CROSS-CULTURAL COMPARISON

Cross-cultural comparisons vary along four dimensions: (1) geographical scope of the comparison—whether the sample is worldwide or is limited to a geographic area (e.g., a region such as North America); (2) size of the sample—two-case comparisons, small-scale comparisons (fewer than 10 cases), and larger comparisons; (3) whether the data used are primary (collected by the investigator in various field sites explicitly for the comparison) or secondary (collected by others and found by the investigator in ethnographies, censuses, and histories); and (4) whether the data on a given case pertain to (or date from) just one time period (a synchronic comparison of cases) or two or more time periods (a diachronic comparison). Although all combinations of the four dimensions are technically possible, some combinations are quite rare. Worldwide cross-cultural comparisons using secondary synchronic data (one “ethnographic present” for each case) are the most common in anthropology. Comparative research is not just done in anthropology. Worldwide studies using ethnographic data are increasingly done by evolutionary biologists, sociologists, political scientists, and others. Cross-cultural psychologists often compare people in different cultures. And various kinds of social scientists compare across nations. The cross-national comparison is narrower than the worldwide cross-cultural comparison because the results of a cross-national comparison are generalizable only to a limited range of cross-cultural variation—that which encompasses only the complex societies (usually multicultural nation-states) of recent times. The results of a cross-cultural study are generalizable to all types of society, from hunter–gatherers—with populations in the hundreds or a few thousand—to agrarian state societies with populations in the millions—to modern nation-states—with populations in the hundreds of millions.1 Cross-national research differs from cross-cultural research in ways other than generalizability. Economists, sociologists, and political scientists usually use secondary data when they study samples of nations, but the data are not generally ethnographic. That is, the measures used are not based on cultural information collected by anthropologists or other investigators in the field. Rather, the data used in cross-national comparisons are generally based on censuses and other nationally collected statistics (crime rates, gross national product, etc.), often documented over time. Cross-cultural psychologists are most likely to collect their own (primary) data, but their comparisons tend to be the most limited; very often only two or a few cultures are compared. Cross-historical studies are still comparatively rare in anthropology (but see Naroll et al. 1974; Peregrine 2001; Peregrine et al. 2004). Few worldwide or even within-region cross-cultural studies have employed data on a given case for more than one time period. But, as noted, some cross-national studies have been cross-historical studies as well, and cross-historical studies using archaeological data are becoming increasingly common (see, e.g., Peregrine 2003; the papers in M. E. Smith 2012). Both cross-national and archaeologically based cross-historical studies can be undertaken by scholars in a wide variety of disciplines (e.g., sociology, history, political science) because there are accessible historical databases. Since primary data are so hard and expensive to collect, it

14_107-Bernard.indb 566

6/9/14 10:03 AM

18: Cross-Cultural Research

567

is hardly surprising that primary comparisons are likely to be small in scale. If you have to collect your data by yourself, going to several places to do so takes a lot of time and money. Large-scale comparisons, which almost always rely on secondary data (collected or assembled previously by others), are generally much less expensive. Let us turn now to the advantages and disadvantages of the different types of crosscultural comparison. Our discussion compares the different types with particular regard to theory formulation and theory testing. (Cross-cultural research may also be done to establish the incidence or relative frequency of something.)

Advantages and Disadvantages of the Different Types of Comparison Worldwide cross-cultural comparisons have two major advantages, compared with other types of comparative research (M. Ember 1991). The major one, as already noted, is that the statistical conclusions drawn from a worldwide comparison of all types of society are probably applicable to the entire ethnographic record, assuming that the sample is more or less free of bias. (See the section on Sampling, below.) This contrasts with results of a within-region comparison, which may or may not be applicable to other regions. And it contrasts with results of a cross-national comparison, which may or may not be applicable to the ethnographic record. The worldwide type of crosscultural comparison, then, has a better chance than other types of coming close to the goal of knowing that a finding or an observed relationship has nearly universal validity, consistent with the general scientific goal of more and more comprehensive explanations. (Most cross-cultural studies undersample modern industrial societies, but this deficiency will decrease as the ethnographic record increasingly includes the results of ethnographic field studies in industrial countries.) The other advantage of worldwide cross-cultural comparison is that it maximizes the amount or range of variation in the variables investigated. This may make the difference between a useful and a useless study. Without variation, it’s impossible to see a relationship between variables. Even if there’s some variation, it may be at just one end of the spectrum of variation. We may think the relationship is positive or negative, because that is all we can observe in one region or in one type of society, but the relationship may be curvilinear in the world, as John Whiting (1954, 524–25) noted. This is what occurs when we plot socioeconomic inequality against level of economic development. There is little socioeconomic inequality in hunter–gatherer societies; there is a lot of inequality in premodern agrarian societies; and inequality is lower again—but hardly absent—in modern industrial societies (M. Ember et al. 1997). If we only looked at industrial societies, it would appear that more equality goes with higher levels of economic development. But if we only looked just at preindustrial societies, it would appear that more equality goes with lower levels of economic development. Thus, to be sure about the nature or shape of a relationship, we have to conduct a worldwide cross-cultural comparison because that shows the maximum range of variation and is the most reliable way to discover the existence and nature of relationships between or among variables pertaining to humans. When a researcher compares many societies from different parts of the world, she or he is unlikely to know much about each one. If a tested explanation turns out to be supported, the lack of detailed knowledge about the sample cases isn’t much of a problem.

14_107-Bernard.indb 567

6/9/14 10:03 AM

568

Carol R. Ember et al.

However, if the cross-cultural test is disconfirming, it may be difficult to come up with an alternative explanation without knowing more about the particular cases. More familiarity with the cases may help in formulating a revised or new theory that could be tested and supported (Johnson 1991). Narrowing the scope of a comparative study to a single region may mean that you can know more about the cases and therefore may be more likely to come up with a revised theory if your first tests are unsuccessful. Restricting the study to a region doesn’t allow you to discover that the results of the comparison apply to the whole world. (See Burton and White [1991] for more discussion of regional comparisons.) Even a regional comparativist may not know all the cases in the region; that depends mostly on the size of the region. If the region is as large as North America, the comparativist is likely to know less about the cases than if the region studied is the American Southwest. And if you need to look at a sample of the rest of the world to discover how generalizable your results are, you might as well do a worldwide study in the first place! The objective of large-scale within-region comparisons (using data on all or most of the societies in the region) is usually different from the objective of a worldwide crosscultural study (Burton and White 1991). Using large numbers of cross-cultural traits, within-region comparativists generally try to arrive at classifications of cultures in order to make inferences about processes of diffusion and historical ancestry. Instead of trying to see how culture traits may be causally related to each other, within-region comparativists are usually more interested in trying to see how the cultures in the region are related to each another. But some regional comparativists are interested in pursuing both objectives at the same time (Jorgensen 1974). The importance of looking at worldwide as well as regional samples, especially if you are interested in relationships between or among variables, is indicated by the following discussion. Consider the discrepancy between the findings of Driver and Massey (1957) and the findings of M. Ember and C. R. Ember (1971), and Divale (1974) with regard to the relationship between division of labor by gender and where couples live after they get married. Driver and Massey found support in aboriginal North America for Murdock’s (1949, 203ff.) idea that division of labor by gender determined matrilocal versus patrilocal residence, but the Embers (and Divale) found no support for this idea in worldwide samples. The Embers found that the relationship varies from region to region. In North America, there is a significant relationship, but in other regions the relationship is not significant. And in Oceania, there is a trend in the opposite direction—matrilocal societies there are more likely to have men doing more in primary subsistence activities. What might account for the difference in the direction of the relationship in different regions? C. Ember (1975) found that the relationship between a male-dominated division of labor and patrilocal residence did hold for hunter– gatherers; hence, she suggested that the frequent occurrence of hunter–gatherers in North America (see Witkowski N.d.) may account for the statistically significant relationship between division of labor and residence in North America. Fred Eggan (1954) advocated small-scale regional comparisons, which he called “controlled comparisons” because he thought they would make it easier to control on similarity in history, geography, and language. He presumed that the researcher could readily discern what accounts for some aspect of cultural variation within the region if

14_107-Bernard.indb 568

6/9/14 10:03 AM

18: Cross-Cultural Research

569

history, geography, and language were held constant. However, the similarity of cases within a region may be a major drawback. A single region may not show sufficient variability in the aspect of culture (or presumed causes) the researcher is investigating. Unless a substantial number of the cases lack what you are trying to explain, it would be difficult or impossible to discern what the phenomenon at issue may be related to. For example, suppose almost all the cases in a region share beliefs about sexuality being somewhat harmful. It would be difficult or nearly impossible to be able to figure out what this belief is related to because you could not tell which of the other regularly occurring practices or beliefs in the region might explain the sexual beliefs. Only if some cases lack what you are trying to explain might you see that the hypothetical causes are also generally absent when the presumed effect is absent. Unless there is sufficient variation in all possibly relevant variables, the controlled comparison strategy is a poor choice for testing theory. Obviously, the controlled comparison is also a poor choice for describing the worldwide incidence of something (unless the region focused on is the only one of interest). While the strategy of controlled comparison may seem analogous to controls in psychology and sociology (which hold some possible causes, and their effects, constant), the resemblance is only superficial. Psychologists and sociologists typically eliminate certain kinds of variation (e.g., in race, religion, ethnicity, or gender) only when they have prior empirical reasons to think that these factors partially predict the variable they are trying to explain. In contrast, those who do controlled comparisons in the anthropological sense usually only presume that common history, language, or geography have made a difference. If researchers aren’t really controlling on the important predictors when they do a controlled comparison, they aren’t necessarily getting any closer to the causal reality by restricting their study to a particular region. The researcher must collect primary data, in the field if he or she is interested in topics that are rarely (if ever) covered adequately by ethnographers. This was the major reason why John and Beatrice Whiting (see, e.g., B. B. Whiting 1963), who were interested in children’s behavior and what it was related to, decided that they had to collect new data in the field for the comparative study that came to be known as the six cultures project. Many aspects of socialization (such as particular practices of the mother) weren’t typically described in ethnographies. Similarly, researchers interested in internal psychological states (such as sex-identity, self-esteem, and happiness) couldn’t find out about them from ethnographies and therefore would need to collect the data themselves, in the field. How time is allocated to various activities is a nonpsychological example of information that is also not generally covered in ethnographies, or not in sufficient detail. Although it may always seem preferable to collect primary data as opposed to secondary data, the logistics of cross-cultural comparisons using primary data are formidable in time and expense. And the task of maintaining comparability of measures across sites is difficult (R. L. Munroe and R. H. Munroe 1991a). If a researcher thinks that something like the needed information is already available in ethnographies, a comparison using secondary data is more economical than comparative fieldwork in two or more places. But comparative fieldwork may be the only viable choice when the information needed is not otherwise available.

14_107-Bernard.indb 569

6/9/14 10:03 AM

570

Carol R. Ember et al.

Similarly, although it may seem preferable to use historical diachronic data to test the temporal ordering implied in causal theories, such data are not often readily available. Because most societies that cultural anthropologists studied lacked native writing, there are usually no historical documents to use for measuring variables for an earlier time. The alternative is to reconstruct the situation in a prior time period using oral history and the occasional documents left by travelers, traders, and other visitors. Such reconstructions are notoriously subject to bias (because of wishful thinking by the researcher). It is also difficult to find diachronic data because different ethnographers have different substantive interests, so different ethnographers who may have worked in the same place at different times may not have collected information on the same variables. For these reasons, most cross-culturalists think it is more efficient to test causal theories with synchronic data first. If a theory has merit, the presumed causes and effects should generally be associated synchronically. If they are, then we might try to make a diachronic or cross-historical test. If they are, then we might try to see if the presumed causes antedated the presumed effects unless we see first that the synchronic results show correlation. Diachronic studies may get a lift from eHRAF Archaeology (http://ehrafarchaeology .yale.edu)—a database with archaeological traditions across the span of prehistory. In recent years, a large body of diachronic archaeological data have been assembled that could be used to test causal theories (Peregrine 2003). Several such tests have been undertaken (e.g. Peregrine 2006; Peregrine et al. 2007). However, linking the archaeological record to cultural traits of interest to cross-culturalists is difficult and often controversial. There are established archaeological correlates of matrilineal descent, sedentarism, and warfare frequency (see Peregrine 2004 for a review), but few others, and some that have been proposed continue to be questioned (e.g., Longacre 1970). In addition, many societies in the HRAF Collection of Ethnography have more than one time focus.2 Diachronic data should become increasingly available as the ethnographic record expands with updating and archaeological data are made easier to use, so we should see more cross-historical studies in the future. SAMPLING

Whatever questions cross-cultural researchers want to answer, they always have to decide what cases to compare. How many cases should be selected for comparison and how should they be selected? If we want to answer a question about incidence or frequency in the world or in a region, it is critical that all the cases (countries or cultures) be listed in the sampling frame (the list to be sampled from, sometimes also called the “universe” or “population”). And all must have an equal chance to be chosen (the major reason sample results can be generalized to the larger universe of cases). Otherwise, it is hard to argue that sample results are generalizable to anything. It is often not necessary to investigate all cases or even to sample a high proportion of cases from the sampling frame. The necessity of sampling a high proportion of cases depends largely on the size of the population to generalize to. When the population is small, the percentage of cases that must be included to ensure generalizability can be high. When the population is large, the sample size, proportionately, can be quite

14_107-Bernard.indb 570

6/9/14 10:03 AM

18: Cross-Cultural Research

571

small. Except when relationships between contiguous cases are of interest, investigating all the cases would be a colossal waste of time and resources. Political opinion polling is a case in point. The number of voters in the United States is a very large number. Yet, very accurate results can usually be obtained by randomly sampling a few hundred to a few thousand individuals. The size of the sample is not as important as selecting the cases in some random way from a more or less complete list of voters. If you are looking for a large difference between one kind of voter and other, or if the relationship you are examining is strong, you do not have to have a very large sample to obtain statistically significant results. The most important consideration is to sample in an unbiased way, preferably using some kind of random sampling procedure (M. Ember and Otterbein 1991).

Sampling in Comparisons Using Primary Data There have been relatively few comparisons using primary data (collected by the researcher in two or more field sites) in anthropology. There have been a considerable number of two-case comparisons in cross-cultural psychology, usually comparing subjects in the United States with subjects in some other place. Generally, sampling in comparisons using primary data has been purposive rather than random. That is, the cases compared have not been randomly selected from some sampling frame. This is understandable, given the political realities of gaining permission to do fieldwork in certain countries. In terms of the cost of fieldwork, it is not surprising that two-case comparisons are more common than any other kind of comparison using primary data. Unfortunately, the scientific value of two-case comparisons is dubious. Years ago, Donald Campbell (1961, 344) pointed out that a difference between two cases could be explained by any other difference(s) between the cases. Let us consider a hypothetical example. Assume we are comparing two societies with different levels of fertility. We may think that the difference is due to a need for child labor because much agricultural and household work has to be done. As plausible as this theory may sound, we should be skeptical about it because many other differences between the two societies could be responsible for the difference in fertility. The high-fertility society may also have earlier weaning, a shorter postpartum sex taboo, better medical care, and so on. There is no way, using aggregate or cultural data on the two societies, to rule out the possibility that any of the other differences (and still others not considered here) may be responsible for the difference in fertility. If, however, you have data on a sample of mothers for each society, and measures of fertility and the possible causes for each mother, we could do statistical analyses that would allow us to narrow down the causal possibilities in these two societies. But as suggestive as these results might be, we cannot be sure about what accounts for the difference at issue because we still have only two sample societies. What is the minimum number of societies for a comparative test using primary data? If two variables are related, the minimum number of cases that might provide a statistically significant result—assuming unbiased sampling, errorless measurement, and a hypothesis that is true—is four (see R. L. Munroe and R. H. Munroe 1991b). Examples of four-case comparisons using primary data, which employed theoretical criteria for case-selection, are the four-culture project on culture and ecology in East

14_107-Bernard.indb 571

6/9/14 10:03 AM

572

Carol R. Ember et al.

Africa, directed by Walter Goldschmidt (1965), and the Munroes’ four-culture project on socialization (R. H. Munroe et al. 1984; R. L. Munroe and R. H. Munroe 1992). In the East Africa project, which was concerned with the effect of ecology/economy on personality and social life, two communities (one pastoral and one agricultural) in each of four cultures (two Kalenjin speaking and two Bantu speaking) were selected. The Munroes selected four cultures from around the world to examine the effects of variation in degree of father-absence and the degree of male-centered social structure.

Sampling in Comparisons Using Secondary Data You have to decide what your sampling frame is and what list of cases you want to generalize the sample results to. Will it be worldwide (all countries or all societies)? Will it be regional (a broad region like North America, or a narrower one like the cultures of the American Southwest)? When you specify your sampling frame, you’re also specifying your unit of analysis. A country isn’t necessarily equivalent to a society or culture in the anthropological sense. A country (or nation-state) is a politically unified population; it may, and often does, contain more than one culture or society. Conventionally, a culture is the set of customary beliefs and practices characteristic of a society, which, in turn, is a population that occupies a particular territory and speaks a common language not generally understood by neighboring populations. Once you know what you want to generalize the sample results to, you should sample from a list containing all the eligible cases. Cross-national researchers have no problem constructing a list of countries. Crosscultural researchers don’t yet have a complete list of the world’s described cultures. But there are large lists of cultures to sample from. Several published lists of societies have served as sampling frames for most crosscultural studies of the ethnographic record (largely the recent past). A few claim to accurately represent the world’s cultures, but we argue below that these claims are problematic and that cross-cultural researchers cannot yet generalize sample results to all cultures. Any claim about a relationship or about the proportion of societies that have a particular trait should be tempered by the recognition that the generalization is only applicable to the list sampled from, and only if the particular cases investigated constitute an unbiased sample of the larger list. Currently, available cross-cultural samples of the ethnographic record include the following (from largest to smallest): (1) the “Ethnographic Atlas” (1962ff., beginning in Ethnology 1 [1962], 113ff. and continuing intermittently over succeeding years and issues of the journal), with a total of 1,264 cases; (2) the “Summary” version of the “Ethnographic Atlas” (Murdock 1967), with a total of 862 cases; (3) the “World Ethnographic Sample” (Murdock 1957), with 565 cases; (4) the Atlas of World Cultures (Murdock 1981), with 563 cases; (5) the annually growing HRAF Collection of Ethnography, which covered 385 cultures as of 2013 (the HRAF sample is a collection of texts grouped by culture and indexed by topic for quick information retrieval; no precoded data are provided for the sample cases, in contrast to the situation for all of the other samples except the next one); (6) the “Standard Ethnographic Sample” (Naroll and Sipes 1973 and addenda in Naroll and Zucker 1974), with 273 cases; (7) eHRAF World Cultures, which covered 280 cases as of 2013; (8) the “Standard Cross-Cultural Sample”

14_107-Bernard.indb 572

6/9/14 10:03 AM

18: Cross-Cultural Research

573

(Murdock and White 1969), with 186 cases; and (8) the “HRAF Probability Sample” (HRAF 1967; Lagacé 1979; Naroll 1967—also now included and updated in eHRAF World Cultures), with 60 cases (this sample is also called the “HRAF Quality Control Sample,” for which some precoded data are available). Before we examine some of the claims made about these various samples, we first need to realize why it is necessary to use random sampling procedures. According to sampling theory, only random sampling provides an unbiased or representative sample of some larger population or sampling frame (Cochran 1977, 8–11; see also Kish 1987, 16). For example, simple random sampling (using a table of random numbers or a lottery type of selection procedure) guarantees that every case in the sampling frame has had an equal chance to be chosen. (Equiprobability of selection is assumed in most tests that estimate the statistical significance, or likely truth-value, of sample results.) To sample in a simple random fashion, all you have to do is make sure that all cases in the sampling frame are numbered uniquely (no repeats, no cases omitted). Researchers may sometimes choose other kinds of random sampling, such as systematic sampling (every nth case is chosen after a random start) or stratified random sampling (first dividing the sample into subgroups or strata and then randomly sampling from each). There are two kinds of stratified random sampling. In proportionate stratified random sampling, each subgroup is represented in proportion to its occurrence in the total population; in disproportionate stratified random sampling, some subgroups are overrepresented and others are underrepresented. Disproportionate stratified random sampling is used in cross-cultural research when the researcher needs to overrepresent a rare type of case in order to have enough such cases to study (as in a comparison of relatively rare hunter-gatherers with more common agriculturalists) or when a researcher wants to derive an accurate estimate of some parameter (e.g., mean, variance, or strength of an association) for a rare subgroup. Proportionate random sampling may reduce the sample size needed when there are marked differences between subgroups. However, stratified random sampling may not improve much on the accuracy obtainable with a simple random sample (Kish 1987, 33). In addition to the samples of the ethnographic record, there are now two samples describing “traditions” in prehistory. The first is a more-or-less complete description of the archaeological traditions in the prehistoric record. All the traditions are described in the Encyclopedia of Prehistory (Peregrine and M. Ember 2001–2002) with brief summaries and brief bibliographies. The second is eHRAF Archaeology. Modeled after eHRAF World Cultures, eHRAF Archaeology has documents subject-indexed by paragraph, eHRAF Archaeology now contains a random sample of the world’s prehistoric traditions, plus over six complete temporal sequences.

Comparing the Available Samples Three of the existing cross-cultural ethnographic samples were said to be relatively complete lists at the times they were published. The largest is the complete “Ethnographic Atlas” (with 1,264 cases), published from 1962 on in the journal Ethnology). But as its compiler (Murdock 1967, 109) noted, not even the atlas is an exhaustive list of what he called the “adequately described” cultures; he acknowledged that East

14_107-Bernard.indb 573

6/9/14 10:03 AM

574

Carol R. Ember et al.

Eurasia, the Insular Pacific, and Europe were not well represented.3 For the smaller summary version of the atlas (Murdock 1967), he dropped all cases he considered poorly described. So, if you want your sampling frame to include only well-described cases (in Murdock’s opinion), then the 1967 Atlas Summary (with 862 cases) is a reasonable list to sample from. Raoul Naroll set out to construct a list of societies that met his stringent criteria for eligibility. Some of his criteria were: The culture had to lack a native written language; it had to have an ethnographer who lived for at least a year in the field; and the ethnographer had to know the native language. The resultant sample, which he called the “Standard Ethnographic Sample” (Naroll and Sipes 1973; see also Naroll and Zucker 1974), contains 285 societies; Naroll and Sipes claimed that this list was about 80–90% complete for the cultures that qualified at the time (eastern Europe and the Far East were admittedly underrepresented). Three of the existing samples (from largest to smallest: the Atlas of World Cultures, the Standard Cross-Cultural Sample, and the HRAF Probability Sample Files) were developed to give equal weight to each of a number of culture areas (areas of similar cultures) in the world. Technically, the samples mentioned in this paragraph are all disproportionate stratified samples (only the HRAF Probability Sample Files uses random sampling to select cases for each culture area identified). The sampling is disproportionate from the strata because the number of cases selected for each identified culture area is not proportionate to the real number of cases in the culture area. The presumption behind all of these stratified samples is that the cultures in a given area are bound to be very similar by reason of common ancestry or extensive diffusion. The designers of these samples wanted to minimize Galton’s Problem. In the next-to-last section of this chapter, we discuss whether or not Galton’s Problem really is a problem as well as nonsampling solutions to the presumed problem. There are difficulties with these disproportionate stratified samples. First, exactly how we should define and separate culture areas requires empirical testing; Burton et al. (1996) have shown that social structural variables do not cluster consistently with Murdock’s major cultural regions. Second, disproportionate stratified sampling is a less efficient way to sample (i.e., it requires more cases) than simple random sampling. Third, even if the disproportionate sample uses random sampling from each stratum, every case selected will not have had an equal chance to be chosen. This makes it difficult or impossible to estimate the commonness or uniqueness of a particular trait in the world. If we do not know how common a trait is in each culture area, we cannot correct our counts by relative weighting, which we would need to do to make an accurate estimate of the frequency of the trait in the world. Many have used all or some of the cases in the Standard Cross-Cultural Sample (SCCS) (Murdock and White 1969) for cross-cultural studies, at least partly because the published literature contains a large number of codes (ratings of variables) on those cases; many of these codes were reprinted in Barry and Schlegel (1980). This sample is claimed to be representative of the world’s known and well-described cultures (as of 1969), but that claim is dubious for two reasons. First, disproportionate sampling does not give an equal chance for each culture to be chosen. Second, the single sample case from each cluster was chosen judgmentally, not randomly. However, Gray (1996) com-

14_107-Bernard.indb 574

6/9/14 10:03 AM

18: Cross-Cultural Research

575

pared results from the SCCS with a thousand random samples from the Ethnographic Sample and found little evidence of bias in the SCCS. He claims that the only bias found was the inclusion of better-described societies. (Judgmental criteria were also used to choose the five cases per culture area for the Atlas of World Cultures [Murdock 1981].) Of the three ethnographic samples discussed here, the HRAF Probability Sample is the only one employing random sampling within strata (the 60-culture sample includes a random selection from each identified culture area). However, the other criteria for selection were so stringent (e.g., at least 1,200 pages of cultural data focused on a community or other delimited unit; substantial contributions from at least two different authors) that only 206 societies in the whole world were eligible for inclusion as of the time the sample was constructed (in the late 1960s). Two other samples should be briefly discussed, because researchers have used them as sampling frames in cross-cultural studies. One is the World Ethnographic Sample (Murdock 1957). In addition to being based on judgmental sampling of cases within culture areas, it has one other major drawback. A time and place focus is not specified for the cases, as in the two Ethnographic Atlas samples (full and summary), the Standard Cross-Cultural Sample, and the Standard Ethnographic Sample. If researchers want to use some of the precoded data for the World Ethnographic Sample, they would have no idea what the “ethnographic present” is for a case. As we discuss later, focusing on the same time and place for all the measures on a case is usually called for in testing for an association; you may be introducing error if you do not make sure that the measures used pertain to the same time and place for the case. Finally, let’s turn to the entire HRAF Collection of Ethnography. Like most of the other samples, it, too, was based on judgmental selection. But because it covers many cultures all over the world and provides ethnographic materials that are complexly indexed for rapid information retrieval, the collection has often been used as a sampling frame for cross-cultural studies. (As of 2013, the HRAF collection covered 385 cultures, at least 44% of the world’s well-described cultures if you go by Murdock’s [1967] total of 862 in the summary version of the Ethnographic Atlas.) The major advantage of the HRAF collection, as compared with the other available lists or samples of cultures, is that only HRAF provides ethnographic texts on the cases. The other samples provide only coded data (usually) and only limited bibliography (usually). If the codes constructed by others don’t directly measure what you are interested in, but you use them anyway, you may be reducing your chances of finding relationships and differences that truly exist. Hence, if you need to code new variables or you need to code something in a more direct way, you are likely to do better if you yourself code from the original ethnography (C. R. Ember et al. 1991). Library research to do so would be very time consuming, which is why HRAF was invented in the first place. If you use the HRAF collection, you don’t have to devote weeks to constructing bibliographies for each sample case; you don’t have to chase down the books and other materials you need to look at, which might otherwise have to be obtained by interlibrary loan; and you don’t have to search through every page of a source (that often lacks an index) to find all the locations of the information you seek. The HRAF collection gives you the information you want on a particular topic, from all of the sources

14_107-Bernard.indb 575

6/9/14 10:03 AM

576

Carol R. Ember et al.

processed for the culture, in a single place. That place, with the electronic HRAF, is now online. If you want to examine the original ethnography on a case (with all the context), and particularly if you want to construct your own measures, there is no substitute for the HRAF collection. If you are starting out to do your first cross-cultural study, how should you sample? If you want to use some of the data already coded for one of the available samples, by all means use that sample as your sampling frame, particularly if it is one of the larger lists (such as the summary version of the Ethnographic Atlas [Murdock 1967]). The sampling frame becomes the list you are claiming to generalize to. (A larger claim that you are generalizing to the world is inappropriate.) If you want to code all of your variables yourself, you can do so most economically by sampling from the HRAF collection. If a sample size of 60 is large enough, the HRAF Probability Sample has the advantage of giving you a randomly selected case from each of the 60 culture areas. If you want to code some variables yourself and use some precoded variables, you can sample from the intersection between HRAF and the sample with the precoded variables. Whatever sampling frame you use, you should select your cases in some standard random fashion, because only random sampling entitles you to infer that your statistically significant sample results are probably true for the larger universe. If, for some reason, you cannot sample randomly from some list, be sure to avoid selecting the cases yourself. After a random sample, the next best thing is a sample constructed by others (who could not know the hypotheses you want to test). The wonderful thing about a random sample is that wherever your research stops, after 20 or 40 or 200 randomly selected cases, you will always be entitled to conclude that a statistically significant result in your sample is probably true for the larger universe. MEASUREMENT

How you choose to measure some variable of interest to you depends at least partly on your implicit or explicit theory. After all, why are you interested in variable X in the first place? Your implicit or explicit theory specifies which variables are of interest and provides a model of how they may be related. Theories are generally evaluated by testing hypotheses derived from them. (What does the theory imply in the way of relationships?) The variables in a test can be fairly specific, such as whether or not a culture has a ceremony for naming a newborn child, or they may be quite abstract, such as whether the community is harmonious. Whether the concept is fairly specific or not, no variable is ever measured directly. We are so used to a thermometer measuring heat that we may forget that heat is an abstract concept that refers to the energy generated when molecules are moving. A thermometer reflects the principle that as molecules move more, a substance in a confined space (alcohol, mercury) will expand. We don’t see heat; we see only the movement of the substance in the confined space. So all measurement is indirect. But some measures are better (more direct, more predictive, fewer errors) than others. The three most important principles in designing a measure are: (1) try to be as specific as possible in deciding how to measure the theoretical variable you have in mind; (2) try to measure the variable as directly as possible; and (3) if possible, try to measure the variable in a number of different ways.

14_107-Bernard.indb 576

6/9/14 10:03 AM

18: Cross-Cultural Research

577

The first principle recognizes that science depends on replication; if we are to be confident about our findings, other researchers must be able to repeat them. To facilitate replication, the original researchers have to be quite explicit about what they intended to measure and exactly how they measured it. The second principle recognizes that although all measurement is indirect, some measures are more direct than others. To measure how “rainy” an area is, you could count the number of days that it rains while you are there for a week’s vacation and then multiply by 52. But it would be better to count the number of rainy days, on average, over a number of years, and it would be best if you measured the total number of inches of rain per years, on average, over a number of years. The third principle of good measurement is that few measures exactly measure what they are supposed to measure, so it is better to use more than one way to measure the theoretical variable of interest. However, much of cross-cultural research is limited by what is available in ethnographies. So, the availability of relevant information is the major constraint on the number of supposedly equivalent measures one might construct. Measures have to be specified for each variable in the hypothesis. Devising a measure involves at least four steps: (1) theoretically defining the variable of interest (in words or mathematically); (2) operationally defining the variable, which means spelling out the empirical information needed to make a decision about where the case falls on the scale that the researcher has devised for measuring it; (3) pretesting the measure to see if it can be applied generally (to many if not most cases) (designing a measure requires some trial and error, and if the scale is too confusing or too hard to apply, because the required information is too often lacking, the measure needs to be rethought); and (4) performing reliability and validity checks (reliability involves the consistency, replicability, and stability of a measure; validity involves the degree to which the measure reflects what it is supposed to reflect; for more discussion of these issues; see C. R. Ember et al. 1991). Because most attention has been paid to measurement issues in secondary comparisons, we focus mainly on them in what follows. To illustrate processes involved in measurement, let’s consider that a researcher has an idea about why many societies typically have extended families. Although the concept of extended families may appear straightforward, it needs to be defined explicitly. The researcher needs to decide whether to focus on extended family households or to include extended families that are not co-residential. The choice should depend on the theory. If the theory discusses labor requirements that would favor an extended family staying together (see Pasternak et al. 1997, 237–39), then extended family households should be measured. When this is decided, the researcher still needs to state what an “extended family” means and what a “household” means. And she or he has to decide on the degree (relative frequency) to which a sample society has extended family households. The first thing to decide is what is meant by an extended family. The researcher may choose to define a family as a social and economic unit consisting minimally of at least one or more parents and children; an extended family might be defined as consisting of two or more constituent families united by a blood tie; and an extended family household might be defined as an extended family living co-residentially—in one house,

14_107-Bernard.indb 577

6/9/14 10:03 AM

578

Carol R. Ember et al.

neighboring apartments, or a separate compound. Having defined the concepts, the researcher must then specify the counting procedure—how to measure the degree to which a society has extended family households. All of these steps are involved in operationalizing the variable of interest. Definitions are not so hard to arrive at. What requires work is evaluating whether an operational definition is useful or easily applied. For example, suppose by degree (of extended “family-ness”) we operationally mean the percentage of households in the community that are extended families. The range of possible scale scores is from 0 to 100%. Suppose further that we instruct our coders to rate a case only if the ethnographer specifies a percentage or we can calculate a percentage from a census of the households. If we did a pretest, we would find that very few ethnographers tell us the percentage of extended family households or the results of censuses. Rather, they usually say things like “extended family households are the norm.” Or, “extended families are typical, but younger people are beginning to live in independent households.” Our operational definition of percentage of extended family households, although perfectly worthy, may not be that useful if we cannot find enough societies with reports based on household censuses. What can we do? There are three choices. We can stick to our insistence on the best measure and study only societies for which a percentage is given (or can be calculated); we may have to expand our search (enlarge our sample) to find enough cases that have such precise information. Or we can redesign our measure to incorporate descriptions in words that are not based on census materials. Or we can choose not to do the study because we can’t measure the concept exactly how we want to. Faced with these choices, most cross-cultural researchers would opt to redesign the measure so as to incorporate word descriptions. Word descriptions do convey information about degree, even if not so precisely. If an ethnographer says “extended family households are typical,” we don’t know if that means 50% or 100%, but we can be very confident it does not mean 0–40%. And we can be fairly sure it does not mean 40–49%. If the relative frequency of extended families (measured on the basis of words) is related to something else, we should be able to see the relationship even though we are not able to use a percentage measure based on numerical information. A measure, going by words, might read something like what follows. Code extended family households as: 4. Very high in frequency if the ethnographer describes this type of household as the norm or typical in the absence of any indication of another common type of household. Phrases like “almost all households are extended” are clear indicators. Do not use discussions of the “ideal” household to measure relative frequency, unless there are indications that the ideal is also practiced. If there is a developmental cycle, such as the household splitting up when the third generation reaches a certain age, do not use this category. Rather, you should use scale score 3 if the extended family household remains together for a substantial portion of the life-cycle or scale score 2 if the household remains together only briefly. 3. Moderately high in frequency if the ethnographer describes another fairly frequent household pattern but indicates that extended family households are still the most common.

14_107-Bernard.indb 578

6/9/14 10:03 AM

18: Cross-Cultural Research

579

2. Moderately low in frequency if the ethnographer describes extended family households as alternative or a second choice (another form of household is said to be typical). 1. Infrequent or rare if another form of household is the only form of household mentioned and if the extended family form is mentioned as absent or an unusual situation. Do not infer the absence of extended families merely from the absence of discussion of family and household type. Don’t know if there is no information on form of household, or there is contradictory information.

The next step is to pretest this measure, preferably with coders who haven’t had anything to do with creating the scale. Four distinctions may be too difficult to apply to the word descriptions usually found in ethnographies, so a researcher might want to collapse the scale a little. Or, two coders may not agree with each other frequently. If so, the investigator may have to spell out the rules a little more. And if we decide to use the scale described above, what do we do when the ethnography actually gives us numbers or percentages for a case? It is usually easy to fit those numbers into the word scale (or to average two adjacent scale scores). For instance, if 70% of the households have extended families, and 30% are independent, we would choose scale score 3. But we might decide to use two scales: a precise one based on numerical measurement (percentages) for cases with numbers or percentages, the other scale relying on words (when the ethnography provides only words). C. R. Ember et al. (1991) recommend using both types of scale when possible. The advantage of using two scales of varying precision is that the more precise one (the quantitative scale) should be more strongly related to other variables than the less precise one. (The less-precise scale should be less accurate than the more precise one, assuming that the former sometimes has to rely on ambiguous words.) Stronger results with the more precise scale would increase our confidence that the relationship observed even with the less-precise one is true. We discuss the issue of validity in the next section and the issue of reliability in the section titled Minimizing Coder Error.

How to Minimize Error in the Design of Measures Two kinds of measurement errors are usually distinguished: systematic and random errors. They have different effects on data analysis and each type is handled differently (Zeller and Carmines 1980, 12). Systematic error or bias exists if there is a consistent, predictable departure from the “true” score. Examples would be a scale that inflates everyone’s weight by half a pound or a tendency by observers to ignore certain types of aggression in behavior observations. In cross-cultural research, systematic error can come from ethnographers, informants, design errors in measurement, or coders. Ethnographers may not mention something or may underreport it (e.g., not mention the number of Western objects in the village). Or they may overreport (e.g., overemphasize the unilineality of the descent system to fit reality into an ideal type). Informants may over- or underreport (e.g., not mention an illegal or disapproved activity). Coders may interpret ethnography from the point of view of their own culture, their gender, or their personality. As discussed below, we may be able to detect possible bias by hypothesizing and testing for it. However, one major

14_107-Bernard.indb 579

6/9/14 10:03 AM

580

Carol R. Ember et al.

type of systematic error can’t be detected so easily—the error that’s introduced because the measure consistently under- or overestimates the theoretical variable (Blalock 1968; Cook and Campbell 1979, 64; Zeller and Carmines 1980, 11). Random error, which is error in any direction, weakens correlations (Blalock 1972, 414). Naroll (1962) called random errors “benign,” perhaps because he and other social scientists commonly worry more about accepting a relationship as true that is false (Type I error) than about failing to accept a true relationship (Type II error). As Naroll (1962) pointed out, systematic error in two variables in the same direction could conceivably create a result when none is really there. (C. R. Ember has unpublished results showing that systematic error has to be enormous to produce a significant result when there really is none.) However, if our purpose is to find relationships when they are really there, we should take steps to minimize random error. But let us turn first to the problem of systematic error that is due to the lack of fit between the theoretical construct and the measure that presumes to tap that construct. Even though all measurement is indirect so validity can’t ever be established beyond a doubt (Campbell 1988), some measures are more direct and therefore more likely to be valid than others. More direct measures involve little question that they are measuring what they are supposed to measure. Other things being equal, we suggest that crossculturalists try to use measures that are as direct as possible, because less inference and less guesswork generally yield more accuracy and hence stronger results (assuming you are dealing with a true relationship). For example, when a cross-culturalist wants to measure whether a husband and wife work together, it is more direct to use a measure based on explicit ethnographers’ reports of work patterns rather than to infer the work pattern from general statements about how husbands and wives get along. Some measures pose few validity problems. Where the operationalization is very close to the theoretical variable, for example, the sex of a person (Blalock 1968, 20), the measure requires so little inference that there is little question about its validity. Other measures seem to have “face validity” too—few researchers would question their validity. For example, if the theoretical variable is the rule of residence in a society, we usually think that the ethnographer’s identification of the rule of residence is an operational measure with high face validity, despite the apparent lack of agreement between ethnographers Ward Goodenough (1956) and John Fischer (1958). The disagreement between them over Chuuk (Truk) appears to have been interpreted by some anthropologists as an indication that a fieldworker’s conclusions are bound to be subjective and therefore unreliable. We think that there are two reasons why this interpretation is incorrect. First, the two investigators were in the field at different times, and practices can vary even over just a few years. Second, both Goodenough and Fischer found the Chuukese to be predominantly matrilocal (Goodenough 71%, Fischer 58%). Indeed, they differed only with respect to 13% of the cases, which Goodenough classified as avunculocal and Fischer as patrilocal. (In some people’s eyes, avunculocal residence is patrilocal because living with husband’s mother’s brother is living with husband’s relatives.) With respect to the major or predominant residence pattern, the cross-culturalist would not misclassify the Chuukese using Fischer’s or Goodenough’s data, even though they apparently disagreed on some details.

14_107-Bernard.indb 580

6/9/14 10:03 AM

18: Cross-Cultural Research

581

Unlike experimental psychologists, cross-cultural researchers using secondary data are not able to use a great variety of validation techniques. There is rarely a standard measure to evaluate a new measure against. (If there were, we would always use the standard as the measure.) If there is little information in ethnographies on particular topics, it is sometimes difficult to think of multiple measures of the same theoretical variable. Given this situation, it is best to use measures that are the most direct and therefore have the highest face validity. However, many concepts of interest to cross-cultural researchers are not easily measured directly. Two situations in which indirect measures might justifiably be used are when the theoretical variable can be measured only “projectively” (e.g., unconscious fear of something as measured by how often, proportionately, folktales mention it) or when very few ethnographies give information allowing a more direct measure. Unconscious variables, in particular, do not lend themselves to direct measurement. Psychologists (and others) employ projective testing and disguised measures when subjects cannot or will not give honest answers to certain questions. For example, a field investigator (comparative or not) cannot simply ask a boy “Do you wish to be a woman?” and expect to get an answer that will reflect the boy’s degree of feminine identification, any more than a survey researcher can ask “Are you prejudiced against blacks?” and get an accurate response most of the time. Cross-cultural researchers may conclude, therefore, that culturally shared personality dispositions might be more accurately coded from folktales than from presumptions by ethnographers about unconscious feelings. Although a researcher may well be aware that a more direct measure would be preferable, more indirect measures may also be chosen because a more direct measure may be usable only for a small proportion of cases. Whatever the reason, the decision to employ an indirect or proxy measure should be made only if the investigator can justify its use on the basis of explicit and strong reasoning about why we should accept the validity of the proxy. More on this point below. When the cross-culturalist decides to use a more indirect measure, we strongly recommend that he or she develop direct measures for some proportion of the cases (even if only a minority) to evaluate the validity of the more indirect measure. If it’s not possible to correlate the two measures to validate the more indirect or proxy measure (because both kinds of information are hardly ever available for a particular case), the researcher could still see if the more indirectly measured cases show weaker correlations than the more directly measured cases. As noted above, if the more direct measure produces higher coefficients of association than the more indirect measure, and the directions of the results are similar with both measures, we can be more confident that the indirect measure is probably tapping the same variable as the more direct one. Researchers sometimes choose proxy measures over more direct measures because the proxy measures are readily available in precoded databases. But unless it is shown that they correlate with more direct measures (which requires going back to the original ethnographies), we’re skeptical about the validity of opportunistic proxy measures. We acknowledge that there’s enormous value in having databases with codes provided by previous researchers. It isn’t always necessary to code things anew, and using available codes when they fit your interests can allow time to code additional variables. But

14_107-Bernard.indb 581

6/9/14 10:03 AM

582

Carol R. Ember et al.

the practice of using precoded variables as proxies, without any attempts to validate them, deserves our suspicion. (Of course, the investigators originally responsible for the available codes are not to be blamed for how others use them.) The designer of a measure also needs to consider the degree to which the information required is available in ethnographies. No matter how direct a measure may seem conceptually, it may require a high degree of inference by the coder if there is little relevant information in ethnographies. The terms “high-inference” and “low-inference” variables, first introduced by J. W. M. Whiting (1981), are useful for discussing this aspect of measurement design. Variables that require low inference on the part of the coder tend to deal with visible traits or customs, usually reported by ethnographers, and are easily located in ethnographies (Bradley 1987, 1989; Burton and White 1987; J. W. M. Whiting 1981; and see White 1990). High-inference variables often require complex coding judgments and are therefore difficult to code reliably. The codings of low-inference variables are less likely to contain error, and independent coders are therefore more likely to agree on codings. For example, Bradley (1987) compared her codings of presence versus absence of the plow with Pryor’s (1985) codings for 23 societies; there was 96% agreement between the two data sets. The only disagreement was about a case that Pryor himself expressed uncertainty about. Thus, presence or absence of the plow, which ethnographers can observe and record without interpretation, is a low-inference variable. Others include the type of carrying device for infants, shapes of houses, domestic animals and major crops, and many elements of material culture (Bradley 1987; Burton and White 1987; J. W. M. Whiting 1981). Note that the dimension of low versus high inference may be uncorrelated with the dimension of more versus less direct measurement. Presence or absence of the plow may be measurable with low inference but if you consider it to be a proxy measure for the judgment that men do most of the agricultural work, then your measure would be low-inference but indirect. Because the measurement of some variables requires moderate levels of inference, such measures are more subject to random error. Relevant information is also missing more often, therefore you are likely to be able to code only a small proportion of the sample. In addition, when variables require moderate levels of inference, coders usually agree with each other less. Bradley (1989) presents evidence that coding the gender division of labor in agriculture requires moderate inference. Other examples of variables that require a moderate degree of inference are the proportion of day an infant is carried (J. W. M. Whiting 1981) and warfare frequency (C. R. Ember and M. Ember 1992a; Ross 1983). For moderate-inference variables, coders usually have to read through a considerable amount of ethnographic material that is not explicitly quantitative in order to rate a case. Because of the imprecision, the coding decision is more likely to contain some error. The highest degree of inference is required when researchers are interested in assessing general attitudes or global concepts such as “evaluation of children” (Barry et al. 1977). Such global concepts don’t have obvious empirical referents to guide coders, and it’s easy to see how coders focusing on different domains might justifiably code the same society differently. The most appropriate solution here, we believe, is to develop a series of more specific measures with clear empirical referents, as Whyte (1978) did

14_107-Bernard.indb 582

6/9/14 10:03 AM

18: Cross-Cultural Research

583

for the various meanings of women’s status and Ross (1983) did for the various dimensions of political decision making and conflict. Random errors may be more likely if the investigator does not precisely specify time and place for all variables. Divale (1975), acting on the Embers’ suggestion, has shown with a few examples how lack of time and place focus, which presumably increases random error, tends to lower correlations. The same time and place for a sample case should be attended to whether previous codes are used or new codes are developed. For example, M. Ember (1974, 1984/85) has presented evidence that polygyny is favored by a shortage of men because of high male mortality in war. If polygyny were present in a particular society as of 1900, but you measured the sex ratio in a later ethnography (after warfare ceased), you would be likely to find a more or less balanced sex ratio at the later time. Would this case be exceptional to the theory that high male mortality (and an excess of women) favors polygyny? The answer is yes, but it would not be appropriate to measure the two variables in this way. Each of the two variables should be measured synchronically (for more or less the same time period), or you could measure male mortality in war or the sex ratio for a slightly earlier time than you measure form of marriage. Otherwise, you may be introducing so much measurement error that the relationship between excess women and polygyny could be masked (C. R. Ember and M. Ember 2009, 76). (C. R. Ember et al.’s [1992] concordance between cross-cultural samples can help you match time and place foci across samples.) Requiring that your measurements on each case pertain to the same time and place (or the appropriate relative times for a diachronic [over time] test) can only maximize your chances of seeing a relationship that truly exists.

Minimizing the Effect of Ethnographer (or Informant) Error Measurement error caused by ethnographer or informant error is not often considered something we can deal with in the measurement process. But that is not necessarily true. Naroll (1962) proposed methods to deal with such errors, and others have developed additional methods. The supposedly poor or uneven quality of the ethnographic record is often cited as invalidating cross-cultural research. This notion is puzzling, given the usual high regard ethnographers have for their own work. If most anthropologists have high regard for their own work, how could the bulk of ethnography be poor unless most anthropologists were deluding themselves (C. R. Ember 1986, 2)? Certainly there are errors in the ethnographic record, and we must try to minimize their effects, but the critics’ worry may derive from their ignorance about the effect of error on results. Space here does not allow it, but we could show statistically that even a great deal of random error hardly ever produces a statistically significant finding. And even systematic error would not normally produce a statistically significant finding that was false. (There is always the possibility of deliberate cheating; but this probably does not happen often and other investigators’ attempts to replicate a result will eventually reveal it.) Statistically speaking, random error generally reduces the magnitude of obtained correlations. This means that more error lessens the likelihood of finding patterns that are there. And if error makes us less likely to infer statistical significance, it should not

14_107-Bernard.indb 583

6/9/14 10:03 AM

584

Carol R. Ember et al.

be assumed that significant cross-cultural results are generally invalid. It may seem paradoxical, but the more random error there is, the more likely the “true” results are better than the observed results. Naroll (1962) proposed an indirect method—data quality control—to deal with systematic informant and ethnographer errors. His basic procedure involved identifying factors that might produce biases in the reporting of certain variables (e.g., a short stay in the field would presumably make for underreporting secret practices such as witchcraft). Indeed, Naroll found that short-staying ethnographers were significantly less likely to report witchcraft than were long-staying ethnographers. This suggests the possibility of systematic error due to length of stay. However, as Naroll himself was aware, there are other possible interpretations of the correlation. One is that short stays may be more likely in more complex cultures. If more complex cultures are less likely to have witchcraft beliefs than less complex ones, the correlation between length of stay and the presence of witchcraft would be spurious, not due to systematic underreporting by short-staying ethnographers. Still, data quality factors could account for some results. For example, to exclude the possibility that a data quality factor may account for a correlation because that factor is correlated with both variables in the correlation, Naroll advised the researcher to control statistically for the data quality factor. Naroll’s concern with the possible influence of data quality persuaded some researchers to test for systematic biases of various kinds (gender of ethnographer, type of training, knowing the native language, etc.) in evaluating their results. Naroll believed that the coding of information on qualities of the ethnographer and on conditions of the fieldwork should be a regular part of the coding process. But C. R. Ember et al. (1991) do not agree, for two reasons. First, as Naroll (1977) also noted, it is very expensive for researchers to code for a large number of features that could, but probably do not, produce false correlations. Second, of the large number of studies done by Naroll, his students, and others (see Levinson [1978] for substantive studies employing data quality controls), hardly any have found that a data quality feature accounts for a correlation [but see also Divale 1976; Rohner et al. 1973]). Therefore, Ember et al. (1991) recommend that before investigating relationships between data quality variables and substantive variables, researchers should have plausible theoretical reasons for thinking they may be related. If we cannot imagine how a data quality variable could explain a correlation, it is not necessary to spend time and money coding for it. Only plausible alternative predictors should be built into a research design. However, Ember et al. (1991) recommend that researchers develop a specific data quality code for each substantive variable (for each case), to provide a direct assessment of the quality of the data in regard to that variable. To illustrate the suggested procedure, compare the information in the following three statements, which a coder might use to measure the frequency of polygyny among married men: “Polygyny is the form of marriage that men aspire to,” “Only the senior men have more than one wife,” and “The household survey indicates that 15% of the married men are married polygynously.” Although none of these statements may contain error, and we can’t assume that the one based on quantitative information is the most accurate, a coder trying to rate frequency of polygyny according to an ordinal

14_107-Bernard.indb 584

6/9/14 10:03 AM

18: Cross-Cultural Research

585

scale would have the most trouble coding the first statement because it tells us only that polygyny is present and preferred (at least by men). The researcher could try to minimize error by instructing the coders not to make frequency judgments based on statements about what people prefer (ideal culture). But the researcher could also develop a data quality code for the measure of frequency of polygyny. The highest-quality score would be given to a code based on a census. The lowest-quality score would be given to information pertaining to ideal culture (such as the first statement listed above) or to a judgment based on inference (e.g., polygyny is inferred to be not so common because only senior men are said to have more than one wife). A middle-quality score might be given to information such as in the following statement: “The typical married man has only one wife.” With a data quality code by case for each variable in a correlational test, we could analyze the results with and without the “poorer” quality data. The omission of cases with scores based on poor-quality data (e.g., vague ethnographic statements) should yield stronger results than the data set that includes poor-quality data. (For how cases with poorer data quality can be omitted, see our discussion toward the end of the next section on Minimizing Coder Error.) And because the standard errors will be higher with more random error, our chances of finding a statistically significant relationship are greater with higher-quality data. We think the data quality control procedure suggested here provides two important advantages. First, it taps the quality of the ethnographic information more directly than Naroll’s suggested strategy, which may not make any difference at all in regard to a particular correlation. Second, the data quality coding we suggest can be done quite efficiently, at the same time substantive variables are coded, because reading additional material is not required. Another problem the ethnographic record poses to cross-cultural researchers is that in addition to errors in what is reported, there may also be problems about what is not reported. Ethnographers may not have gone to the field with a comprehensive guide to what kinds of information could be collected, such as is provided by the Outline of Cultural Materials (Murdock et al. 2008). For this and other reasons, ethnographers often pay little or no attention to a question that is of interest later to the cross-cultural researcher. What should a cross-culturalist do? We don’t recommend inferring that something is absent if it is not reported, unless the cross-cultural researcher can be quite sure that it would have been reported if it had been present. For example, if an ethnographer didn’t mention puberty rites but thoroughly discussed childhood and adolescence, absence of puberty rites could reasonably be inferred. If, however, the ethnographer didn’t collect any information on adolescence, the fact that no puberty rites are mentioned shouldn’t be taken to mean that they were absent. Researchers need to specify coding rules for inferring absence (see the measure of extended family households described above) or they need to instruct their coders not to make inferences. A further strategy to deal with missing data is to interview the original ethnographers themselves (or others who have worked for extended periods in the society) to supplement the information not present in the published sources (Levinson 1989; Pryor 1977; Ross 1983). The cross-cultural researcher has to be careful to keep to the

14_107-Bernard.indb 585

6/9/14 10:03 AM

586

Carol R. Ember et al.

same time and place foci of the published data; if you call a recent ethnographer about a case, you should ask only about the time and place foci of the published information. Finally, if data on some of the sample societies are missing, the cross-cultural researcher may decide to impute missing values. Burton (1996) and Dow and Eff (2009) have described and evaluated a number of procedures for doing so. The cross-cultural researcher needs to remember that any method of imputation is likely to increase measurement error. Therefore, the advantage of imputation (to increase sample size) has to be weighed carefully against the possible increase of error. Researchers who impute some data should consider doing analyses with and without the imputed data to see if the imputing has misleadingly improved the results, just as we can compare data sets with and without dubious codings to see if including them has transformed a borderline or nonsignificant result into a significant one.

Minimizing Coder Error The coding process itself can produce measurement error. If the investigator is also the coder, there is the possibility of systematic bias in favor of the theory being tested (Rosenthal 1966; Rosenthal and Jacobson 1968). For that reason alone, many researchers prefer to use “naive” or theory-blind coders. However, naive coders may not be as likely as experienced coders to make accurate judgments. Experienced researchers have skills that should make for more accurate coding because they are more likely to be aware that an ethnographer’s words should not always be taken at face value. For example, an experienced researcher is more likely to know that avunculocal residence might be called patrilocal residence, and that hunter–gatherers may get plenty of food (even if they have to move their camps frequently, it doesn’t necessarily mean their food supply is precarious). Furthermore, experienced coders are more likely to pay attention to time and place foci—to know that when one ethnographer describes what Samoans do, he or she may not be talking about the particular time period or particular group of Samoans a previous ethnographer has studied and described (M. Ember 1985). Bradley (1987) argues that naive coders can make systematic errors when coding instructions are insufficiently precise, especially when coding high-inference variables. For example, differences between her codes for the division of labor in agriculture and those of the coders for Murdock and Provost (1973) could be explained by the possibility that the Murdock and Provost coders were not instructed to consider differences between crop types or which phase of the agricultural sequence was to be coded. Naive coders might also be more likely to make judgments that are systematically biased toward their own cultural assumptions, as suggested by the experimental findings presented by D’Andrade (1974). Researchers can try to minimize the error of inexperienced coders by being as explicit as possible in their coding instructions and by making sure that the codes do not surpass the information that is generally available in the ethnographic literature (Tatje 1970). Coding should have to make as few inferential leaps as possible. The process of trying to spell out all the possible obstacles to coding is an important part of the research design. It may be that having at least one relatively inexperienced coder provides an advantage—it may force the researcher to be as clear as possible in the operationalization of theoretical concepts.

14_107-Bernard.indb 586

6/9/14 10:03 AM

18: Cross-Cultural Research

587

Researchers may worry that coders will introduce systematic errors because of their gender, their political ideology, their personality, or their faulty assumptions about different types of societies. But such contamination may not be so likely. Whyte (1978) did not find more significant relationships between gender of coder and his many indicators of women’s status than would be expected by chance. His research suggests that systematic coding bias is most likely to occur when codes are very general (requiring a high degree of inference), which may allow the coder’s personal background to exert an influence on the coding process. Whyte suggests that personal/cultural biases can be avoided if coders are asked to rate concrete or specific customs and behaviors. Designing studies to test systematically for coder biases is normally quite expensive because to do so properly requires more than one coder for each bias-type. Thus, it’s more cost effective for the investigator and a naive coder to rate the cases. Not only could we then compare the two sets of ratings for reliability, we could also see if both sets of ratings give similar results. If they do not, that would be something to worry about. It might only be that the naive coder’s ratings contained more error; the results using only that coder’s ratings should be weaker than the results using only the investigator’s ratings. Perhaps the best strategy is to test hypotheses using only those cases that both raters agreed on. That way, you would probably be omitting the cases with more random error. This brings us to the concept of interrater reliability—the extent to which different persons using the same measure achieve the same score or the same relative ranking for each case rated (Nunnally 1967, 172; Zeller and Carmines 1980, 6). Researchers who assess interrater reliability usually show a correlation coefficient between the two raters’ judgments for at least a sample of the rated cases. Or, they may show the percentage of agreement. Both of these measures have advantages and disadvantages. The percentage of agreement may detect systematic error, whereas a correlation coefficient may not. Suppose one coder always gives a score that’s one point higher score than the other. The correlation coefficient will be perfect (1.00); the percentage of agreement will be zero. On the other hand, the percentage of agreement measure can’t distinguish between substantial disagreements and small disagreements. On a 10-point scale, a disagreement of 1 point counts the same as a disagreement of 10 points (Rohner and Katz 1970, 1068). But percentage of agreement will detect differences between coders due to systematic inflation or deflation by one of the coders, whereas this kind of systematic bias will not affect a reliability correlation. So, it is useful to compute both kinds of measures of interrater reliability. There is no clear decision point as to what is an acceptably high coefficient or percentage of agreement. Coefficients and percentages of agreement over .80 appear to be considered good and are usually reported without comment; more than .70 appears to be considered minimally acceptable; less than .70 leaves a feeling of unease. Neither method of measuring interrater reliability is a good way to deal with the situation where one rater does not assign a scale score (says “don’t know”) and the other rater does. Correlation coefficients are easily computed; if some form of Pearson’s r is used, it is easy to interpret—the square of the coefficient equals the proportion of variance explained. We always expect that there will be some interrater inconsistency (a study without any at all would probably be suspect). How do researchers deal with disagreements?

14_107-Bernard.indb 587

6/9/14 10:03 AM

588

Carol R. Ember et al.

If a researcher does a reliability check on a small portion of the cases and the reliability coefficient is reasonably high, she or he usually uses the scores of the one coder who rated all of the cases. If a researcher has two or more coders for each case, then a variety of strategies may be followed. Scores may be summed or averaged, or disagreements may be resolved by discussion. An advantage of the summing or averaging method is that both coders are given equal voice; personality differences between the coders cannot influence resolutions. A second advantage is that the “effective reliability” is increased. Rosenthal and Rosnow (1984, 163–65) indicate that if the correlation coefficient between two judges is .75, the reliability of the mean of the two judges’ ratings is actually higher (.86). This is a better estimate of the reliability of the measure; it is an example of the Spearman-Brown prophecy formula, also known as Cronbach’s alpha (Nunnally 1978, 211–16; Romney 1989). A disadvantage of the summing or averaging procedure is that one coder may be a better coder (more careful, more knowledgeable about the material), and the rating by that person might objectively deserve more weight. If coders are asked to discuss their disagreements and come to a resolution, the more knowledgeable coder might point to information that the other coder missed. This strategy might be particularly useful when each coder has read a large amount of ethnographic material. There are also some disadvantages in the resolution method. First, as mentioned, one coder may have undue influence over the other. Certainly, if one coder was the investigator and the other a paid assistant, the resolution method might not be unbiased. But even if the coders were roughly equal in status, one personality may dominate the other. Rohner and Rohner (1981) discuss a procedure for testing for the influence of one coder over another. However, a major problem with their method is that it cannot distinguish between influence because of power and influence because of more information. Another disadvantage is that it may add measurement error if the coders feel obliged to come to some resolution, even when the data are too ambiguous to justify resolution. The Embers have found that results get stronger when they use only those cases with the most reliable initial scores.4 This should not be surprising. The most reliable scores are presumably those that independent coders will agree on initially, before any attempts to resolve disagreements. When we eliminate cases with disagreement, our results should get stronger because we are probably eliminating the more ambiguous cases, which are probably the ones more likely to be coded inaccurately. An interrater reliability coefficient may be acceptable, but you could still have a lot of cases that have been measured inaccurately. Even though the size of the sample is reduced when we eliminate the least reliable scores in some way, our chances of finding a true relationship are improved. We think the best way to maximize the reliability of ratings and results is to use only those ratings (and cases) that the independent coders initially rated in exactly the same way (or very similarly), before any attempts at resolution. The researcher could provide a code that tells the reader how closely the raters agreed initially on the variable for a particular case (see C. R. Ember and M. Ember 1992b). This kind of reliability code (by variable, by case) would allow subsequent users of the data code to choose their own degree of reliability. Just as it is likely that results are more robust when we omit cases

14_107-Bernard.indb 588

6/9/14 10:03 AM

18: Cross-Cultural Research

589

that did not have higher quality data, so they should be more robust when we omit cases that were not rated in much the same way by the coders initially. Chances are, more ambiguous ethnographic information will generally be coded with more error, and it does us no good to cloud the situation by including them in our analyses.

Minimizing Error Due to Sampling: Galton’s Problem Another major reason some question cross-cultural findings is referred to as “Galton’s Problem.” In 1889, Francis Galton heard Edward Tylor’s presentation of what is generally considered the first cross-cultural study. Galton (see Tylor 1889, 270–72) suggested that many of Tylor’s cases were duplicates of one another because they had similar histories, so Tylor’s conclusions were suspect because the sample size was unjustifiably inflated. Since the 1960s, Raoul Naroll and others such as James Schaefer, Colin Loftin, Malcolm Dow, and E. Anthon Eff have considered Galton’s Problem a serious threat to cross-cultural research. They have devised several methods to test for the possible effects of diffusion and historical relatedness (Naroll 1970; for earlier references, see C. R. Ember 1990; more recent references can be found in Dow 2007, 2008; Dow and Eff 2008, 2009; Eff 2004). The concern behind these methods is that statistical associations may not be valid if the correlations could be attributed mostly to diffusion (cultural borrowing) or common ancestry. How serious is Galton’s Problem? Cross-culturalists disagree (see M. Ember and Otterbein 1991 for references; see also C. R. Ember 1990). Most but not all crossculturalists think Galton’s Problem is a serious one (see the names in the previous paragraph). We and others (Strauss and Orans, Otterbein—for references, see C. R. Ember 1990) think it is not serious, albeit for different reasons. A great deal of the disagreement hinges on different theoretical assumptions about causality—if you believe that cultures are slow to change and people are strongly influenced by their past history and by their neighbors, the more important you think the problem is. If you think that cultures change as circumstances change and borrow because something is perceived to be functional or adaptive, Galton’s Problem is not perceived as important. Statistical tests require two kinds of independence, which we can call sampling independence and measurement independence (Blalock 1972; Kish 1965, 1987). Sampling independence requires that every case has an equal chance to be chosen. Measurement independence requires that the measure for one or more variables for one case is not influenced by the measures of another case. Copying answers on an exam is a clear case of nonindependence. Blalock (1972, 144–45) suggests that crime rates for different census tracts may be nonindependent if crimes are committed by some of the same individuals across boundaries. It is measurement independence that Galton appeared most concerned with; the assumption is that common history created similarity. However, some who argue against Galton’s Problem (M. Ember and Otterbein 1991, 223–24) assert that usually different societies have mutually unintelligible languages and therefore their speech communities have been separated for at least 1,000 years. If two related languages began to diverge 1,000 or more years ago, many other aspects of the cultures will also have diverged. Therefore, such cases could hardly be duplicates of each other. If you push Galton’s Problem to the limit and avoid any two cases that share a common

14_107-Bernard.indb 589

6/9/14 10:03 AM

590

Carol R. Ember et al.

history and language, then psychological studies with more than one individual per culture would be suspect! Until recently, whether or not you worried about Galton’s Problem made a big difference about how you would do a study. Naroll’s tests for the possibility of diffusion were quite time consuming to carry out; probably for this reason, most cross-culturalists altered their sampling strategy to eliminate multiple cases from the same culture area. For example, the Standard Cross-Cultural Sample (Murdock and White 1969) and the HRAF Probability Sample both contain only one culture per identified culture area. Unfortunately, this solution is not ideal because sampling from geographical clusters does not give all societies an equal chance to be chosen (a principle of statistical independence). Another sampling strategy is to take a relatively small random sample from a large list, such as the Ethnographic Atlas (Murdock 1962ff.) or the summary Ethnographic Atlas (Murdock 1967). By chance, there are likely to be only a few closely related cases; if there are, some can be eliminated randomly (C. R. Ember and M. Ember 2009, 109; M. Ember and Otterbein 1991). Recently, however, mathematical anthropologists have developed statistical solutions and computer programs that treat the proximity of societies (in distance or language) as a variable whose influence can be tested in a multiple regression analysis. (This is called “testing for spatial autocorrelation.” For newer treatments based on network autocorrelation, see Dow 2007, 2008; Eff 2008; Eff and Dow 2009). Whether or not a researcher agrees that Galton’s Problem is a problem, the recent mathematical and computer solutions do not require a special sampling strategy, nor do they require expensive time-consuming controls. If you worry about Galton’s Problem, all you have to do is test statistically for the possibility that proximity or common ancestry accounts for a result (Burton et al. 1996; Dow et al. 1984). Even without a statistical control for autocorrelation, cross-culturalists who randomly sample from a larger sampling frame can redo their analyses by randomly omitting more than a single case from the same culture area. If the results do not change substantially after multiple cases from an area are omitted, the original result cannot be due to duplication of cases. Indeed, duplication may weaken results because a set of historically related cases may be exceptional to a cross-cultural generalization rather than consistent with it. (For some evidence on this possibility, see M. Ember 1971.) MAXIMIZING THE INFORMATION VALUE OF STATISTICAL ANALYSIS

Most cross-cultural studies aim to test hypotheses, so some kind of inferential statistic is usually used to make a decision about whether the tested hypothesis should be accepted or rejected. This is where statistical researchers resort to the concept of level of significance. Cross-culturalists generally conform to the social science convention of accepting a hypothesis (at least provisionally) if there were five or fewer chances out of a hundred (p < .05) of getting the result just by chance. That is, if 100 different random samples were examined, only 5 or fewer would show the same result or one stronger. Early cross-cultural studies usually relied on contingency tables and chi-square as a test of significance for the relationship between two variables (bivariate analyses), but with the advent of statistical software packages for personal computers, increasing numbers of cross-culturalists realized that they could achieve more powerful and

14_107-Bernard.indb 590

6/9/14 10:03 AM

18: Cross-Cultural Research

591

more informative statistical results in testing hypotheses if they measure ordinally (or intervally) rather than nominally. In addition, multivariate analyses, such as logistic regression (for nominal dependent variables) and multiple regression (for interval and sometimes ordinal dependent variables), can be used to evaluate the independent effects of two or more variables.5 Nominal measurement is putting a case into an appropriate set without implying that one set is higher or lower than another on some scale (e.g., female vs. male; extended family households vs. independent family households). Often, nominal variables can be appropriately transformed to ordinal variables. Rather than classify societies as just having independent family households versus extended family households, our discussion earlier showed how to measure the degree to which a society has extended family households in terms of a four-point ordinal scale. In ordinal-scale measurement, a higher (or lower) number implies more (or less) of the variable measured. Cross-cultural researchers can rarely measure variables in terms of interval or ratio scales, where the distance between two adjacent numbers is equal to the distance between any other two adjacent numbers (the ratio-scale has a meaningful zero-point). But there are plenty of possible interval or ratio scales: number of people in the community, population density, average rainfall, average mean temperature (the Fahrenheit or Celsius scale is an interval scale because zero does not mean the absence of temperature), altitude, and others. Labovitz (1967, 1970) has suggested that statistical tests originally designed for interval-level data may be used with ordinal data when the number of ordered scale scores is not very small. For example, C. R. Ember and M. Ember (1992a, 1994) used multiple regression analysis when the dependent variables had five or more ordinal scale scores. Indeed, cross-cultural researchers are now more likely to use multivariate techniques to discover the relative effects of two or more predictors. These techniques are especially important for evaluating alternative theories that seem to be equally supported by the bivariate results. Most cross-culturalists don’t use mathematical formulas to decide on sample size in advance of their research. This is a pity, for a researcher often spends a good deal of effort trying to code a large number of cases in the belief that a large sample size is necessary. Generally, you don’t need a large sample to obtain a significant result. A small one can give you a trustworthy or significant result if it is strong. A large sample is necessary if you want to detect a weak association or effect. Indeed, with a large enough sample, you will often find trivial effects that may hinder interpretation of results. Nominallevel tests (like chi-square) require the largest sample sizes; ordinal and interval-level tests generally require smaller sample sizes. Kraemer and Tiemann (1987) have provided a master table to calculate approximately what sample size you will need, if you can specify how big an effect or correlation you are looking for, and how much possibility of error you can tolerate. (They also provide specific formulas for particular measures and tests of significance.) Suppose a researcher is looking for a correlation of .50 or better and wants to be 90% confident that he or she has found a true result, and one that has only a .05 chance of being false. The approximate number of cases required would be 30. If the researcher insists on a p value of .01, 45 cases would be required. To detect a weak correlation of only .24, a p value of .05 requires 144 cases; a p value of .01 requires 219 cases (Kraemer and Thiemann 1987).

14_107-Bernard.indb 591

6/9/14 10:03 AM

592

Carol R. Ember et al.

Statistical techniques (factor analysis, multidimensional scaling, correspondence analysis) can evaluate whether there are one or more dimensions in a set of related measures. These techniques are especially useful for constructing complex scales from multiple indicators (e.g., indicators of social complexity, status of women) and for exploring patterns in the data. Most of these techniques are not used for hypothesis testing. While cross-culturalists aim to reject causal theories when the hypotheses derived from them are not supported by correlational tests, they generally cannot differentiate between causes and effects even when hypotheses are supported. However, much more can be done than usually is. First, partial correlation and path analysis are designed to help evaluate alternative causal models. Only a few cross-cultural studies have used such techniques (C. R. Ember and M. Ember 1992a, 1994; Kitahara 1981). Second, more cultures than you might think have been studied for more than one time period.6 Crossculturalists can use this diachronic information to study whether changes occur in time as predicted by our causal theories, but there’s been little research of this kind as yet. (More will probably be done as the HRAF Collection of Ethnography increasingly includes information on more than one time period in the history of a case.) In the future, crossarchaeological studies will also be used to test theories diachronically (M. Ember and C. R. Ember 1995) as eHRAF Archaeology grows and contains more tradition sequences. Causal theories can also be tested using “phylogenetic” methods (Borgerhoff Mulder et al. 2001; Mace and Holden 2005). Phylogenetically based comparative methods employ an existing phylogeny that describes a hypothetical relationship between cultures based on some known variable, most commonly language. The known variable is used to create a phylogenetic tree upon which specific cultural traits are mapped. The tree of mapped traits is then examined or compared to other mapped traits to determine whether or not a specific hypothesis is supported. Because the trees are based on known relationships, they can be used to infer the evolution of various traits found among a group of cultures; that is, cultures grouped in a phylogeny based on language that also share a particular cultural trait might be hypothesized to have inherited that trait from the common ancestral culture that spoke the ancestral language (e.g., see the papers in Mace et al. 2005). Since the phylogenetic tree also hypothesizes ancestral states, phylogenetic methods can also be used as a method of reconstructing ancient cultural systems (e.g., Currie et al. 2010; Jordan 2011). Phylogenetically based comparative methods are not without controversy (Moore 1994), and their use to date has been limited both in geographic scope and scholarly impact. However, they have the potential to allow cross-culturalists to explore complex evolutionary and causal relationships. They also are unique among cross-cultural methods in avoiding Galton’s Problem altogether, as defining the relationships between cases (i.e., the extent to which they are independent) is a foundation of the method (Borgerhoff Mulder et al. 2001). Because of these strengths, we anticipate phylogenetically based comparative methods will be more widely used in the coming years. CONCLUSION

Cross-cultural researchers do not deny the uniqueness of particular cultures, but they look at cultures in a different way—focusing on qualities or quantities that vary along some specified dimension. These variables do not capture everything about cultural

14_107-Bernard.indb 592

6/9/14 10:03 AM

18: Cross-Cultural Research

593

attitudes, beliefs, values, or behaviors, but they do exhibit some distinguishable similarities and differences. A large part of the art of cross-cultural research is learning how to focus on dimensions of variation; so is learning how to frame a meaningful and answerable question. Questions range from the descriptive, dealing with the prevalence or frequency of a trait, to questions about causes, consequences, or relationships without specified causal direction. In addition to believing that comparison is possible, cross-culturalists generally assume that all generalizations require testing on some unbiased sample of cases. There are many types of cross-cultural comparison, from regional to worldwide, small or large, using primary or secondary data, synchronic or diachronic. They all have advantages and disadvantages. Large samples aren’t necessarily better than small ones—what is most important is that the studied cases should fairly represent the universe of cases to which the results are generalizable; some kind of random sampling is the best way to ensure representativeness. Because theoretical variables are never measured directly, we have devoted a lot of space to various issues of measurement, including how to minimize error in designing measures and how to minimize the effects of ethnographer and coder error in cross-cultural tests using ethnographic data. We advocate the design of the most direct measures possible, those that require the least coder inference, and data quality scores for each variable for each culture. Cross-cultural researchers don’t unanimously agree about everything they do. They disagree about the seriousness of Galton’s Problem, and how to deal with it, for instance. And they disagree often about causal interpretations. But with all their disagreement, they agree that it is necessary to do cross-cultural research (preferably a variety of cross-cultural studies—worldwide, regional, primary, secondary, synchronic, diachronic) to arrive at trustworthy and comprehensive explanations of human behavior, explanations that are probably true because they apply to the vast majority of cultures. We need to test our explanations in other ways too—within cultures, ethnohistorically, cross-historically, experimentally, by computer simulations. But cross-cultural research (of the usual synchronic variety) is a necessary part of the social scientific enterprise because only a cross-cultural test gives us a relatively low-cost opportunity to discover that a theory or explanation does not fit the real world of cultural variation and therefore should be rejected, at least for now. NOTES

1. Although it might be argued that the Standard Cross-Cultural Sample does not adequately cover contemporary national cultures, it does contain some cultures that constitute the core of countries today—Japanese, Koreans, Russians, Egyptians, Vietnamese, Chinese, and Haitians. Considering that the number of well-described cultures probably exceeds 2,000, we would not expect that many to be national cultures. 2. Swanson (1980) reported that about one-third of the probability sample files cases were well described for more than one time period; since then, more recent time frames have been added. Researchers can use C. R. Ember et al.’s (1992) cross-cultural concordance that matches time and place foci across samples to find cultures described for more than one time period. 3. However, Bondarenko et al. (2005) added cultures of Russia. 4. This was the case in the results reported in C. R. Ember and M. Ember 1992a. Although the initial results that included all resolved ratings were not shown, they were considerably

14_107-Bernard.indb 593

6/9/14 10:03 AM

594

Carol R. Ember et al.

weaker with all resolved ratings. The reported results were based on omitting the less reliably rated variables. 5. Autocorrelation methods referred to Galton’s Problem is a form of multiple regression. 6. See note 2. REFERENCES

Barry, H., III, L. Josephson, E. Lauer, and C. Marshall. 1977. Agents and techniques for child training; cross-cultural codes 6. Ethnology 16: 191–230. Barry, H., III, and A. Schlegel, eds. 1980. Cross-cultural samples and codes. Pittsburgh: University of Pittsburgh Press. Blalock, H. M., Jr. 1968. Measurement problem: A gap between the languages of theory and research. In Methodology in social research, ed. H. M. Blalock and A. B. Blalock, 5–27. New York: McGraw Hill. Blalock, H. M., Jr. 1972. Social statistics, 2nd ed. New York: McGraw-Hill. Bondarenko, D., A. Kazankov, D. Khaltourina, and A. Korotayev. 2005. A corrected ethnographic atlas. Ethnology 44: 261–89. Borgerhoff Mulder, M., M. George-Cramer, J. Eshleman, and A. Ortolani. 2001. A study of East African kinship and marriage using a phylogenetically based comparative method. American Anthropologist 103: 1059–82. Bradley, C. 1987. Women, children and work. Ph.D. dissertation. Irvine: University of California Press. Bradley, C. 1989. Reliability and inference in the cross-cultural coding process. Journal of Quantitative Anthropology 1: 353–71. Burton, M. L. 1996. Constructing a scale of female contributions to agriculture: Methods for imputing missing data. Cross-Cultural Research 30: 3–23. Burton, M. L., C. C. Moore, J. W. M. Whiting, and A. K. Romney. 1996. Regions based on social structure. Current Anthropology 37: 87–123. Burton, M. L., and D. R. White. 1987. Cross-cultural surveys today. Annual Reviews of Anthropology 16: 143–60. Burton, M. L., and D. R. White. 1991. Regional comparisons, replications, and historical network analysis. Behavior Science Research 25: 55–78 (Special issue: Cross-Cultural and Comparative Research: Theory and Method). Campbell, D. T. 1961. The mutual methodological relevance of anthropology and psychology. In Psychological anthropology: Approaches to culture and personality, ed. F. L. K. Hsu, 333–52. Homewood, IL: Dorsey Press. Campbell, D. T. 1988. Methodology and epistemology for the social sciences: Selected papers. Chicago: University of Chicago Press. Cochran, W. G. 1977. Sampling techniques, 3rd ed. New York: John Wiley. Cook, T. D., and D. T. Campbell. 1979. Quasi-experimentation: Design and analysis issues for field settings. Chicago: Rand McNally. Currie, T. E., S. J. Greenhill, R. D. Gray, T. Hasegawa, and R. Mace. 2010. Rise and fall of political complexity in island South-East Asia and the Pacific. Nature 467: 801–4. D’Andrade, R. 1974. Memory and the assessment of behavior. In Measurement in the social sciences, ed. T. Blalock, 149–86. Chicago: Aldine-Atherton. Dickson, H. R. P. 1951. The Arab of the desert: A glimpse into Badawin life in Kuwait and Sau’di Arabia. London: George Allen & Unwin. Divale, W. T. 1974. Migration, external warfare, and matrilocal residence. Behavior Science Research 9: 75–133.

14_107-Bernard.indb 594

6/9/14 10:03 AM

18: Cross-Cultural Research

595

Divale, W. T. 1975. Temporal focus and random error in cross-cultural hypothesis tests. Behavior Science Research 10: 19–36. Divale, W. T. 1976. Female status and cultural evolution: A study in ethnographer bias. Behavior Science Research 10: 19–36. Divale, W. T. 2007. Galton’s problem as multiple network autocorrelation effects: Cultural trait transmission and ecological constraint. Cross-Cultural Research 41: 336–63. Divale, W. T. 2008. Network autocorrelation regression with binary and ordinal dependent variables. Cross-Cultural Research 42: 394–419. Dow, M. M., M. L. Burton, D. White, and K. Reitz. 1984. Galton’s problem as network autocorrelation. American Ethnologist 11: 754–70. Dow, M. M., and E. A. Eff. 2008. Global, regional, and local network autocorrelation in the standard cross-cultural sample. Cross-Cultural Research 42: 148–71. Dow, M. M., and E. A. Eff. 2009. Cultural trait transmission and missing data as sources of bias in comparative survey research: Explanations of polygyny re-examined. Cross-Cultural Research 43: 134–51. Driver, H., and W. C. Massey. 1957. Comparative studies of North American Indians. Transactions of the American Philosophical Society 47: 165–456. Eff, E. A. 2004. Does Mr. Galton still have a problem? Autocorrelation in the standard crosscultural sample. World Cultures 15: 153–70. Eff, E. A. 2008. Weight matrices for cultural proximity: Deriving weights from a language phylogeny. Structure and Dynamics: eJournal of Anthropological and Related Sciences 3(2). http:// escholarship.org/uc/item/13v3x5xw (accessed September 15, 2012). Eff, E. A., and M. M. Dow. 2009. How to deal with missing data and Galton’s problem in crosscultural survey research: A primer for R. Structure and Dynamics: eJournal of Anthropological and Related Sciences 3(2). http://www.escholarship.org/uc/item/7cm1f10b (accessed September 15, 2012). Eggan, F. 1954. Social anthropology and the method of controlled comparison. American Anthropologist 56: 655–63. Ember, C. R. 1975. Residential variation among hunter-gatherers. Behavior Science Research 10: 199–227. Ember, C. R. 1978a. Men’s fear of sex with women: A cross-cultural study. Sex Roles: A Journal of Research 4: 657–78. Ember, C. R. 1978b. Myths about hunter-gatherers. Ethnology 17: 439–48. Ember, C. R. 1986. The quality and quantity of data for cross-cultural studies. Behavior Science Research 20: 1–16. Ember, C. R. 1990. Bibliography of cross-cultural methods. Behavior Science Research 24: 141–54. Ember, C. R., and M. Ember. 1992a. Resource unpredictability, mistrust, and war: A crosscultural study. Journal of Conflict Resolution 36: 242–62. Ember, C. R., and M. Ember. 1992b. Warfare, aggression, and resource problems: Cross-cultural codes. Behavior Science Research 26: 169–226. Ember, C. R., and M. Ember. 1994. War, socialization, and interpersonal violence: A crosscultural study. Journal of Conflict Resolution 38: 620–46. Ember, C. R., and M. Ember. 2009. Cross-cultural research methods, 2nd ed. Lanham: AltaMira. Ember, C. R., and D. Levinson. 1991. The substantive contributions of worldwide cross-cultural studies using secondary data. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 79–140.

14_107-Bernard.indb 595

6/9/14 10:03 AM

596

Carol R. Ember et al.

Ember, C. R, with the assistance of H. Page, Jr., T. O’Leary, and M. M. Martin. 1992. Computerized concordance of cross-cultural samples. New Haven, CT: Human Relations Area Files. Printed on demand. Ember, C. R., M. H. Ross, M. L. Burton, and C. Bradley. 1991. Problems of measurement in cross-cultural research using secondary data. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 187–216. Ember, M. 1971. An empirical test of Galton’s problem. Ethnology 10: 98–106. Ember, M. 1974. Warfare, sex ratio, and polygyny. Ethnology 13: 197–206. Ember, M. 1984/85. Alternative predictors of polygyny. Behavior Science Research 19: 1–23. Ember, M. 1985. Evidence and science in ethnography: Reflections on the Freeman-Mead controversy. American Anthropologist 87: 906–10. Ember, M. 1991. The logic of comparative research. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 143–53. Ember, M. 1997. Evolution of the Human Relations Area Files. Cross-Cultural Research 31: 3–15. Ember, M., and C. R. Ember. 1971. The conditions favoring matrilocal residence versus patrilocal residence. American Anthropologist 73: 571–94. Ember, M., and C. R. Ember. 1995. Worldwide cross-cultural studies and their relevance for archaeology. Journal of Archaeological Research 3: 87–111. Ember, M., C. R. Ember, and B. Russett. 1997. Inequality and democracy in the anthropological record. In Inequality, democracy, and economic development, ed. M. Midlarsky, 110–32. Cambridge: Cambridge University Press. Ember, M., and K. F. Otterbein. 1991. Sampling in cross-cultural research. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 217–35. Ethnographic Atlas. 1962–. Ethnology 1:113ff. and intermittently thereafter. Fischer, J. L. 1958. The classification of residence in censuses. American Anthropologist 60: 508–17. Goldschmidt, W. 1965. Theory and strategy in the study of cultural adaptability. American Anthropologist 67: 402–8. Goodenough, W. 1956. Residence rules. Southwestern Journal of Anthropology 12: 22–37. Gray, J. P. 1996. Is the Standard Cross-Cultural Sample biased? A simulation study. CrossCultural Research 30: 301–15. HRAF. 1967. The HRAF quality control sample universe. Behavior Science Notes 2: 63–69. Johnson, A. 1991. Regional comparative field research. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 3–22. Jordan, F. 2011. A phylogenetic analysis of the evolution of Austronesian sibling terminologies Human Biology 83: 297–321. Jorgensen, J. G. 1974. On continuous area and worldwide sample cross-cultural studies. In Comparative studies by Harold E. Driver and essays in his honor, ed. J. G. Jorgensen, 195–204. New Haven, CT: HRAF Press. Kish, L. 1965. Survey sampling. New York: John Wiley. Kish, L. 1987. Statistical design for research. New York: Wiley. Kitahara, M. 1981. Men’s heterosexual fear due to reciprocal inhibition. Ethos 9: 37–50. Kraemer, H. C., and S. Theimann. 1987. How many subjects? Statistical power analysis in research. Newbury Park, CA: Sage. Labovitz, S. 1967. Some observations on measurement and statistics. Social Forces 46: 151–60. Labovitz, S. 1970. The assignment of numbers to rank order categories. American Sociological Review 35: 515–24.

14_107-Bernard.indb 596

6/9/14 10:03 AM

18: Cross-Cultural Research

597

Lagacé, R. O. 1979. The HRAF probability sample: Retrospect and prospect. Behavior Science Research 14: 211–29. Levinson, D. 1978. Holocultural studies based on the Human Relations Area Files. Behavior Science Research 13: 295–302. Levinson, D. 1989. Family violence in cross-cultural perspective. Newbury Park, CA: Sage. Longacre, W. 1970. Archaeology as anthropology: A case study. Anthropological papers of the University of Arizona, Number 17. Tucson: University of Arizona Press. Mace, R., and C. Holden. 2005. A phylogenetic approach to cultural evolution. TRENDS in Ecology and Evolution 20: 116–21. Mace, R., C. Holden, and S. Shennan, eds. 2005. The evolution of cultural diversity: A phylogenetic approach. London: UCL Press. Meggitt, M. J. 1964. Male-female relationships in the highlands of Australian New Guinea. American Anthropologist 66: 202–24. Moore, J. H. 1994. Putting anthropology back together again: The ethnogenetic critique of cladistic theory. American Anthropologist 96: 925–48. Munroe, R. H., H. S. Shimmin, and R. L. Munroe. 1984. Gender understanding and sex role preference in four cultures. Developmental Psychology 20: 673–82. Munroe, R. L., and R. H. Munroe. 1991a. Comparative field studies: Methodological issues and future possibilities. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 155–85. Munroe, R. L., and R. H. Munroe. 1991b. Results of comparative field studies. (Special issue. Cross-cultural and comparative research: Theory and method.) Behavior Science Research 25: 23–54. Munroe, R. L., and R. H. Munroe. 1992. Fathers in children’s environments: A four culture study. In Father-child relations, ed. B. S. Hewlett, 213–29. New York: Aldine de Gruyter. Murdock, G. P. 1949. Social structure. New York: Macmillan. Murdock, G. P. 1957. World ethnographic sample. American Anthropologist 59: 664–87. Murdock, G. P. 1962ff. Ethnographic atlas. Ethnology 1 ff. Murdock, G. P. 1967. Ethnographic atlas: A summary. Ethnology 6: 109–236. Murdock, G. P. 1981. Atlas of world cultures. Pittsburgh: University of Pittsburgh Press. Murdock, G. P., C. S. Ford, A. E. Hudson, R. Kennedy, L. W. Simmons, and J. W. M. Whiting. 2008. Outline of cultural materials, 6th rev. ed. with modifications. New Haven, CT: HRAF Press. Murdock, G. P., and C. Provost. 1973. Factors in the division of labor by sex: A cross-cultural analysis. Ethnology 12: 203–25. Murdock, G. P., and D. R. White. 1969. Standard Cross-Cultural Sample. Ethnology 8: 329–69. Nadel, S. F. 1954. Nupe religion. London: Routledge & Kegan Paul. Naroll, R. 1962. Data quality control—a new research technique: Prolegomena to a cross-cultural study of culture stress. New York: Free Press. Naroll, R. 1967. The proposed HRAF probability sample. Behavior Science Notes 2: 70–80. Naroll, R. 1970. Galton’s problem. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 974–89. Garden City, NY: Natural History Press. Naroll, R. 1977. Cost-effective research versus safer research. Behavior Science Research 11: 123–48. Naroll, R., V. L. Bullough, and F. Naroll. 1974. Military deterrence in history: A pilot crosshistorical pilot survey. Albany: State University of New York Press. Naroll, R., and R. G. Sipes. 1973. Standard ethnographic sample. Current Anthropology 14: 111–40. Naroll, R., and H. Zucker. 1974. Reply. Current Anthropology 15: 316–17.

14_107-Bernard.indb 597

6/9/14 10:03 AM

598

Carol R. Ember et al.

Nunnally, J. C. 1967. Psychometric theory. New York: McGraw-Hill. Nunnally, J. C. 1978. Psychometric theory, 2nd ed. New York: McGraw-Hill. Pasternak, B., C. R. Ember, and M. Ember. 1997. Sex, gender, and kinship: A cross-cultural perspective. Upper Saddle River, NJ: Prentice Hall. Peregrine, P. 2001. Cross-cultural comparative approaches in archaeology. Annual Review of Anthropology 30: 1–18. Peregrine, P. 2003. Atlas of cultural evolution. World Cultures 14: 2–88. Peregrine, P. 2004. Cross-cultural approaches in archaeology: Comparative ethnology, comparative archaeology, and archaeoethnology. Journal of Archaeological Research 12: 281–309. Peregrine, P. 2006. Synchrony in the New World: An example of ethnoarchaeology. CrossCultural Research 40: 6–17. Peregrine, P. 2007. Modeling state origins using cross-cultural data. Cross-Cultural Research 41: 1–12. Peregrine, P., C. R. Ember, and M. Ember. 2004. Universal patterns in cultural evolution: An empirical analysis using Guttman scaling, American Anthropologist 106: 145–49. Peregrine, P., and M. Ember, eds. 2001–2002. Encyclopedia of prehistory. New York: Kluwer Academic/Plenum Publishers. Pryor, F. L. 1977. The origins of the economy: A comparative study of distribution in primitive and peasant economies. New York: Academic Press. Pryor, F. L. 1985. The invention of the plow. Comparative Studies in Society and History 27: 727–43. Rohner, R. P., B. R. DeWalt, and R. C. Ness. 1973. Ethnographer bias in cross-cultural research: An empirical study. Behavior Science Notes 8: 275–317. Rohner, R. P., and L. Katz. 1970. Testing for validity and reliability in cross-cultural research. American Anthropologist 72: 1068–73. Rohner, R. P., and E. C. Rohner. 1981. Assessing interrater influence in cross-cultural research: A methodological note. Behavior Science Research 16: 341–51. Romney, A. K. 1989. Quantitative models, science and cumulative knowledge. Journal of Quantitative Anthropology 1: 153–223. Rosenthal, R. 1966. Experimenter effects in behavioral research, enl. ed. New York: Irvington. Rosenthal, R., and L. Jacobson. 1968. Pygmalion in the classroom. New York: Holt, Rinehart, and Winston. Rosenthal, R., and R. L. Rosnow. 1984. Essentials of behavioral research: Methods and data analysis. New York: McGraw-Hill. Ross, M. H. 1983. Political decision making and conflict: Additional cross-cultural codes and scales. Ethnology 22: 169–92. Seligman, C. G., and B. Z. Seligman. 1932. Pagan tribes of the Nilotic Sudan. London: George Routledge & Sons. Smith, M. E., ed. 2012. The comparative archaeology of complex societies. Cambridge: Cambridge University Press. Special Issue. Cross-Cultural and Comparative Research: Theory and Method. 1991. Behavior Science Research (now Cross-Cultural Research), 25 (1–4). Stout, D. B. 1947. San Blas Cuna acculturation: An introduction. New York: Viking Fund Publications in Anthropology. Swanson, E. C. 1980. A note about temporal foci in the HRAF Probability Sample Files. CrossCultural Research 15: 295–307. Tatje, T. A. 1970. Problems of concept definition for comparative studies. In A handbook of method in cultural anthropology, ed. R. Naroll and R. Cohen, 689–96. Garden City, NY: Natural History Press.

14_107-Bernard.indb 598

6/9/14 10:03 AM

18: Cross-Cultural Research

599

Tylor, E. B. 1889. On a method of investigating the development of institutions applied to the laws of marriage and descent. Journal of the Royal Anthropological Institute of Great Britain and Ireland 18: 245–72. White, D. R. 1990. Reliability in comparative and ethnographic observations: The example of high inference father-child interaction measures. Journal of Quantitative Anthropology 2: 109–50. Whiting, B. B., ed. 1963. Six cultures: Studies of child rearing. New York: Wiley. Whiting, J. W. M. 1954. The cross-cultural method. In Handbook of social psychology, Vol. 1, ed. G. Lindzey and E. Aronson, 523–31. Cambridge, MA: Addison-Wesley. Whiting, J. W. M. 1981. Environmental constraints on infant care practices. In Handbook of cross-cultural human development, ed. R. H. Munroe, R. L. Munroe, and B. B. Whiting, 155–79. New York: Garland. Whyte, M. K. 1978. Cross-cultural studies of women and the male bias problem. Behavior Science Research 13: 65–80. Witkowski, S. R. N.d. Environomental familiarity and models of band organization. Unpublished manuscript. New Haven, CT: Human Relations Area Files. Zeller, R. A., and E. G. Carmines. 1980. Measurement in the social sciences: The link between theory and data. New York: Cambridge University Press.

14_107-Bernard.indb 599

6/9/14 10:03 AM

14_107-Bernard.indb 600

6/9/14 10:03 AM

CHAPTER 19

Geospatial Analysis Eduardo S. Brondizio and Tracy Van Holt

Implicitly and explicitly, space has always had its place in anthropological research. Anthropologists talk about space in many ways: It provides context in ethnographic research on the distribution of populations, social groups and territories, environmental parameters, natural resources, and human behavior on the landscape. Space-related variables are key in providing information for research design, whether examining causality (e.g., space as variables related to extent, pattern, distances, temporal trajectories of change), cognitive dimensions of human–environment interaction (e.g., emic understanding of space, ethnoecology of landscapes, comparing emic and etic perspectives to landscapes), and/or symbolic (e.g., distribution of sacred places, paths of pilgrimage). Spatial data and analysis have become increasingly central to one’s interaction with local communities, such as when providing support to territorial recognition or participatory mapping of cultural resources or discussing visions of environmental management. Remote sensing (RS), geographical information systems (GIS), and global positioning systems (GPS) are complementary geospatial technologies and methods, collectively used in geospatial analysis (Brondizio and Roy Chowdhury 2010; Goodchild and Janelle 2004; Jensen 1996). Geospatial analysis is the analysis of features and their attributes that are distributed in space and are tied to a coordinate system (Bolstad et al. 2005). Geospatial analysis can be used to integrate data, generate data, summarize data, create maps, design research, and analyze data. A GIS helps organize, display, maintain, store, model, and analyze spatially interrelated data (Bolstad et al. 2005). The power of GIS is to link and synthesize information from different types of data: spatial, tabular, graphical, textual, and audiovisual. RS involves the collection of surface reflectance data from the Earth at a distance, such as in analog/digital aerial photography or satellite imagery recording information about the Earth’s surface such as location of urban areas or forest cover, in gridded (raster) products. RS provides primary data, including historical data, otherwise unavailable, thus representing an essential part of geospatial analysis. Finally, GPS provides real-time locational coordinates and altitude information through the use of a hand-held device that receives signals from an orbiting network of 24 satellites. Initially developed during the early 1970s, since the 1990s GPS devices have become an integral part of anthropological fieldwork allowing worldwide use for recording not only coordinate information, but also mapping of sites of interest, trails, pathways, and the

601

14_107-Bernard.indb 601

6/9/14 10:03 AM

602

Eduardo S. Brondizio and Tracy Van Holt

location of photographs. GPS data, from most types of devices, are usually compatible and can be readily uploaded into GIS and image processing software as illustrated for various applications below. A sample of anthropological research topics where geospatial analysis has come to the fore includes settlement pattern, tracking movements, landscape change, resource use and foraging, neighborhood analysis, participatory mapping, and social network studies. While these research topics are not restricted to anthropologists, anthropologists are well positioned to integrate their rich ethnographic experiences and research methods designed to assess and understand human behavior in ways not addressed by other scientists working on these topics. The purpose of the chapter is to provide an introductory outline of the main concepts underlying geospatial analysis and to overview basic steps and processes involving in producing and analyzing spatial data relevant to anthropology. We also discuss different types of questions in geospatial analysis that are relevant for anthropologists. We are not trying to cover the software skills required to accomplish the tasks below but to outline basic concepts and processes involved in using geospatial data for anthropological research, from data acquisition and generation to data analysis. Our aim is to provide tools for anthropologists to work as effective collaborators in interdisciplinary teams. Let’s start by expanding on the definitions of GIS and remote sensing presented above and some basis characteristics. Typically, GIS is software that organizes layers of data sharing a common coordinate system and containing attributes. The attributes are data attached to each individual unit in a layer. For a layer containing land parcel features, attributes may include, for instance, year of ownership, area of land, total taxes in a given year, and the size of the building on that parcel. Specific attributes can be selected to be visually, numerically, or textually displayed in a layer. Thus, in a GIS, layers can be displayed, synthesized, and analyzed to create maps, model phenomena, and generate new data layers. Each layer can be either vector or raster data formats. A vector layer can have three types of features: point (e.g., point location of houses), line (e.g., a road), or area features (e.g., an indigenous territory). Raster data represent continuous data across an area and are organized into equally sized grid cells (usually called pixels). In the case of a map layer, the resolution of the pixels can be defined by the user or a given characteristic of the data, such as the level of detail of the data. Remotely sensed images, such as from the Landsat satellite system, are always organized in raster format (i.e., an image represents a continuous layer or matrix of pixels) (e.g., a Landsat Thematic Mapper [TM] image has a 30 x 30 m resolution for each pixel). In general, remote sensing refers to the activities of recording/observing/perceiving (sensing) objects or events at faraway (remote) places (Jensen 1996). The information recorded is carried through electromagnetic radiation. The vast majority of applications of remote sensing in anthropology rest on optical remote sensing (ORS). ORS depends on the sun as the source of illumination. ORS mostly makes use of visible and infrared wavelengths captured by sensors that detect the solar radiation reflected from targets on the ground to form images of the Earth’s surface. Remote-sensing technologies enable the mapping of the Earth’s surface and subsurface features based on their differential reflectance of incident electromagnetic energy (i.e., their spectral profile or signature).

14_107-Bernard.indb 602

6/9/14 10:03 AM

19: Geospatial Analysis

603

A fundamental distinction among remote-sensing technologies is based on the source of that energy: Passive sensors record reflectance of incident solar radiation (i.e., the energy source is independent of the measurement device, exemplified by optical systems [ORS] such as sensors in Landsat satellite series); active sensors record the return of wavelengths generated by the sensor system itself, such as in microwave RADAR or LiDAR sensors. The underlying principle of remote sensing is that different materials reflect and absorb electromagnetic energy differently at different wavelengths (as explained below, an image usually has several bands each representing a wavelength interval). Thus, the targets can be differentiated by their spectral properties in the remotely sensed image bands that produce visually distinct features (like in any photograph) and different reflectance or spectral signatures. UNDERSTANDING CONCEPTS AND BASIC STEPS IN GEOSPATIAL ANALYSIS

The field of geospatial analysis is large and complex and a full technical explanation is beyond the scope of this chapter. We focus on three components of geospatial analysis to introduce basic techniques of data processing to anthropologists: understanding data structure (types of layers, spatial extent, and resolutions), searching and preparing data, and generating secondary datasets.

Understanding the Basics of Geospatial Data Structure: Types of Layers, Spatial Extent, and Resolutions Geospatial data and the articulation of spatial information with nonspatial attributes create opportunities and challenges in research design. Defining units of analysis and matching between social and environmental phenomena has been a long-term concern of anthropologists for concepts like culture area, niche, population and ecosystems, ethnic boundaries, social groups, nation-states, reserves, and social networks. Anthropologists help bring attention to local heterogeneity within regional studies, thus contributing ways to address challenges involved in linking levels of analysis (Brondizio and Moran 2012; Guyer et al. 2007). These include matching spatial data resolution(s) and social phenomena; accounting for historical differences across space; and considering observable and non-observable variables, visible, and invisible landscape features to avoid misguided interpretations of causality, determinism, and ecological fallacy (i.e., a mismatch of units of analysis when inferring cause and effect in the analysis of environmental change). Thus, it is important for anthropologists to understand the structure of different types of layers and trade-offs in layer properties to make the best use of geospatial data. Geospatial analysis, and remote-sensing data, in particular, can help reverse or reinforce these interpretations. For instance, Figure 19.1 in Box 19.1 illustrates different representations of intensively managed açaí agroforestry (i.e., by showing it as not distinctive from unmanaged forests and, conversely, as a separate land use category). Vector files are classified as a point, line, or area layers. The vector layer shown here is of fisheries management areas in Valdivia, Chile (Figure 19.2). A layer with the location of cities, which is a typical point layer, represents cities with an x,y coordinate for each point (city). The cities are sized according to population levels in 2002. The road layer, a typical line layer, is generated when multiple x,y coordinates are connected

14_107-Bernard.indb 603

6/9/14 10:03 AM

604

Eduardo S. Brondizio and Tracy Van Holt BOX 19.1. MAKING THE INVISIBLE VISIBLE IN AN AMAZONIAN LANDSCAPE

During the past 40 years, the growth of urban populations in the Amazon has created significant market demand for regionally preferred food sources such as the açaí palm fruit (Euterpe oleracea Mart.). Açaí is a regional staple food consumed widely by rural and urban populations of the Amazon. Since the mid1990s, açaí has gained national and international markets as a gourmet and fashion food item. Intensification of the production system followed increasing market demand, thereby changing the regional economic profile and the regional landscape. Today, açaí fruit is the most important source of income for the vast majority of riverine households and a significant sector of urban populations in the region. The evolution of the açaí economy in the past four decades has created a complex structure of production, distribution, commercialization, and processing that is specialized and socially ranked (Brondizio 2008). Building on existing ethnobotanical knowledge and agroforestry techniques, the significant increase in fruit production has been a direct result of forest management and agroforestry planting by local farmers. As a consequence, management and planting strategies have transformed floodplain forest areas into acaizais, managed forest areas planted with mainly açaí palm. Despite their high economic productivity and level of agroforestry intensity, areas such as these are often treated as areas of extraction and classified as unmanaged natural forest. Extractivism and extractivists are social categories associated with the long history of extractivist economies in the Amazon, which has been extended to local farmers in the context of the expanding açaí fruit economy. Such categories perpetuate the perception that these areas are native forests, which provides fruits without human input, thus being seen as unproductive from an agronomic point of view. Consequently, their managers are referred to as passive extractors rather than active producers who manage forest intensively in response to market demand. These social and environmental categories have implications for local producers, such as limiting their access to agricultural credit or limiting the recognition of land tenure rights based on previous productive use of the land. Images at the top of Figure 19.1 illustrate thematic classes of unmanaged and managed floodplain forests overlaid on top of satellite images. Images on the bottom represent two levels of detail in land cover classification and their consequences for depicting types of land use (and, thus land users). On the left, the land-cover classification disregards acai agroforestry as a distinct land use (i.e., as part of unmanaged floodplain forest). The image on the right shows a classification system that accounts for açaí agroforestry as a separate class. From an invisible land use system, this depiction of the landscape shows it instead as the most important and intensive land use activity of the area. Revealing the role of local farmers in intensifying a food production system based on local knowledge and without deforestation helps highlight their economic contribution to the region and reconsiders the economic identity of local producers from extractivists to forest farmers (Brondizio 2008). How users of geospatial technologies classify land cover and land use is essential to not only make the invisible visible but also to further our understanding of the underlying processes of social–economic and landscape change.

14_107-Bernard.indb 604

6/9/14 10:03 AM

19: Geospatial Analysis

605

Figure 19.1. Visibility and invisibility of forest management and land managers: Color composite and thematic representation of landscape management in a subregion of the Amazon estuary.

together (e.g., when you use a GPS to map a path). A map of counties or municipalities is a typical area layer. An area layer contains multiple x,y coordinates that are interconnected in a circumscribed way. Depending on the scale of analysis or type of data, the area of a city, however, could be represented as an area, while a county could be represented as a point. The user chooses which format to use to represent the data and the choice of point, line, area, vector layers or a raster layer, which depends on available data (and, in some cases, the scale of a printed map), and the intended analysis and mapping goals. The information in each layer, and in each feature within a layer, is connected to a set of attributes. Attributes can be derived by using operations in a GIS to calculate spatial information about the layer, such as the length of a given segment of road or the area size of a county. Attributes can also be linked up from other databases, such as a census dataset that has the demographic information of counties or from assessments in the field, like whether the road is paved or not, or the quality of the fisheries resources in the management area, or the demographic composition of a household or community. Vector and attribute information are linked

14_107-Bernard.indb 605

6/9/14 10:03 AM

14_107-Bernard.indb 606

6/9/14 10:03 AM

Figure 19.2. Nearshore fisheries management areas and the Valdivian Watershed, Chile. The villages are point layers that are scaled by population size. The road layers are a line layer. Fisheries management areas, the Valdivian watershed, and lakes are represented as areas. The user’s decision for file format (point, line, or area) depends on convention, the types of analyses, and the overall objective of the map.

19: Geospatial Analysis

607

through a common identifier. The location data are stored in a coordinate system that contains x,y locations; sometimes z information (height) is also included. For digital imagery (e.g., digital aerial photography or digital satellite images), the level of detail can be defined in terms of its spatial, spectral, radiometric, and temporal resolutions. These resolutions trade off with each other and with the overall areal extent covered by each image scene (see Figure 19.3). Spatial resolution refers to the size of individual pictorial elements (pixels), corresponding to rectangular portions of the Earth’s surface being imaged. Higher spatial resolution systems such as IKONOS (ranging from ~0.8 to 4 m) or Quickbird (ranging from ~0.6 to 2.8 m) satellite images map Earth’s surfaces in great detail, with submeter pixels, but offer limited areal extent. Spectral resolution refers to the precision with which sensors partition the electromagnetic spectrum into bands to record reflectance. The human eye is sensitive to wavelengths in the visible range, or 300–750 nm (blue to red wavelengths). NASA’s Landsat Thematic Mapper (TM) sensor system captures the visible spectrum in three ranges or reflectance bands (blue, green, and red light), and additionally records reflectance in four infrared bands (three near/mid- infrared and one thermal), as opposed to up to 220

Figure 19.3. Trade-offs in spatial, spectral, and temporal resolution as seen in this composite image of the city of Altamira in the State of Para in the Brazilian Amazon. The city is located in the center of the image (shown here in bright white) along the Xingu river (dark), surrounded by roads on the left (also bright white here), agricultural fields (light gray), and forests (dark gray). Figure prepared by Scott Hetrick, ACT-Indiana University.

14_107-Bernard.indb 607

6/9/14 10:03 AM

608

Eduardo S. Brondizio and Tracy Van Holt

bands in hyperspectral remote sensing (e.g., NASA’s EO-1 Hyperion sensor). Radiometric resolution refers to the levels of quantization (referred to as bit depth) with which the sensor captures spectral reflectance intensities in a pixel, typically ranging from 0 to 255 (8 bits), or 256 levels of gray (i.e., whereas zero represents black and 255 represents white). Temporal resolution refers to the revisiting frequency of the sensor for a specific point on the Earth. Landsat TM has a relatively high spatial resolution (~30 m) and temporal resolution of 16 days, compared to the NOAA AVHRR satellite’s spatial resolution of ~1.1 km but a high temporal resolution at twice-daily imaging frequency. Some sensors, such as those from the French SPOT series or Quickbird, can generate images based on customers’ orders. These trade-offs are illustrated in Figure 19.3. The figure shows differences in spatial resolution (from sub-meter to kilometer) and its impact on the level of detail of the same location (the city of Altamira, located between the Trans-Amazon Highway and the Xingu River in the State of Pará in Brazil) across images representing eight commonly used sensors. As ground features (e.g., parts of an urban area, roads, river, forest, and agricultural sites) vary in spatial extent, spatial resolution has differential impact on our ability to observe and collect data about them, as illustrated by this figure. However, higher spatial resolution may come at the cost of temporal and spectral resolution, and image areal extent. Usually, the most expensive commercial images have the highest spatial resolution and lowest the temporal resolution. This helps explain, in part, the popularity of Landsat images, particularly from the TM (thematic mapper) and ETM+ (enhanced thematic mapper) sensors. The Landsat program, dating back to 1972, has offered continuous and comparable remote sensing data of the entire globe at low or no cost; the availability of 40 years of data is invaluable for anthropological and environmental research. Offering a relatively high spatial resolution of 30 m (or higher for panchromatic bands), spectral bands from the visible to the mid-infrared spectrum, a temporal frequency of 16 days, and a spatial extent of 185 × 185 km, Landsat TM and ETM+ are suitable to most applications of interest to anthropologists, excluding some types of urban and archaeological studies. The good news for Landsat users is that NASA and the USGS are committed to continuing the long history of the program through the new Landsat Data Continuity Mission (LDCM). The LDCM satellite will include a combination of sensors offering comparable images to the Landsat historical series and additional products. It was launched successfully in February 2013.

Search, Preparation, and Preprocessing UNDERSTANDING THE BASIS OF COORDINATE AND PROJECTION SYSTEMS

Many geospatial analyses are conducted on a single layer. However, most users integrate information from multiple layers to create maps and databases, generate new layers, or analyze data. To overlay different layers together or to integrate GPS coordinates taken from the field or from other maps, the information needs to be in the same datum and coordinate system (either geographic or projected). The geographic coordinate system mimics the three-dimensional globe, where the projection coordinate system projects the three-dimensional globe in two dimensions (as on a piece of paper). This transformation distorts the globe and therefore distinct

14_107-Bernard.indb 608

6/9/14 10:03 AM

19: Geospatial Analysis

609

projections are designed to control for the distortion in shape, area, distance, direction, or conformality (Krygier and Wood 2001). One of the most commonly used projections in the field is the Universal Transverse Mercator or UTM; UTM preserves shape well. When this projection is used in areas that are a few square miles, it is basically distortion free. UTM is a metric system, which makes its coordinate system easy to understand and facilitates navigation in the field, such as when locating a coordinate collected with a GPS on a map or printed or digital satellite image. Datums are an element of a coordinate system. They explain the model of the Earth using a sphere, a spheroid, or ellipsoid that is defined by major and minor axes (the distance from the center of the Earth to the surface). They also help explain where the origin is centered. For more information on map projections, see Maher (2010). While software such as ArcGIS can project data on the fly based on the similarity of their geographic coordinate systems, often we find that the datum associated with different maps or layers is not the same. This difference in datums can result in errors, such as with lining up points between layers. For example, the coordinates of Seattle in North American Datum 1927 (NAD27) are 122° 19' 1.2", 47° 35' 20.4" but this same location in North American Datum 1983 (NAD83) is ~95 m away (122° 19' 5.7", 47° 35' 19.8"). Is 95 m an acceptable error? Possibly, however, the errors are distinct across the globe. The coordinates of New York (−74° 6' 0.0", 40° 45' 0.0"), for instance, are basically the same in these two datums Models of the Earth (geodesy) are becoming more precise, and there is an international movement toward one datum, World Geodetic System 1984; the latter is often used when this information is not available to the user (as may be common in maps or spatial data sets). People often assume that if they have coordinates such as 122° 19' 1.2", 47° 35' 20.4" they have all the info needed, but without the datum and coordinate system information, these points may not line up with others. This means that if someone provides an Excel sheet with coordinates, or passes on a file for use with GIS, or writes down coordinates from a GPS unit, the information is incomplete without the coordinate-system information. Coordinates come in mainly three formats. Coordinates 43°30" 0', 3°3" 0' are in degrees, minutes, seconds format and most likely represent a geographic coordinate system. The same coordinates in decimal degrees are 43.32, 3.33 and are typically associated with a projected coordinate system; this format is useful for many mathematical calculations since it is in the decimal system (as opposed to degrees). The same coordinates transformed to the UTM projection are (x or easting) 622483, (y or northing) 4735440. This projection is in meters, which is useful for distance calculations and for fieldwork. SEARCHING FOR GEOSPATIAL DATA

In considering the selection of data for a research project one should consider: (1) the research question of interest and the ideal and minimum data requirements; (2) the time frame of interest and/or the sequence of relevant events affecting an area; (3) the seasonality of the area; how climate or land use and land cover may change through the year; and (4) the spatial dimensions (extent, resolution, and

14_107-Bernard.indb 609

6/9/14 10:03 AM

610

Eduardo S. Brondizio and Tracy Van Holt

pattern of feature distribution in a landscape) of important social and environmental features. Brondizio and Roy Chowdhury (2010) offer guidelines to consider these features during research design. Once the questions and data needs are outlined, potential sources of eligible RS/GIS data may be identified. Image scenes for specific regions may be quickly located using global referencing systems (e.g., the World Reference System [WRS I or II] to identify Landsat images) and/or locational (e.g., UTM or geographic) coordinates. Several search engines and public data archives are available (see Appendix 1 for a sample). We suggest that a good place to start is the United States Geological Survey (USGS) Earth Explorer data gateway. Data quality concerns are important when selecting among alternate short- or long-term RS datasets that may match research needs. These include considerations of scene cloud cover and patterns of geometric or radiometric corrections that are likely necessary. Quality and informational content is equally important in ancillary GIS data, and researchers need to assess the type and detail of thematic information about features of interest as well as their cartographic aspects (e.g., projection system, scale, and minimum mapping units). These features will fundamentally influence data input as well as the quality of registration/overlay with other data layers. Some of the most accurate geospatially explicit datasets can be found on government and nongovernment organization websites. Usually these datasets have metadata associated with them. Metadata will explain how the data were generated, what the attributes were, what the accuracy of the data is, who created the dataset, what the datum and projection of the data are, and possibly a contact person or agency if questions arise about the data. Metadata may even explain how you can analyze the data (e.g., U.S. Census datasets). These datasets are relatively straightforward to use and easy to integrate with other layers. Some datasets are not geospatially explicit but can be transformed to be geospatially explicit based on information acquired from their metadata. Anthropologists often have access to a variety of printed regional maps (including historical sources) produced by government organizations and/or specific projects. We also create maps in the field in conjunction with local collaborators (with various levels of cartographic accuracy or lacking it altogether). Accessible scanning technology today permits relatively easy integration of such maps into a GIS. Their integration into GIS and/or to remote-sensing data (for overlay on aerial photography or satellite images) is based on assigning the scanned source to a cartographic reference system through the process of georeferencing, discussed below. The level of precision of their overlay with other layers will depend on the cartographic quality of the original source, but GIS digitalization tools, also described below, also allows for on-screen corrections and adjustments based on user needs. STEPS OF IMAGE PROCESSING

Processing satellite imagery or digital aerial photographs of varying spatial, spectral, and radiometric characteristics involves three main phases: (1) preprocessing: geometric correction (georeferencing), subsetting or mosaicking, radiometric correction (e.g., noise removal), and atmospheric correction; (2) visual or digital thematic classification relevant to the chosen research question; and (3) post-classification: analyzing and extracting relevant thematic data from classified images (e.g., land cover types and extent

14_107-Bernard.indb 610

6/9/14 10:03 AM

19: Geospatial Analysis

611

in a particular area). Preprocessing is discussed next, while classification and change detection are covered in Basics of Data Analysis. These sections draw on Brondizio and Roy Chowdhury (2010), where additional information can be found. The bibliographic references below also provide detailed information about each step. PREPROCESSING

Geometric correction (often used interchangeably with the term georeferencing)1 involves spatial transformations to reassign the matrix of pixels forming an image (with x,y positions) to a geographic (latitude/longitude) coordinate system or a standard map projection, such as the Universal Transverse Mercator (UTM) (Jensen 1996; Wilkie and Finn 1996). While images can be ordered with provider-applied geometric correction, in-house corrections enable greater control of positional accuracy and error distribution. Geometric correction is based on locating ground control points (GCP) representing known locations (e.g., from field-derived GPS points) on a target image. The user identifies key locations/landmarks (GCPs) that are clearly visible in both reference map and target images. To minimize the magnitude of locational inaccuracy, captured as root mean square error (RMS or RMSE) assessed during coordinate transformation (Jensen 1996), many GCP points need to be evenly selected across the entire scene. A basic rule of thumb is to aim for a RMSE equal to or less than half the image pixel size and use the lowest-order transformation function feasible (e.g., the nearest neighbor or affine transformation, for faster processing and least alteration of pixel attribute values). After georeferencing, the transformed image may be overlaid with the reference maps to check for the quality of their spatial matching. Depending on the specific research objectives, data source, and study region, at this stage the researcher makes choices regarding whether image mosaicking (i.e., combining two adjacent images) or sub-setting (i.e., cropping a segment of an image) are necessary to cover a study area in adequate spatial extent and/or detail. At this stage, the user also considers which spectral bands to use (i.e., whether all bands are necessary and whether transformed bands are needed), and whether radiometric and/or atmospheric correction algorithms are necessary to improve signal-to-noise ratios before further analysis (Green et al. 2005). Preprocessing may further entail band transformations and/or the integration of GIS techniques. For instance, a variety of algorithms allow the generation of transformed bands representing indices of vegetation (e.g., NDVI or VI), soil, impervious surfaces and spatial texture.

Preparing Basic Data Layers GEODATABASES

Geodatabase is a term used by ESRI (ArcGIS parent company) to designate a collection of different types of geographic datasets used in ArcGIS and that can be managed as a file folder or a relational database. Geodatabases are efficient ways to organize data and offer stability, portability, and the ability to standardize data entry. This organization often speeds up processing time. Project management is facilitated since the entire database can be moved as one piece of data. This reduces the issue of having files throughout a hard drive (or multiple hard drives) or having data in multiple projections. ESRI offers personal, file, and scalable geodatabases. The file geodatabase is

14_107-Bernard.indb 611

6/9/14 10:03 AM

612

Eduardo S. Brondizio and Tracy Van Holt

preferred because there is a limit of 1 TB per table, whereas the personal geodatabases have a max of 2 GB; the scalable geodatabases are more complex and require advanced specialized knowledge. Geodatabases also take advantage of the integrated data sets. For example, users can take advantage of topology, that is, how feature classes may share geometry (boundaries, or corners). For example, a concession area to harvest mussels shares a topology (boundary) with a river area because the concession area is located within the river borders. If the user verifies the concession-area boundary with a GPS unit in the field and subsequently edits the concession-area layer in the geodatabase to be wider than the river feature, then the river boundary would also change. Geodatabases also allow for more sophisticated network analyses and allow point and line features to be transformed to edge and junction features. For more on geodatabases, see Zeiler and Murphy (2010). GENERATING A POINT, LINE, AND AREA FEATURE LAYERS FROM GPS FIELD POINTS OR MAPS

Collecting point locations (i.e., the coordinates of a given location) using a GPS has become one of the most used routines for anthropologists doing fieldwork. GPS points representing houses, village centers, agricultural fields, sacred sites, schools and health facilities, location of people, areas of conflicts, and neighborhood landmarks can yield a variety of data for map production, site description, and several types of spatial analysis. GPS points represent spatial coordinates collected within a given coordinate and projection system and can be imported into a map layer or used to generate a new one. Each point can be linked to qualitative and/or quantitative attribute data; they can also be associated with a particular record, like a photograph or image or an object. Points can also be generated from existing maps or from Google Earth.2 Points can be connected to generate additional data. For instance, point locations can help calculate distances (e.g., between points) or serve to construct and represent a path, or a network (e.g., origin and destination of migrants). Three or more points can be connected together to identify the location of household plots (i.e., areas). This layer can then be used to summarize the percentage of each land use in a household plot available from another overlaid layer. Points help generate other layers. Points can be used to create buffer zones that contain equidistant intervals around the point. In turn, these buffer zones can serve as units of analysis to, for example, calculate the number and size of a land use class or other features at different distances from a household or the center of a community (see examples in Box 19.2). GPS units can help with field navigation and data collection. Recreational units cost <$500 and have a 3–5+ m accuracy; mapping units cost several hundred to several thousands of dollars and have a 1–3 m accuracy in the field; survey units cost thousands of dollars and have sub-meter accuracy in the field. GPS units may not work well under forest canopy, inside buildings, or down in some steep topography because the signal cannot penetrate through these materials and the GPS needs to communicate/triangulate with multiple satellite images for the best results. If research is conducted under a forest canopy, however, additional equipment can help the signal penetrate the canopy. In areas of rugged and steep terrain, the signal may bounce to another portion of a mountain, rather than directly to the satellite. The signal from satellite images varies depending on the time of the day because a different number of satellites may be available.

14_107-Bernard.indb 612

6/9/14 10:03 AM

19: Geospatial Analysis

613

BOX 19.2. NEIGHBORHOOD STUDIES THAT INTEGRATE POINT AND AREA DATA WITH INTERVIEWS The following neighborhood studies geocode food, drink, and tobacco store locations (point data) to test how a person’s proximity to these stores affects culture change, health choices, and health outcomes. Pereira et al. (2010) examined food acquisition of African refugees in Australia. Their local collaborators recorded food diaries. Pereira et al (2010) mapped the location of food sources in neighborhoods using online satellite images and direct observations. They concluded that although migrants did not live in food deserts (places with little available food), their food intakes remained inferior to the Australian counterparts in the study. Their interviews showed that physical proximity does not guarantee a good diet. Refugees chose less healthy food options, which were actually farther away, because they were not integrated into the Australian diet. Through interviews, they concluded that refugee orientation events were where refugees could be introduced to foods more typical of the Australian population, improving their access to healthier food choices. Franzen and Smith (2010) compared food access and grocery-purchasing events of Thai/Lao immigrants that varied in the amount of time they lived in the United States with U.S.-born individuals. They mapped grocery store locations and integrated this with the consumer price index and a survey about food purchases. They found that immigrants who spent more time in the United States bought food closer to home in stores that sold more U.S. products. During interviews, more culturally integrated immigrants reported that stores were selected for their one-stop shopping potential. West et al. (2010) tested whether proximity to alcohol and tobacco stores influenced use. They integrated surveys with census data and locations of alcohol and tobacco places. Use was negatively associated with distance from respondents’ home to the nearest alcohol and tobacco retailer and positively related to peer use of tobacco and alcohol. These studies show the relatively basic integration of geospatial data using simple distance calculations. They also illustrate the potential to include more complex analysis involving geospatial correlations. Census tract data are available across many countries and typically provide basic demographic information about the people in a given area. Many countries also have additional census data on agricultural and economic activities. Freisthler et al.’s study (2007) integrated census tract data (areas) with child maltreatment rates (i.e., social service attributes in an Excel format without geospatial coordinates) and point locations of alcohol outlets (which were geocoded). They found out that the rate of child maltreatment was different across racial groups and was strongly correlated with poverty level, proximity to alcohol outlets, single parents on unemployment, and a high ratio of kids to adults in a home. They argue that this approach can help in developing programs targeted to specific sectors of the population so that maltreatment is reduced.

14_107-Bernard.indb 613

6/9/14 10:03 AM

614

Eduardo S. Brondizio and Tracy Van Holt

The GPS unit identifies the accuracy of each point based on the number of satellite images consulted, and your research question can help identify what level of accuracy is appropriate. The geospatial industry is moving toward GNSS (global navigation satellite system) to increase accuracy and reliability of GPS points. This system will integrate the U.S., Russian, and European Union’s system of satellites and other technologies that help users identify locations on the ground. The datum and projection in each GPS can be changed. Researchers often use UTM coordinates in the field because the units are in meters and can help guide researchers in the field. It is challenging to walk 0.2° north, which is why UTM is so widely used. To walk 20 m north from an original point, 622463 (N), 4735440 (E), the GPS unit can be set to head to 622483 (N), 4735440 (E). In most topographic maps, both UTM and latitude longitude in degrees, minutes, and seconds format are provided. Most GPS units can be connected to a computer so you can download the data easily, although software may be necessary for this step. Some software will download data in decimal degrees, but the display may show UTM. Depending on the GIS and/ or image-processing software one is using, GPS-downloaded points can be imported and uploaded directly as a layer. Another common procedure is to generate a spreadsheet where each line represents a point location (with columns representing respective point ID, x and y coordinates, and attributes associated with each point); this spreadsheet can be imported into GIS software (or image-processing software) to create a new point layer. If you have geographic coordinates (lat/lon) written down as field notes, you can enter them into a spreadsheet to generate a map. Free software is available on the Internet to transform between decimal degrees, degrees minutes seconds, and UTM coordinates, among others. Most RS and GIS software offer coordinate converters. Some cameras can also provide coordinates, and coordinates can be taken from phones and/or GPS units. As with collecting point-specific coordinates, a GPS unit can be set to collect continuous data (i.e., linked point coordinates collected at very short time intervals) to produce a line, such as representing a road, streets, or a tracking path. These data are extremely useful for anthropologists as fine-scale information about such features are usually not available from cartographic sources or maps that are outdated. A line layer can also be generated from digitizing lines in an existing map or by tracing (on-screen) over a satellite image or aerial photograph. The editing feature of ArcGIS, for instance, allows you to generate new layers (i.e., point, line, or area). You can either use a feature recognition program such as ArcScan, which is an extension for ESRI’s ArcGIS suite of geospatial software, to help draw the lines or the lines can be drawn by hand. There are options to assist you in snapping line vertices to other line vertices so there are no gaps. Area layers can be generated from GPS continuous data points or from existing maps. The conversion of a raster thematic image (i.e., a satellite image classified in thematic categories or classes) to vector will also generate areas (i.e., each area or polygon representing the area of a thematic class). Digitizing areas from a map (or points and lines) is a common area- or polygon-generating procedure. For instance, in participatory mapping people draw boundaries over an image or aerial photograph or even on a blank piece of paper. These notations can be digitized or scanned to create an area layer (or line and point layers). Like in the United States, many census bureaus offer area vector layers of

14_107-Bernard.indb 614

6/9/14 10:03 AM

19: Geospatial Analysis

615

political-administrative and census units for download and/or acquisition. Likewise, many organizations offer area layers of features of interest to anthropologists, such as community areas, conservation reserves, and indigenous territories, among others. LINKING AND RELATING ATTRIBUTE DATA TO POINT, LINES, AND AREA LAYERS

Associating features with quantitative and/or qualitative data is perhaps most relevant to anthropology (see various types of examples for instance in Fox et al. 2003). Each feature (point, line, and area units in a layer) has associated attribute data that are organized into a table. Attribute tables (and relational databases) are based on the association of a given feature ID (unique identifier) and the respective relevant variables. For instance, a point feature representing a house may have attributes containing information about size, composition, origin, type of household, income, and so on. If several houses are mapped in a layer, each one will be an observation (i.e., a row) associated with different variables (i.e., columns) that together form an attribute table associated with that layer. A line feature representing a road may have an attribute table with information about type of pavement and whether the road is public or private. An area representing a neighborhood may have an attribute table with census information associated with it. One example is cesarean section rates at Florida hospitals (Figure 19.4). This nongeospatially explicit dataset comes in an Excel format from the Florida Agency for Health Care Administration (AHCA) and shows the C-section rates at Florida hospitals. No geographic coordinates are provided. The Florida Geographic Data Library (FGDL) at the University of Florida, however, has geocoded the locations of all hospitals in Florida. The FGDL database can be joined or related to the hospitals layer. The match needs to be exact. The hospitals are matched by name, then the AHCH code is added to the C-section rate data by hand. The two tables are then either joined (1-to-1 relationship) or related (1-to-many relationship). In addition to attribute data, features on a layer can also be linked to digital representations or illustration of items, such as photographs, videos, and illustrations of objects, audio files, or scanned documents. In this case, a point, for instance, serves as a

Figure 19.4. Data that are not geospatially explicit can often be converted to geospatially explicit datasets. The Florida Agency for Health Care Administration dataset of C-section rates in Florida hospitals has a code for each hospital. The Florida Geographic Data Library has a geospatially explicit point layer of hospitals in Florida. This dataset can be manually modified to include the AHCA# and these two datasets can be joined.

14_107-Bernard.indb 615

6/9/14 10:03 AM

616

Eduardo S. Brondizio and Tracy Van Holt BOX 19.3. NETWORKS AND SPATIAL RELATIONSHIPS

Network metrics are used to understand power, knowledge, and diffusion of information among many other topics. While network metrics and interpretation are well established, how these metrics interact across space and how the surrounding environment influences networks is less well understood and studied. Social network analysis quantifies social structure. Integrating geospatial analysis with network analysis requires one to have a location for the actors or nodes of interest. A leading edge of network analysis is to understand when geography, geospatial orientation of people, and the environment match social network metrics and when these cases do not and why. In network analysis, a connection or tie is present when there is a relationship between two individuals (e.g., based on a questionnaire or if people participate in similar meetings, etc.). The structure of the network itself or the structure of an individual’s position in the network can be measured. Whole network measures include the size (number of individuals) or density (how closely connected individuals are to each other). Individual position measures include degree centrality (the number of connections an individual has) and betweenness centrality (a measure of a person’s role in bridging different parts of the network). Faust et al. (1999) characterized the spatial arrangement of social and economic networks in Thailand. Tractor-hiring networks brought together larger sets of villages and spanned longer distances. They integrated the locations of the tractor-hiring network together with a thematic map that showed land cover and use (e.g., rice, upland agriculture, forest, water, other). They found that the networks in upland agricultural areas were denser than those in rice agricultural areas. By integrating network data with thematic maps, we can ask questions such as: Do the land use patterns influence the density of the tractor hiring networks or the number of individuals in key bridging positions? Are there geospatial features such as river barriers to network formation? We can also test whether technology eliminates distance limitations in forming social networks. Mok et al. (2010) asked whether distance matters in social interactions in the age of the Internet. They compared the role of distance in social networks pre- and post-Internet. They found out that email contact is generally insensitive to distance. However, face-to-face contact remains strongly related to short distances (>5 mi). Might cultural anthropologists integrate ethnographic interviews and social network type data with geospatial features to detect key distances for certain types of social interactions? Does the surrounding environment influence the network structure? Are there influences particular to rural areas and not to urban areas?

14_107-Bernard.indb 616

6/9/14 10:03 AM

19: Geospatial Analysis

617

link to a file representing the object of interest; depending on the GIS software, objects can be visualized in conjunction with map data. BASICS OF DATA ANALYSIS

Counting and Cross-Tabulating Features An example of a basic GIS analysis is counting points or calculating distance or area within a boundary. The boundary can be an area feature that is already established (such as a census block) or a buffer zone that can be generated around a point, line, or area features. Buffers are zones of distance around a given feature (e.g., a one-mile buffer around a city hospital). They can also represent, as proxy, units of analysis. Buffer zones are invaluable for a variety of analyses within and across layers. For instance, buffer zones can be created at a given distance interval (e.g., 500 m or 1 km) from a point location such as a hospital or at different distances from a given road or within and around a conservation unit. These layers of buffer zones or any other area can be used to query and/or crosstabulate features in other layers. For example, areas (a polygon or buffer zone) can be used to query the amount of different land use classes within a farm or reserve (crosstabulate operation) or query how many households exist at different distances from a health clinic (count operation). The user sets whether one or multiple buffers are used in an operation. Buffer distances should be meaningful relative to the research use question. If a survey of hunters in a village reports that they travel on average less than 10 km per day when hunting, then the buffer distance should take these data into account when determining buffer distances. Area Calculations Area calculations are among the most used estimates in geospatial analysis, when done both synchronically or diachronically. A grid of farm lots (a vector layer) can be overlaid on a thematic map of land-use types, for instance. Using a cross-tabulate operation as mentioned above, we can extract thematic information from the map at the level of each farm (or for a community, a county, a province, etc.). So, cross-tabulation will produce a table showing the amount (area) of each land use class (e.g., different types of agriculture, forest, water, pasture, etc.) for each farm. Diachronically, one can calculate changes in spatial features across time, such as the growth of an urban area or settlement mapped using remote sensing data in combination with historical maps or the deforestation rate within given units of analysis. As the area calculation of landscapes may be affected by relief in regions of rugged topography, the use of digital elevation models (DEM) will improve area estimates significantly. DEMs of different resolutions are available for most of the globe (or can be generated from a topographic map) and allow sophisticated 3-D visualization and modeling depending on the software. Distance Calculations Anthropologists are interested in a variety of distance calculations as distance represents important variables for people’s movement, access to services and resources, and patterns of settlement. There are two main kinds of distances: linear and cost-distance.

14_107-Bernard.indb 617

6/9/14 10:03 AM

618

Eduardo S. Brondizio and Tracy Van Holt BOX 19.4. INTEGRATING LANDSCAPE PATTERNS IN OPTIMAL FORAGING STUDIES

Anthropologists have used optimal-foraging theory to explain resource use behavior, for instance, for hunters and fishers (Aswani 1998). Optimal-foraging theory suggests that people will not travel greater distances or exert greater effort than necessary to harvest natural resources needed for subsistence. By integrating distance measures with counting and cross-tabulating features, buffers, and classification of satellite images, scientists have been able to advance this field of research, creating more nuanced models of optimal foraging. Alvard (1995) used optimal-foraging theory to examine the concept of the ecologically noble savage by testing whether hunters selectively harvested species to conserve wildlife. Traditionally, scientists characterized hunting behavior by calculating the distance from home to the center point of the hunting area (Alvard 1995; Turner and Hienaux 2002). Parry et al. (2009) and Turner and Hienaux (2002) expanded on traditional optimal-foraging measures and integrated satellite image data and GPS points in a GIS. They extracted information from the satellite images about the surrounding landscape and demonstrated that human behavior doesn’t simply follow an optimal ring, line, or specified point. Parry et al. (2009) integrated land cover/use information (they classified whether the landscape was fallow field, plantations, or native forest) with hunter catch per unit effort (CPUE) data. Parry et al. (2009) concluded that hunters had the highest CPUE in fallow fields, followed by native forest. The lowest CPUE was found in landscapes with tree plantations. In another study, Turner and Hienaux (2002) mapped livestock activities across agropastoral landscapes in semi-arid Africa. Like Parry et al. (2009), they found out that instead of pastoralists’ behavior following a simple point distribution model, distribution reflected local patterns of land use, topography, vegetation, settlements, and water points.

The first, as the name implies, can be an estimation between two or more points or along a particular path. The latter, on the other hand, takes into account distance relative to time and terrain conditions. For instance, while distance between two points may be the same, the time or the path it takes to move between them may be mediated by type of transportation, type of access, or quality of the road or topography. Costdistance estimates will consider these features in estimating distances.

Producing Thematic Layers through Image Classification Digital-image classification produces thematic images or maps that entail three main suites of techniques: supervised, unsupervised, and hybrid classification; hybrid classifications entail iterative combinations of the first two. Unsupervised classification allows the definition of classes without requiring knowledge of the image or study area, enabling the user to classify a large region into distinct categories of land cover in a time- and cost-effective way at the beginning of research. The process identifies natural groupings of pixels in an image based on their location in multispectral feature space (i.e., their reflectance patterns in multiple bands). The

14_107-Bernard.indb 618

6/9/14 10:03 AM

19: Geospatial Analysis

619

researcher, according to research questions, defines the desirable number of clusters. An algorithm (e.g., ISODATA) will generate an unsupervised classification that will require the user to assign actual names (i.e., thematic classes) as part of the classification process (Jensen 1996). The classified image generated by the software with 1, 2, 3, . . ., n classes listed will be assigned thematic names such as paved surface, forest, pasture, water, and the like. Spectral classes may be difficult to interpret sometimes, with little resemblance to local reality; considering how various local features are aggregated into broad land cover categories is a key part of the analytical process, such as illustrated in Box 19.1. This assignment of unsupervised spectral classes to informational, real world classes entails ancillary knowledge of the area and GPS-assisted field visits. Google Earth has become increasingly important in this process. In areas where high spatial resolution images are available, Google Earth can help greatly with the interpretation of classifications. The first step in supervised classification is to develop a target classification scheme of land cover classes of regional interest. Such schemes are usually developed in an interactive process, which involves the examination of land cover variability within an image as well as literature review and field knowledge. Many researchers prefer to adopt hierarchically nested classification categories: The advantage of a hierarchical scheme is that detailed land cover classes may be aggregated to address questions about land use and cover change at different levels of analysis. In supervised classification, based on target classes the user uses image classification software, such as Erdas Imagine or Multispec (very good and free image-processing software), to select training samples (TS) of known identity to classify pixels of unknown identity (Jensen 1996). The user defines areas of interest representing TS for each target class, which can then be used by the selected classification algorithm to classify pixels of unknown identity and to support accuracy assessment. TS need to be collected for as many samples as possible for each of the proposed land cover classes; TS should be large relative to image resolution (e.g., minimally 9 pixels or about 1 ha in the case of a Landsat TM image). Depending on field conditions, TS can be collected by observing areas along a roadside or river, during field visits with local farmers or informants, and many other creative ways (see Note 2). It is also important to ensure a wide spatial distribution of TS for land cover classes of interest around the area covered by the image to avoid clustered samples biased toward only one part of the image/landscape. TS areas can be located with GPS in georeferenced image printouts or through real-time links to the digital imagery loaded on portable laptops to allow for immediate adjustments/revisions. Most users prefer to use a combination of supervised and unsupervised methods (i.e., a hybrid approach). For instance, the statistics generated in the unsupervised classes may be used as TS for supervised classification. In this case, unsupervised classes are added to the list of reference [supervised] classes (which includes those based on TS selected from fieldwork or high resolution images) to be used for supervised classification. In any case (i.e., unsupervised, supervised, or hybrid), the user should use a combination of visual, spectral, and statistical tools to evaluate classes and improve class discrimination. Most image-processing software provides options for creating graphs of spectral signatures and different options to analyze their statistical separability. Overlaying a classified image

14_107-Bernard.indb 619

6/9/14 10:03 AM

620

Eduardo S. Brondizio and Tracy Van Holt

on a color composite image (i.e., color display of an image using three spectral bands) can be used to evaluate whether the classes are appropriately categorized. Decisions about image classification depend as much on the visual examination of context and pattern of classes on the landscape as on statistical and spectral analyses. Visual, spectral, and statistical analysis complement each other during the classification process. In addition to thematic categorical classification, spectral data can be transformed to create various types of continuous classification and indices representing biophysical gradients of the landscapes. While these processing techniques are usually done by more advanced users, several software packages offer straightforward routines to facilitate this type of data classification. Spectral band math operations can generate images where pixels represent different proportions of greenness, soil exposure, humidity, or different layers representing gradients of each of these features.

Multi-Temporal Change Multi-temporal change analysis is a common procedure in image analysis and in GIS. In a GIS, point, line, and area layers representing different periods of time can be compared and overlaid to generate maps of change or new maps derived from comparing categories over time. In RS, images of the same location from different dates can be overlaid for multi-temporal comparison and classification. Most commonly, images are classified independently, then compared and used to generate a map of change. However, spectral bands from different dates can be used to carry out simultaneous multi-temporal classification. In this case, images from different dates were georeferenced to each other (see above) and stacked together to form a single multi-temporal image file. One can combine multiple image dates into a single image file, although the level of accuracy between image dates (i.e., precision of georeferencing alignment) will decrease as the number of layers increases. An image file containing multiple dates can be used for multi-temporal image displays (i.e., up to three dates can be displayed at once in one image). In this case, displaying the same spectral band but representing different dates can help to identify areas on the landscape undergoing change (e.g., where urban areas are expanding, the extent of flooding, where deforestation and reforestation are occurring, fire scars, etc.). Land change studies are frequently concerned with the extent, direction, and rate of land use/cover change across multiple units/scales of analysis (e.g., buffer zones, farm lots, settlements, community areas, indigenous reserves, municipalities, states, countries, etc.) (Sirén and Brondizio 2009). Analysis of land transformation may refer to continuous or categorical changes using multi-temporal datasets. Techniques vary with research/application goals (e.g., urban expansion, agricultural intensification, deforestation, fire processes, etc.). The reliability of change analysis results is linked to the change detection technique employed and depends fundamentally on the quality of image georeferencing and registration (i.e., precision of alignment between dates), the quality of atmospheric and radiometric correction (especially when comparing spectral reflectance directly across dates), and the consistency and robustness of the classification system used for each date. The categorical change detection produces a third transition image depicting spatial changes along with a transition matrix indicating the frequencies of transi-

14_107-Bernard.indb 620

6/9/14 10:03 AM

19: Geospatial Analysis

621

tions from classes in date one to those in date two (Brondizio 2006; Brondizio et al. 1994). For instance, a pixel classified as forest in date one and soil in date two can be interpreted to reflect deforestation, while the reverse may indicate reforestation. It is worth remembering that spatial errors compound multiplicatively with data integration: Two land cover maps, with 90% overall classification accuracy, when overlaid for change detection will yield a transition image whose accuracy may not exceed 81% overall. BOX 19.5. LANDSCAPE PATTERNS AND CHANGE STUDIES Landscape-change studies examine the drivers and consequences of landscape change, or spatial and temporal patterns of change (Nyerges and Green 2000). Ruf (2011) tested whether small landholders who grow cacao were planting it in an agroforestry system as typically described in the literature. Ruf combined interviews with classified remotely sensed images of cacao farms. From the satellite image analysis and interviews, he concluded that most farmers were adopting sun-loving hybrid cacao trees. The farmers reported that agroforestry approaches limited their ability to compete in today’s global economy. In another study, Kusters et al. (2007) classified land use before and after a decree granted farmers the right to obtain a concession (quasi-tenure) for their agroforestry plots within the state forest in Indonesia. Although most people didn’t get tenure, the decree did influence land use decisions. People felt secure in maintaining their agroforestry despite not having tenure on the land; as a consequence, the native forest had some level of protection from encroachment. In their interviews, Kusters et al. discovered that the perception of tenure security— not actual tenure—was what was most significant in people’s land use decisions. In their study of patterns of landscape use in a Yanomami community, Albert and Le Tourneau (2007) integrated long-term ethnographic data, satellite imagery, and GPS to show how the spatial pattern of geographic land and resource use in this Yanomani community is configured in terms of reticular space, rather than, as previously described in the ethnographic literature, sets of mutually exclusive exploitation zones. Their study also shows how the community’s traditional spatial system has been adapted in creative ways to the displacement and sedentarization imposed by contact with a local government posts implanted in the area in recent decades. Examining different types of questions at the interface of land and seascapes, Van Holt (2012) examined whether fisher success was influenced by landscape change (i.e., tree plantation development and the associated effects of nutrient input on the seascape). In the closed-access Concholepas concholepas (loco) fishery, fisher success depended mainly on the environmental quality of the loco resource, which was related to broader landscape changes. Fisher experience couldn’t help fishers mitigate the environmental effect on their success (price paid per kg for loco). In contrast, in the Genypterus chilensis (congrio) offshore fishery, fisher experience and technology counted, and the environmental influence played a minimal role in fisher success (price paid per kg of congrio).

14_107-Bernard.indb 621

6/9/14 10:03 AM

622

Eduardo S. Brondizio and Tracy Van Holt

Producing Image Maps for Fieldwork and Participatory Mapping Anthropologists and other professionals need to carefully consider their choices depicting people and/or the landscape of an area, locations of particular features, and local social–ecological relationships. These challenges extend to broader uses of geospatial data, as highlighted in a publication of the National Research Council of the National Academy of Sciences (NAS 2007). Images and maps represent a form of power: Selected sectors have access to that power (e.g., datasets), and fewer still have the ability to manipulate it (e.g., represent or transform them). Geospatial data can be used to assert one’s territory over another, formalize ownership, show the distribution of valuable resources revealed by local knowledge, or portray illegal activities or settlements that may endanger local people directly or indirectly (Brondizio and Roy Chowdhury 2010). Anthropologists concerned with environmental cognition are increasingly using geospatial tools to represent local environmental perceptions, local topology (designation of places/features and their interrelations), and symbolic meanings imprinted onto landscapes. Ethnoecological methods have long provided tools to relate etic and emic views of landscapes and land use systems and to learn how social groups understand, materially and symbolically, their local and regional landscapes. These methods, in conjunction with GIS and RS tools, can aid in various forms of participatory resource mapping. Chapin et al. (2005) illustrates examples of working with indigenous communities to map their resources. His approach is to offer appropriate landmarks, such as rivers, roads, or other geographical features as a common base of reference from which to discuss the landscape. One of his efforts, for instance, has been transformed to official regional maps. Widayati et al. (2010) integrated participatory mapping (in this case, hand-drawn maps) with rattan cane harvest data and remotely sensed data that included slope and identified conservation areas in Indonesia. They asked the harvesters to identify where on a satellite image they harvested rattan cane. They extracted information from the satellite image and found out that the slope and distance from their home played a role in the quantity of rattan canes harvested. They also found that conservation areas were disregarded. D’Antona and colleagues (2008) used a combination of multi-temporal satellite images and sketch maps to study farm-level land use along the Trans-Amazon Highway. Farmers provided detailed sketches showing their land use choices and location of land use features not apparent in the satellite image; conversely, multi-temporal images helped with the recall process during interviews. These types of studies illustrate how these tools in conjunction with fieldwork can show variation in emic/etic understandings of resource use. The availability of high-resolution data and easily accessible software has made it increasingly common to use satellite images or aerial photographs to prepare field maps. Black-and-white or color composite images can be overlaid with a coordinate grid and other layers of interest (e.g., municipal/community/reserve boundaries, roads, river systems), and, if applicable, legends (symbols representing different thematic features). Although basic cartographic information is always important, the

14_107-Bernard.indb 622

6/9/14 10:03 AM

19: Geospatial Analysis

623

degree of detail varies according to the goals of the map, but should at least include scale, data source, and legend. Additional considerations in preparing image maps for use in field interviews include the selection of color composite combination, coordinate systems, and scale and extent (with implications for the size of a map). The selection of spectral bands and colors to depict their combinations (composites) should aim to balance familiarity of human vision with detailed information about the environment. Using data from Landsat series 4, 5, or 7 as an example, the most common color composite used for image printouts (one that offers a good trade-off between appearance and spectral information) assigns spectral bands 5 (mid-infrared), 4 (near-infrared), and 3 (red) to the red, green, and blue color display channels, respectively. This will produce an image that resembles somewhat our own perception of the environment (i.e., vegetation will be greenish, water will be bluish, and soil will be reddish). Other combinations relying on visible bands will also produce images with a visual resemblance to reality. Field maps can also be prepared with images from multiple dates. Figure 19.5 illustrates field maps prepared for interviews with colonist farmers along the TransAmazon Highway (Brondizio et al 2002). This field map shows the same farm lot (and surrounding farm lots) since the arrival of the family in the area. On the left is a sequence of color composite images, which are matched on the right with a thematic classification for each date. These maps were used in combination with a protocol for participatory mapping. These can be very helpful during interviews aimed at recalling previous land use or discussing changes on the landscape. Image printouts should also include a coordinate system (e.g., geographic latitude/longitude or in UTM), that one can use to locate a reference point collected with a GPS. Since distance measures are easily calculated in planar coordinates (e.g., meters), a UTM coordinate system or equivalent is the most practical for field purposes. A GPS locational reading can then be easily located within the image’s UTM grid. It is of foremost importance to consider ways to share images and maps with the population involved with the research. So, we consider it necessary to prepare duplicates of image and maps to be shared with local collaborators, institutions, and schools. Another key consideration is whether images and maps should be used at all during fieldwork or during particular stages of fieldwork. There are many situations, including but not restricted to land or resource conflicts, when it is not advisable to bring and introduce this type of data during fieldwork. CONCLUDING REMARKS AND CAUTIONARY NOTES

Anthropologists have engaged with and benefited from geospatial analysis in different ways. Whether choosing to develop technical skills in GIS and RS or to work in collaboration with experts, it is important to develop a basic understanding of the process involved in using these tools. It is equally important to develop a critical understanding of the limitations and potentials of different data sets and analytical procedures as applied to different research questions and applications. This chapter provides a basic introduction and describes ways of thinking about the use of these tools in anthropological research. It is an introduction; one should not feel intimidated by the challenges involved in learning the software skills that are needed

14_107-Bernard.indb 623

6/9/14 10:03 AM

14_107-Bernard.indb 624

6/9/14 10:03 AM

Figure 19.5. Multi-temporal images of farm lots and thematic classifications prepared for fieldwork and interview with local farmers. This image was used in combination with a protocol for participatory mapping (focusing on the process of forming a farm since settlement). This image was used in combination with two other images representing the larger region (where landmark references could be located) and the subregion around the farm lot under consideration. Each interviewed farmer was presented with a copy of the image and introduced to basic principles to interpret it.

19: Geospatial Analysis

625

to perform these tasks. As software changes constantly, it is more important to understand the underlying process of geospatial analysis and the basic structure of different kinds of geospatial data. There are always multiple pathways of processing and using geospatial data; the effectiveness of use does not correlate with technical complexity. The wide availability of geospatial data and software (including freeware) opens opportunities for anthropologists to ask new questions and to develop new forms of partnership with local collaborators. Similarly, much of the data familiar to anthropologists and part of previous data collection efforts can be made geospatially explicit and integrated into GIS software. At its basic level, geospatial analysis offers a platform for data organization and data analysis; it also offers a way of thinking about research design and how to contextualize the locations and people we study in different ways (i.e., in time and space dimensions). Geospatial data may help us rethink the very phenomena we study at a local level. As concepts of space, place, site, fieldwork, and community, as well as local and global change in anthropology, geospatial analysis offers us tools to rethink and contribute new ways of analyzing questions of anthropological and societal interest. The potential for integration with a wide range of ethnographic methods and new tools such as network analysis, text analysis, and ethnoecological methods, among others, is promising. As with other tools, the outcome is not based on the level of technical sophistication but on the extent to which we use necessary basic and advanced skills to analyze human social phenomena and our relationship to the environment. NOTES

1. It is beyond the scope of this chapter to discuss specific technical differences between these and other terms. 2. It is beyond the scope of this chapter to explore the different possibilities of using Google Earth and associated products. In brief, Google Earth has an extensive catalogue of satellite images across the world for visualization, including high-resolution images for some locations. It can be used to mark locations and extract their coordinates, to collect training samples for image classification, or to draw lines or areas around features of interest in order to create a map. Professional versions of Google Earth provide capabilities for importing and exporting maps, features, and coordinates. While limited in terms of spatial analysis, Google Earth allows for storage and organization of data associated with particular locations, thus serving as a depository of locational information intended for sharing with a wider community; this may include databases, photographs, and other relevant material. REFERENCES

Albert B., and F.-M. Le Tourneau. 2007. Ethnogeography and resource use among the Yanomami. Current Anthropology 48: 584–92. http://halshs.archives-ouvertes.fr/halshs-00168318 (accessed December 22, 2012). Alvard, M. 1995. Intraspecific prey choice by Amazonian hunters. Current Anthropology 36: 789–818. Aswani, S. 1998. Patterns of marine harvest effort in southwestern New Georgia, Solomon Islands: Resource management or optimal foraging? Ocean and Coastal Management 40: 207–35. Bolstad, P. V., A. Jenks, J. Berkin, and K. Horne, 2005. A comparison of autonomous, WAAS, real-time, and post-processed GPS accuracies in northern forests. Northern Journal of Applied Forestry 22: 5–10.

14_107-Bernard.indb 625

6/9/14 10:03 AM

626

Eduardo S. Brondizio and Tracy Van Holt

Brondizio, E. S. 2006. Landscapes of the past, footprints of the future: Historical ecology and the analysis of land use change in the Amazon. In Time and complexity in historical ecology: Studies in the neotropical lowlands, ed. W. Balée and C. Erikson, 365–405. New York: Columbia University Press. Brondizio, E. S. 2008. The Amazonian caboclo and the Açaí palm: Forest farmers in the global market. New York: New York Botanical Garden Press. Brondizio, E. S., and R. Roy Chowdhury. 2010. Spatio-temporal methodologies in environmental anthropology: Geographic Information Systems, remote sensing, landscape changes and local knowledge. In Environmental social sciences: Methods and research design, ed. I. Vaccaro, E. A. Smith, and S. Aswani, 266–98. Cambridge: Cambridge University Press. Brondizio, E. S., S. D. McCracken, E. F. Moran et al. 2002. The colonist footprint: Toward a conceptual framework of deforestation trajectories among small farmers in frontier Amazônia. In Deforestation and land use in the Amazon, ed. C. Wood and R. Porro, 133–61. Gainesville: University Press of Florida. Brondizio, E. S., and E. F. Moran. 2012. Level-dependent deforestation trajectories in the Amazon: 1970 to 2001. Population and Environment DOI 10.1007/s11111-011-0159-8. Brondízio, E. S., E. F. Moran, P. Mausel, and Y. Wu. 1994. Land use change in the Amazon estuary: Patterns of caboclo settlement and landscape management. Human Ecology 22: 249–78. Chapin, M., Z. Lamb, and B. Threlkeld. 2005. Mapping indigenous lands. Annual Review of Anthropology 34: 619–38. D’Antona, A. O., A. D. Cak, and L. VanWey. 2008. Collecting sketch maps to understand property land use and land cover in large surveys. Field Methods 20: 66–84. Faust, K., B. Entwisle, R. Rindfuss et al. 1999. Spatial arrangement of social and economic networks among villages in Nang Rong District, Thailand. Social Networks 21: 311–37. Fox, J., V. Mishra, R. Rindfuss, and S. Walsh, eds. 2003. People and the environment: Approaches to linking household and community surveys to remote sensing and GIS. Boston: Kluwer Academic Press. Franzen, L., and C. Smith. 2010. Food system access, shopping behavior, and influences on purchasing groceries in adult Hmong living in Minnesota. American Journal of Health Promotion 24: 396–409. Freisthler, B., E. Bruce, and B. Needell. 2007. Understanding the geospatial relationship of neighborhood characteristics and rates of maltreatment for black, Hispanic, and white children. Social Work 52: 7–16. Goodchild, F., and D. G. Janelle. 2004. Thinking spatially in the social sciences. In Spatially integrated social science, ed. M. F. Goodchild and D. G. Janelle, 3–22. New York: Oxford University Press. Green, G., C. Schweik, and J. Randolph. 2005. Retrieving land-cover change information from Landsat satellite images by minimizing other sources of reflectance variability. In Seeing the forest and the trees: Human-environment interactions in forest ecosystems, ed. E. F. Moran and E. Ostrom, 131–60. Cambridge, MA: MIT Press. Guyer, J. I., E. F. Lambin, L. Cliggett et al. 2007. Temporal heterogeneity in the study of African land use: Interdisciplinary collaboration between anthropology, human geography and remote sensing. Human Ecology 35: 3–17. Jensen, J. R. 1996. Introductory digital image processing. A remote sensing perspective, 2nd ed. Upper Saddle River, NJ: Prentice Hall. Krygier, J. B., and D. Wood. 2001. Making maps: A visual guide to map design for GIS, 2nd ed. New York: Guilford.

14_107-Bernard.indb 626

6/9/14 10:03 AM

19: Geospatial Analysis

627

Kusters, K., H. de Foresta, A. Ekadinata, and M. van Noordwijk. 2007. Towards solutions for state vs. local community conflicts over forestland: The impact of formal recognition of user rights in Krui, Sumatra, Indonesia. Human Ecology 35: 427–38. Maher, M. M. 2010. Lining up data in ArcGIS: A guide to map projections. Redlands, CA: ESRI Press. Mok, D., B. Wellman, and J. Carrasco. 2010. Does distance matter in the age of the Internet? Urban Studies 47: 2747–83. National Academy of Sciences (NAS). 2007. Putting people on the map: Protecting confidentiality with linked social-spatial data. Washington: National Academy Press. Nyerges, E. A., and G. M. Green. 2000. The ethnography of landscape: GIS and remote sensing in the study of forest change in West African Guinea savana. American Anthropologist 102: 1–19. (for color figures, see: http://www.cipec.org/publications/nyerges_and_green2000.html) Parry, L., J. Barlow, and C. Peres. 2009. Allocation of hunting effort by Amazonian smallholders: Implications for conserving wildlife in mixed-use landscapes. Biological Conservation 142: 1777–86. Pereira, C., N. Larder, and S. Somerset. 2010. Food acquisition habits in a group of African refugees recently settled in Australia. Health & Place 16: 934–41. Ruf, F. 2011. The myth of complex cocoa agroforests: The case of Ghana. Human Ecology 39: 373–88. Sirén, A. H., and E. S. Brondizio. 2009. Detecting subtle change in small-scale tropical forest shifting cultivation systems: Methodological contributions integrating field and remotelysensed data. Applied Geography 29: 201–11. Turner, M., and P. Hiernaux. 2002. The use of herders’ accounts to map livestock activities across agropastoral landscapes in semi-arid Africa. Landscape Ecology 17: 367–85. Van Holt, T. 2012. Landscape influences on fisher success: Adaptation strategies in closed and open access fisheries in southern Chile. Ecology and Society 17: 28. West, J., E. Blumberg, N. Kelley et al. 2010. Does proximity to retailers influence alcohol and tobacco use among Latino adolescents? Journal of Immigrant Minority Health 12: 626–33. Widayati, A., S. Jones, and B. Carlisle. 2010. Accessibility factors and conservation forest designation affecting rattan cane harvesting in Lambusango Forest, Buton, Indonesia. Human Ecology 38: 731–46. Wilkie, D., and J. T. Finn. 1996. Remote sensing imagery for natural resources monitoring: A guide for first-time users. New York: Columbia University Press. Zeiler, M., and J. Murphy. 2010. Modeling our world: The ESRI guide to geodatabase concepts, 2nd ed. Redlands, CA: ESRI Press. APPENDIX 1: SPATIAL DATA SOURCES: A SAMPLE OF SITES OF INTEREST PLACES TO BEGIN

Both satellite images and other spatially explicit data: • • • • •

Earth Explorer http://Earthexplorer.usgs.gov/ GIS Data Map Studio http://gisdata.usgs.net/ Global Land Cover Facility: http://glcf.umiacs.umd.edu/index.shtml Earth Resources Observation and Science (EROS) Center: http://eros.usgs.gov/ Land Processes Distributed Active Archive Center: https://lpdaac.usgs.gov/lpdaac/ get_data • National Geospatial Program http://www.usgs.gov/ngpo/

14_107-Bernard.indb 627

6/9/14 10:03 AM

628

Eduardo S. Brondizio and Tracy Van Holt

• National Map Seamless Server http://seamless.usgs.gov/index.php • USGS Global Visualization Viewer: http://glovis.usgs.gov/ SPECIFIC SATELLITE IMAGES

• • • • • •

ALOS PALSAR: http://www.eorc.jaxa.jp/ALOS/en/about/palsar.htm LPDACC MODIS and ASTER data: https://lpdaac.usgs.gov/lpdaac/products/ RADARSAT: http://www.radarsat2.info/ SeaWiFS http://oceancolor.gsfc.nasa.gov/ Spot Image, French SPOT Satellite Images http://www.spot.com/ Quickbird/Worldview: http://www.digitalglobe.com/

OTHER RESOURCES

• Earth Observatory—Blue Marble: http://Earthobservatory.nasa.gov/Features/Blue Marble/ • Landsat Program Info: http://landsat.gsfc.nasa.gov/ • NASA: http://www.nasa.gov/home/index.html • NASA Earth Observatory: http://Earthobservatory.nasa.gov/ • National Spatial Data Infrastructure: http://www.fgdc.gov/nsdi/nsdi.html • Visible Earth: http://visibleEarth.nasa.gov/ SPATIALLY EXPLICIT DATASETS/INFORMATION

• Global Spatial Data Infrastructure Association: http://www.gsdi.org/SDILinks • Cartographic Boundary Files—U.S. Census Bureau: http://www.census.gov/geo/www/ cob/index.html • FAO: http://www.fao.org/ • Geo.Data.Gov: http://geo.data.gov/geoportal/catalog/main/home.page • GIS Data Depot: http://data.geocomm.com/ • GIS Resources at IUB: http://www.indiana.edu/~gis/ • Global Land Project: http://www.globallandproject.org/ • Natural Resources Canada: http://www.nrcan.gc.ca/Earth-sciences/products-services/ satellite-photography-imagery/11941 • TIGER/Line Shapefiles—U.S. Census Bureau: http://www.census.gov/geo/www/tiger/ • UNEP/GRID: http://www.grida.no/ • U.S. Fish and Wildlife Service Geospatial Data Inventory: http://www.fws.gov/gis/data/ national/index.html • USDA Geospatial Data Gateway: http://datagateway.nrcs.usda.gov/ • Center for International Earth Science Information Network (CIESIN): http://ciesin .columbia.edu • Brazil spatial and tabular data: http://www.ibge.gov.br/ • INPE (Instituto Nacional de Pesquisas Espaciais, Brasil) http://www.inpe.br • Center for Geographic Analysis at Harvard University, AFricaMap: http://worldmap .harvard.edu/africamap/ • Permanent Committee on GIS Infrastructure for Ásia and the Pacific: http://www .pcgiap.org DIGITAL ELEVATION MODELS

• ASTER Generated DEM: http://asterweb.jpl.nasa.gov/gdem.asp • Hydrosheds: http://hydrosheds.cr.usgs.gov/hydro.php

14_107-Bernard.indb 628

6/9/14 10:03 AM

19: Geospatial Analysis

629

• Shuttle Radar topo/DEM data: http://srtm.usgs.gov/index.php • Shuttle Radar Topography Mission Digital Terrain Elevation Data (DTED) · http:// eros.usgs.gov/#/Find_Data/Products_and_Data_Available/SRTM_DTED • Shuttle Radar Topography Mission Water Body Dataset · http://eros.usgs.gov/#/ Find_Data/Products_and_Data_Available/Shuttle_Radar_Topography_Mission_Wa ter_Body_Dataset GIS/RS SOFTWARE

• • • • • • • • • • • • •

14_107-Bernard.indb 629

ArcGIS: www.esri.com ENVI: http://www.ittvis.com/ProductServices/ENVI.aspx ERDAS IMAGINE/ER Mapper: http://www.erdas.com/ Google Earth, free GIS software: http://Earth.google.com/ GPS TrackMaker: http://www.gpstm.com/ GRASS: http://grass.osgeo.org/ IDRISI: http://www.clarklabs.org/ MapWindow Open Source GIS: http://www.mapwindow.org/ Multispec (Purdue/LARS) http://dynamo.ecn.purdue.edu/~biehl/MultiSpec/ NASA World Wind: http://worldwind.arc.nasa.gov/java/ Open Source GIS http://gislounge.com/open-source-gis-applications/ PCI Geomatics: http://www.pcigeomatics.com/ SPRING GIS Software: http://www.dpi.inpe.br/spring/english/index.html

6/9/14 10:03 AM

14_107-Bernard.indb 630

6/9/14 10:03 AM

CHAPTER 20

Social Network Analysis Christopher McCarty and José Luis Molina

Social network analysis is the study of the patterns of interaction between actors. Actors are typically people, such as residents of a village (Hopkins 2011), schoolchildren (Mouttapa et al. 2004), or customers of a cell-phone company (Onnela et al. 2007), but they can be other things as well. Actors can be collections of people, such as companies (Connelly et al. 2011), countries (Kim and Shin 2002), or primates (Kasper and Voelkl 2009). Social network analysis can be a good way to operationalize social structure on many different levels. In this chapter, we review how social networks have been used, how the data are collected, and what you can do with them. For a more complete review of social network analysis, there are several books and resources available. A good introductory text is Scott’s (2000) Network Analysis. Valente’s (2010) text Social Networks and Health is an accessible resource for any discipline. An online tutorial by Hanneman and Riddle (2005) uses the program UCINET (Borgatti et al. 2002), the most widely used social network analysis program, to introduce concepts and measures. Analyzing Social Networks, by Borgatti et al. (2013), provides a good introduction to the analysis of network data with UCinet and other related programs. There is a more in-depth treatment of social network measures in Wasserman and Faust (1994). Kadushin’s book Understanding Social Networks (2012) provides a comprehensive view of the way social network analysis has been applied. Finally, Jackson’s (2008) book Social and Economic Networks applies social network concepts to economic theory and methods, such as game theory. HISTORICAL BACKGROUND AND THEORY

Today, most people think of social networks as social networking sites or services such as Facebook and Twitter. The idea of social networking sites is indeed based on social network research conducted by Stanley Milgram (1967) about the “small world problem.” Milgram and his colleagues showed that completed chains connecting two randomly selected people in the United States contained, on average, about fiveplus people, or six degrees of separation. The first commercial social networking site, sixdegrees.com, was launched in 1997; although it closed a few years later, it started an enormous commercial industry that has changed the way we communicate. Apart from this commercial meaning, social networks and social network analysis have a long history. In his account of the development of the field, Freeman (2004) shows how the roots of social network analysis can be traced simultaneously to the work of Jacob Moreno (1934) and anthropologists from the Manchester School. 631

14_107-Bernard.indb 631

6/9/14 10:03 AM

632

Christopher McCarty and José Luis Molina

Moreno represented social interactions as dots for actors and lines for ties between actors, a procedure he called Sociometry. His work influenced British anthropologists of the 1940s and 1950s. Until then, the structural–functional paradigm had served for the study of single, spatially compact, and small-scale groups but something else was needed to understand new patterns of social interaction among tribal people living in urban settings under colonial rule (Mitchell 1972; Wolfe 1978). In 1947, Max Gluckman, former director of the Rhodes-Livingstone Institute in Northern Rhodesia, started a series of seminars at the University of Manchester in Manchester, England. Gluckman was interested in explaining conflict and change in Africa, taking into account these new social relationships among individuals from different cultures that emerged from British rule. During the development of these seminars, social networks ideas represented a way to bring individual agency into the analysis, representing a source of change and intercultural variation through competition and individual exchange strategies in complex societies. Among the most influential scholars who attended the seminars and applied the idea of networks to their research were J. Clyde Mitchell (1969) (on social structure in central Africa), Elizabeth Bott (1957) (on marriage and kinship in London), John A. Barnes (1954) (on decision-making in Norway), Bruce Kapferer (1972) (on conflict in Zambian factories), Arnold L. Epstein (1969) (on norms and social situations in Zambia), Jeremy Boissevain (1974) (on personal networks in Malta), and Siegfried F. Nadel (1957) (on theory of social structure). Barnes was the first to employ the term social network to characterize the set of relations among institutions in a small fishing village in Norway. These relations bridged different groupings and institutional settings on a personal level. This was similar to Gluckman’s idea of the existence of bridging ties between traditional African cultures and British colonial culture. Mitchell is credited with formalizing social network theory in a way that also suggested how it should be studied. Among other network ideas, he suggested the term ego-centered networks to describe the notion of recording data on the links surrounding a focal individual. In his view, these ego-centered networks combine to form a total network. The Manchester tradition endured for another two decades, but, among other reasons, the lack of methods (particularly, the lack of computers) for dealing with the large amount of network data collected during fieldwork stopped its development. The network approach to the study of social structure developed by Nadel in the Manchester seminars was adopted by American sociologists like Harrison White and others in a series of seminal articles (Boorman et al. 1976; Lorrain and White 1971; White et al. 1976). The University of Chicago and Harvard University began to develop the methods and substantive applications of whole-network analysis. In 1978, Linton Freeman started the journal Social Networks, and in 1977, Barry Wellman organized the International Network for Social Network Analysis (INSNA), the organization at the center of the interdisciplinary field that social network research is today. The first theoretical concern that motivated social network thinking was how to identify and assess the effect of emergent social structures on individual behavior (and norm development). These structures were studied at the egocentric level (i.e., the spread of gossip [Epstein 1969]) or at the whole network level (i.e., the mobilization

14_107-Bernard.indb 632

6/9/14 10:03 AM

20: Social Network Analysis

633

of supporters in a work setting [Boissevain 1972]). This novel perspective allowed researchers to study actual groupings of individuals in different roles instead of focusing solely on normative groups (kinship groups, territorial groups, age groups, etc.; see Srinivas and Béteille 1964). The second theoretical question addressed by anthropologists was the correspondence between cognitive and behavioral data. This task was undertaken by anthropologists influenced by American cognitive anthropology, trained in formal methods for clustering classification patterns and identifying intracultural individual variation (Johnson 1994). These contributions from cognitive anthropology were accelerated by Borgatti’s (1992) development of ANTHROPAC, a set of data collection and analysis routines that made this type of data analysis accessible to a much wider audience. Another contribution in this field was a series of articles about the accuracy of social network data (see, e.g., Bernard et al. 1984; Killworth and Bernard 1976). They found that cognitive information about network relations showed systematic bias and differed from behavioral data. The wealth of network behavioral data available in social networking systems like Facebook and Twitter and the relation of those data to user reports of behavior make this research of special interest for current scholars interested in social media (see, e.g., Gruzd et al. 2011; Lewis et al. 2008). The third theoretical concern came from kinship studies, one of the core subjects in classical anthropology. The question was whether kinship normative systems existed in practice, and whether there were empirically based models that accounted for effective kinship ties. Per Hage studied ethnographic and historical cases of kinship. His collaboration with Frank Harary, a mathematician interested in developing graph theoretical applications for the social sciences, has been influential in the development of social network analysis (Hage and Harary 1997). Douglas White and his students have produced a large corpus of knowledge about the computational representation and analysis of large amounts of kinship data (White and Johansen 2005). They found that it was possible to detect basic structures of relinking intergroup marriage that maintained overall cohesion despite the appearance of fission processes through time. The development of the computer program Pajek by Vlado Batagelj and his colleagues made it possible to process these large data sets to test these models (White et al. 1999). Work in France from the École des Hautes Études en Sciences Sociales has extended this work (Hamberger 2011; Houseman and White 1996). Newly available data on indigenous societies, mainly from Brazil, is fostering the development of network applications to kinship (Barbosa de Almeida and Williams 2010). Another issue that guided anthropological research on networks is how macro (societal) and micro (personal) networks are related. Bernard and colleagues (Bernard et al. 1988; Bernard, Killworth et al. 1990) conducted a series of studies called the Reverse Small World using a personal network approach. These studies extended the work of Milgram, showing that the links in the chains were, in fact, not random and that people’s choice of links was typically guided by those links’ location and occupation. In a related line of research, these researchers were able to estimate the personal network size for each respondent in a population survey. This proved important in estimating the size of hard-to-count populations, like people living with HIV, the homeless, women who had been raped, civilians killed in wars, and so on (Bernard et al. 2010).

14_107-Bernard.indb 633

6/9/14 10:03 AM

634

Christopher McCarty and José Luis Molina

Personal network research has been expanded by McCarty and his colleagues with the development of EgoNet (http://soucerforge.net/egonet), an open source program for collecting, analyzing, and visualizing personal network data (McCarty et al. 2007). In addition, the growing field of transnational studies (Vertovec 2009) and the concepts of “transnational social field” and “transnational social space” are built on ideas from social network analysis (see Molina et al. 2012 for a review). These fields or spaces are emergent cross-border social structures that allow migrants to overcome the hurdles of making a living in a destination country. Finally, the question of factors that account for intracultural variation of ecological knowledge has inspired research that applies social network methods (Berkes and Folke 1998; Ellen et al. 2000). Examples include the study of medicinal plants (Hopkins 2011), the conservation of traditional varieties of plants through the exchange of seeds (Calvet et al. 2012) and the relation of knowledge to status and health in indigenous societies (Reyes-García et al. 2008). These studies have shown that the structural position of an individual in a network of knowledge and/or exchange is related to local ecological knowledge and to actual exchange. Apart from these lines of research, there have been contributions from individual scholars. Larissa Adler-Lomnitz (1975), for example, applied social network analysis to a wide range of ethnographic cases in Chile, Mexico, Russia, and Hungary. Thomas Schweizer (1997) contributed to the development of the theoretical concept of embeddedness based on comparative ethnographies. Robin Dunbar, a psychologist, has done research in the area of personal network size and how it relates to physiological constraints of memory. Dunbar (2008) suggests that the ability of people to know other people is constrained by time and resources as well as the organization of family, friends, and acquaintances in memory, to an average of 150 people. He points to the tendency of various human social groups to fission at around the same size (see Gonçalves et al. 2011 for a review). This is similar to a finding from Bernard and Killworth (1973, 1979). They measured the average number of ties (7.27) between members of a scientific research vessel and developed a model predicting that networks would tend to divide into subgroups when they reach a size greater than 140. The findings from Dunbar and from Bernard and Killworth suggest that large networks are likely best described as collections of smaller networks connected by one or two brokers. Much of the classical work in social network analysis is from sociologists. In 1961, James S. Coleman published The Adolescent Society, an influential book about the divide between adolescent and adult worlds in America. Coleman pointed out the structure of peer interactions as one of the sources of normative reinforcement (against the institution, in this case). Beginning in 1964, Edward Laumann conducted a series of surveys as part of the Detroit Area Study (Laumann 1973). Laumann included questions stemming from his interest in social distance. Laumann was the first to collect personal network data about a small number of network members from a large and representative sample of respondents using a standard survey approach, a method that dominated personal network analysis for decades. In 1973, Mark Granovetter developed the idea of the importance of weak ties and social structure. In his research, people often found out about job opportunities through people they knew. What was unexpected was that these opportunities typically came through acquaintances rather

14_107-Bernard.indb 634

6/9/14 10:03 AM

20: Social Network Analysis

635

than people the respondents knew well (strong ties). This was because acquaintances had access to information about job openings in groups that the close and highly connected network members did not have. Around the same time as Granovetter’s research, Barry Wellman (1979) conducted a large personal network study in a suburb of Toronto using methods similar to Laumann’s. Wellman and his colleagues discovered that communities still existed through sometimes distant network linkages, contrary to common wisdom that people in urban situations lose their sense of community. In To Dwell among Friends, Claude Fischer (1982) explained how personal networks operated in all aspects of life, including work, religion, and attitude formation. Fischer was one of the first to introduce the idea of the personal network life cycle; that is, the ways network composition and structure change as we age. Charles Kadushin (1968) used the personal network approach to construct whole networks of invisible groups. He used the technique of snowball sampling to operationalize Georg Simmel’s (1955) concept of social circles. Social circles are groups of people who share characteristics in common and are therefore influenced by each other either directly or indirectly, although they may not be aware of that influence. Scott Feld was also interested in social circles, but more so in how they actually formed. He introduced the notion of social foci, social situations that result in nonrandom interaction between people. Feld (1981) used variations of this theory to explain the attributes of personal networks and how those attributes are perceived by individuals. Ronald Burt (2004) developed measures of social capital based on the presence or absence of certain types of social ties. In an important work examining the egocentric network of managers in business settings, Burt developed the concept of the structural hole—that is, an opportunity for individuals to act as bridges in otherwise disconnected groups within an organization. Burt’s ideas about how networks stimulate creativity are also applicable outside organizational settings. Thomas Valente (2010) used variability of the characteristics of personal networks to explain contraceptive use among women in Cameroon (1997). His research falls under the umbrella of the diffusion of innovations, another large arena of network research pioneered by Coleman (Coleman et al. 1966) and developed by Everett Rogers and his students (Rice 1994; Richards and Barnett 1993; Rogers 2003). More recently, Christakis and Fowler (2009) analyzed network data from the Framingham heart study and suggested that conditions such as obesity are influenced by social networks. Although this conclusion has been strongly criticized (Aral et al. 2009; Rohila Shalizi and Thomas 2011; VanderWeele 2011), the contributions coming from public health are important and influential. Studies relating individual well-being and social networks include Cohen and Wills (1985), Berkman and Syme (1979), and Kadushin (1982). Studies on the role of network structures on the spread of infectious disease, especially AIDS, include work by Klovdahl (1985) and Latkin et al. (2003). The next sections are devoted to explaining both whole network and personal network analysis. In general, whole network analysis is especially useful when indirect links are supposed to play an important role in the expected outcome or when the behavior under study is expected to be at least partly explained by some institutional context (e.g., at the workplace or a retirement center). By contrast, personal network

14_107-Bernard.indb 635

6/9/14 10:03 AM

636

Christopher McCarty and José Luis Molina

analysis is better suited for studies in which a representative sample of the population is needed or when the whole complex of institutional contexts in which ego is embedded is supposed to be important for a given outcome (like smoking or evincing some ethnic identity). Both types of analysis can be combined as well, as will be explained later. WHOLE NETWORK DATA COLLECTION

Social network analysis is often classified into two types, whole networks and egocentered networks. Whole networks focus on the interactions of actors within a bounded space. Boundary definition—who will be included and who will not—is an important aspect of whole network analysis. Boundaries may be geographic (households in a village) or social (members of an organization). Snowball sampling, where a sample of seed respondents nominate later respondents, is often used as a way to define whole network boundaries when the network to be studied is known (such as a gang) but where network members are not limited to a known geography and there are no lists of network members. The objective of a whole network study is to collect data about each member of the bounded group and their interactions with all other members. Interactions can be measured in many ways. A common approach is to ask members about their interactions with other members using a scale. For example, respondents might report their closeness to every other member on a scale of 0 to 5, where 0 implies they are not at all close and 5 implies they are as close as possible (see Figure 20.1). The first cell intersecting David with David is a 5, reflecting the fact that David is as close to himself as he can be. This is the case for all cells on the diagonal, and these values are typically ignored in the analysis of social network data. The second cell in the first column represents Faith’s assessment of her closeness with David, which she evaluated as a 1. The first cell in the second column is David’s assessment of his relationship with Faith, which he evaluated as a 2. Faith and David do not agree on their relationship. The cells in the bottom half of the matrix (those numbers below the diagonal) are different from those in the upper half. This means that the matrix is asymmetric. If the upper and lower half were the same, the matrix would be symmetric. Symmetry is an important concept in the analysis of network data and must be considered in designing a data collection and analysis strategy. Many network ties are not reciprocal, such as closeness (in Figure 20.1). Exchange of resources can be asymmetrical (A may give more to B than B does to A) or symmetrical (if the question is “Do A and B exchange resources at all?”).

Figure 20.1. Matrix representing interaction between all pairs of members of a class.

14_107-Bernard.indb 636

6/9/14 10:03 AM

20: Social Network Analysis

637

Table 20.1. Example Tie Questions How close are you to X? How likely are you to borrow money from X? How much do you rely on X for information? How much labor do you exchange with X? How much meat do you exchange with X? Do you exchange tools with X? How long have you known X? How often do you interact with X?

In the example above, only one question represents the tie between actors. Social networks may be represented by many tie definitions, each one potentially resulting in a different network structure. Table 20.1 provides a list of commonly used questions for tie definition in anthropological studies. Each definition is not one question, but one for every member of the group a respondent is evaluating, creating the potential for excessive respondent burden. Anthropologists should be aware of the potential for low variability of responses to some tie questions, such as whether members of a small village know each other. Although the last two questions in Table 20.1 (referring to duration and frequency) may be appropriate in some situations, they are not necessarily correlated. For example, the people someone interacts with frequently may not be those they have known the longest. Similarly, the people someone sees most often may not be the closest to them emotionally. Tie questions must be designed carefully to reflect the network interaction that aligns with the theory and specific circumstances of an individual study. Ideally, one would collect responses from every member of the whole network (all villagers or all members of an organization) since respondents may not agree about their patterns of interaction. However, as networks become larger, it becomes difficult to get responses from all network members. The need to measure a network within a fixed period of time may not be possible without assistance. For example, it may be possible to collect interaction data within a village across several years, but during that time the structure of interactions can change—in which case, the tie data collected earlier do not reflect the structure when the data are analyzed. One of the most widely used approaches for collecting whole network data is the roster method where every respondent evaluates their interactions with every other member on a list. If the list is relatively small, say under 100 members, this is the most accurate and recommended approach. Showing each member of a group the names of all other members of the group triggers information retrieval about relationships. Another approach is called the survey method in which respondents are asked to list the members they interact with. Respondents might be asked to list five households they rely on for water during a drought. This approach has the disadvantage that respondents may forget subtle connections within the whole network, but it is much faster and less burdensome than the roster method. This method is recommended when the network is large. Some researchers do not limit the number of nominations when using the survey method, instead allowing respondents to choose the number of network members. While it is almost certain that the number of households one can trust with child care varies from one household to another, it is not certain that the

14_107-Bernard.indb 637

6/9/14 10:03 AM

638

Christopher McCarty and José Luis Molina

number nominated actually reflects that. Some respondents may list few members because they are uncooperative or simply forget. The researcher has to balance these two possibilities. When the boundary definition results in a very large network, it may not constitute a meaningful one. For example, the city of Mumbai, India, may not comprise a single network in any meaningful way. There is some tolerance for missing data in whole network analysis, particularly when the data are asymmetric. While a single respondent may refuse to be interviewed, other respondents will report on their interactions with the missing respondent. Research suggests that for some structural measures, as little as 50% of responses is sufficient to capture network structure (Costenbader and Valente 2003). As mentioned earlier, the network being studied need not be of people; it can be a collection of people. Anthropologists conducting a community study may want the network to reflect the interactions among community members, particularly when age and gender roles are considerations. Other times, it may be more appropriate to use households as the unit of interaction, as when studying networks of exchange. Even when the household is the unit of analysis, the interviewer will have to talk to someone. Choosing who to talk to about the interactions with the household may have consequences, as any one person may not know about all interactions and those they know about may be biased. One approach is to randomly select respondents from the household, an approach often used in survey research. This will eliminate systematic bias, but it is possible that the randomly selected respondent will not know about the actual interaction of all members. Another approach is to select the person responding for the household who is believed to be the most knowledgeable about certain types of interaction. Still another approach is to try to achieve a consensus opinion among a group of household members characterizing the interactions with other households. This approach may be important when there are highly varying types and amounts of interaction among household members that a single member could not report accurately. Sometimes network data can be collected by observing interactions rather than asking people to report on their interactions. Studies of interaction among preschoolers rely on trained observers who carefully record their interactions (Martin et al. 2005). This approach works when there are a relatively small number of actors to observe or when there are many observers. When using an observation approach, coordination of those observations is important to maintain efficiency. Another approach is to sample times of day and specific locations to observe interactions. Over time, recurring interactions should be observed more frequently. Some data on interaction may exist in an archive. For example, in office environments, email and phone records may be used as measures of interaction. In villages, marriage records may indicate interaction between households. Historians have used records of financial transactions to measure networks of exchange over time (Padgett and Ansell 1993). And measuring the width of roads and paths between households is a way to operationalize interactions. Existing data should always be considered as an alternative or complement to data that require direct interviewing of respondents. The data depicted in Figure 20.1 are referred to as one-mode data, meaning the labels (mode) in the rows are the same as the labels in the columns. Another convenient way

14_107-Bernard.indb 638

6/9/14 10:03 AM

20: Social Network Analysis

639

Figure 20.2. Two-mode data of funeral attendance.

to use archival data is with two-mode data collection. With two-mode data, the labels in the rows are not the same as the labels in the columns. Typically, the rows represent the actors and the columns represent some event linking the actors. Figure 20.2 shows a hypothetical two-mode network of a set of villagers and the funerals they attend. Row 1 indicates that Buseje attended funerals 5 and 6, while Alile attended funerals 1 and 2. Assuming that co-attendance implies some form of social interaction, if two people attend the same funerals they may interact. The data in Figure 20.2 can be transformed into a one-mode network, as seen in Figure 20.3. This one-mode network is symmetric, meaning the upper and lower halves are the same. The numbers down the diagonal represent the number of funerals the person attended; Buseje attended two funerals and Chisulo attended three. The numbers in the off-diagonal indicate the number of funerals the actors attended in common. Buseje and Chisulo did not attend any of the same funerals, while Buseje and Dziko attended two of the same funerals. A two-mode network approach is also useful for collecting data directly from respondents. For example, rather than observing funerals, an interviewer could ask respondents to indicate if they had attended each of a set of funerals. The data collection strategy will also depend on data entry constraints from analysis software. Network data are typically analyzed as a matrix, such as those depicted in Figures 20.1 and 20.3. Although it is possible to type data into a matrix format if the number of actors is small, it will be time consuming and error prone for larger networks. It is also not necessary. Most social network programs, such as UCINET and Pajek,1 accept data collected in different ways. Two of the most frequently used methods are an edge list and a node list. Social network analysis borrowed many concepts from graph theory. In the jargon of graph theory, the actors are nodes and the ties between actors are the edges. As an alterna-

Figure 20.3. One-mode affiliation from two-mode data.

14_107_20_McCarty.indd 639

6/9/14 11:19 AM

640

Christopher McCarty and José Luis Molina

Figure 20.4a. Example of edge list.

Figure 20.4b. Matrix produced from edge list.

tive to filling in all the cells in a matrix, one can list all the ties (edges) that are recorded or observed. Figure 20.4a shows an edge list and Figure 20.4b shows the resulting matrix. A node list is the best approach when entering data collected using the survey method where respondents list actors to whom they are tied. Figure 20.5a shows what a node list might look like. The first name in each row is the respondent and the names following are the ones they list. Figure 20.5b shows the matrix produced from the node list.

Figure 20.5a. Example of node list.

Figure 20.5b. Matrix produced from node list.

14_107-Bernard.indb 640

6/9/14 10:03 AM

20: Social Network Analysis

641

NETWORK VISUALIZATION

One of the first things researchers do with social network data is to create a network visualization. The data from Figure 20.1 are visualized in Figure 20.6 using a program called Netdraw.2 Visualization gives a sense of the overall network structure. In Figure 20.6, the dots (nodes) represent the students and the lines indicate how they evaluated each other in terms of closeness. The thickness of the line indicates the level of closeness and the arrow (present or absent on each end of the line) indicates if the tie was reciprocated. For these data to be visualized, the software must decide whether a line will be drawn. The data from Figure 20.1 are asymmetric, meaning that the students did not necessarily agree about how they were tied to one another. The default for most visualization programs is to draw a line that represents the maximum evaluation. For example, David rated Faith a 2, while she rated him a 1. The program must show a line or not, so it defaults to the maximum of 2. David rated Mark a 1, but Mark rated David a 0. The program inserted a line. Typically, network analysts will use social network software to symmetrize the data to control how those choices are made. Many social network measures are graph based and require dichotomized (binary) data. If the data are not already dichotomous, the software will typically default to any value greater than zero. There are many ways to visualize the connections between actors. In Figure 20.6, the data are visualized in two dimensions using Netdraw (Borgatti 2002). Other programs such as Mage (http://kinemage.biochem.duke.edu) can create three-dimensional

Figure 20.6. Visualization of class network.

14_107-Bernard.indb 641

6/9/14 10:03 AM

642

Christopher McCarty and José Luis Molina

visualizations that may be more appropriate for some types of data. For example, if there were three primary drivers of social interaction, such as sex, age, and income, the two-dimensional visualization would only show two. Most social network visualizations are in two dimensions simply because two dimensions are easier to visualize and to publish. One can choose from several techniques to position the actors within the visualization space. Techniques such as circle graphs do not focus on the position of the node at all, but rather the clustering of lines (Freeman 1999). Other techniques are far more complicated. Another consideration is whether the visualization is based on valued data or dichotomized data. Statistical procedures such as multidimensional scaling position the nodes using coordinates that incorporate the value of the tie between the nodes. Figure 20.6 uses a procedure called a spring embedder that relies on the presence or absence of a tie (dichotomous data). A spring embedder selects a node at random and places it in the middle of the screen. It selects a second node at random; if it is tied to the first node, it puts them next to each other; if they are not tied, it pushes them apart. This procedure continues until all nodes are optimally pulled toward those to which they are connected and pushed apart from those they are not. Some programs, like Netdraw, take nodes that are not tied to any other node (isolates) and stack them in one place on the graph, such as the upper left. Other programs randomly place isolates in the two-dimensional space. All programs randomly place disconnected groups (i.e., components) in the two-dimensional space. When interpreting network visualizations, the length of a line is typically not important. If a line exists between two nodes, it means they are tied. A longer line means that the two nodes (the dyad) tend to be less tied to the same people than two nodes connected by a shorter line. One must be careful about interpreting the position of one component relative to another component. If two groups do not have any ties, then their placement in the two-dimensional space is random. Network visualizations often incorporate the attributes of the actors. For example, you can use color, labels, size, or shape to show how the network structure represented by the visualization can be explained. Some attribute data will come from data collected about each actor. Other attributes, such as network measures, can be calculated using software. Many researchers use network visualizations as a cue to talk to respondents about their group. Network visualizations tend to be intuitive to respondents. Network structures can suggest questions such as why some actors are isolated from others, why some people tend to be very central, and why some groups are separated. These questions would be difficult to frame or to explain to respondents without the aid of network visualization (Molina et al. 2014). SOCIAL NETWORK MEASURES

Although visualization is a great way to get a feel for your data, there is no substitute for objective measures. Social network measures can be broken into four types as depicted in the quadrants shown in Table 20.2. The rows refer to whether the network measure is a single number representing the group or an individual measure for each node within the network. Network density,

14_107-Bernard.indb 642

6/9/14 10:03 AM

20: Social Network Analysis

643

Table 20.2. Types of Social Network Measures

Group Level

Node Level

Graph-Based

Statistical

Centralization Number of Components Number of Isolates Density Centrality

Quadratic Assignment Procedure

Cluster Analysis Multidimensional Scaling

one of the most widely used group level measures, is the percent of all ties that exist out of all those that could exist. It is represented as a proportion between 0 and 1. Node level measures say something about the position of the actor within the entire network. The most widely used node level measure is centrality. While there are many centrality measures, the most commonly used are degree and betweenness centrality. Degree centrality for a node is the number of other nodes it is tied to directly. In the class network depicted in Figure 20.6 the degree centrality for Antonio is 3 because he is tied to three people. Betweenness centrality is calculated by determining the shortest path (the geodesic) between every pair of nodes. For a given node the betweenness is the number of shortest paths they lie on. A node that is on a lot of shortest paths may be in a brokering position. In Figure 20.6, Amber is the most between-central node. Centrality measures are often exported from network software and used as covariates in statistical models. For example, betweennness centrality could be used to explain the variability in a measure of social mobility. Actors with high centrality do not necessarily use that quality to their advantage. Actors can be central for many reasons. They may be important or they may be central because they are exploited by others. They could even become central if they were mutually disliked by most people in the network. When interpreting the structure of a group, it is valuable to consider why some people are marginalized or why others are embedded in closely knit groups. Many social network measures are based on dichotomous data. However, there are many statistical measures that can be used for analyzing valued matrices. For example, cluster analysis uses valued matrices to aggregate nodes into subgroups. Finding subgroups is often an objective of social network analysts, and statistical procedures are useful for doing that. Here is a list of commonly used social network measures with a brief description: Degree Centrality—An alter is highly degree-central to the extent he or she is directly connected to many other alters. Degree Centralization—A measure of the extent to which the network is dominated by a few degree central nodes. Closeness Centrality—An alter is highly close-central if he or she is connected by short paths to most other alters. Closeness Centralization—A measure of the extent to which the network is dominated by a few close central nodes. Betweenness Centrality—A node is highly between-central to the extent he or she lies on many geodesics (shortest paths) between alters. Alters that are between central are brokers.

14_107-Bernard.indb 643

6/9/14 10:03 AM

644

Christopher McCarty and José Luis Molina

Betweenness Centralization—A measure of the extent to which the network is dominated by a few between central nodes. Components—A set of alters who are connected to one another directly or indirectly. Isolates—A node unconnected to any other node.

Another category of measures focuses on the tie (edge) rather than the node. These are important when the research question has more to do with the relationship than with individual actors. For example, a study of exchange networks within a community could focus more on the structural properties of a tie between two actors, such as edge betweenness, rather than the structural position of either actor. SOCIAL NETWORK MODELS

The measures described in the previous section result in single numbers to represent the network or the nodes in the network. These measures, particularly the node level measures, can be used in statistical models to predict, or be predicted by, other variables. A common strategy is to calculate the degree or betweenness centrality of each node and to use that in a model to explain the variability of a dependent variable. A network model is in contrast to a statistical model. Network models determine the fundamental relationships that underlie network structures. In social network analysis, there are models that rely on the distribution of different types of triads within the network. A triad is a set of three nodes, and these three nodes can exhibit any of 16 different types of relationships depending on whether the nodes are tied, and, if they are tied, whether the relationships are reciprocated. An examination of the triads in a network is called a triad census. Some network analysts consider triads to be the building blocks of social structure (Johnsen 1985; Killworth and Bernard 1979; Wasserman and Faust 1994). A level up from triad analyses are blockmodels. Blocks are sets of interacting nodes that are mostly tied relative to the potential ties between blocks. Social network software can be used to identify the optimal number of blocks, or subsets, within a network (see de Nooy et al. 2005 for the software Pajek). Another set of models are p1, p2, and exponential random graph models (ERGM, see Robins et al. 2007; Wasserman and Pattison 1996; Wasserman and Robins 2005). These are used to understand what explains network structure—the attributes of the nodes or the structural position of the nodes. Network models are particularly useful in analyzing longitudinal networks— that is, networks over time. Network software such as RSiena (http://www.stats.ox.ac .uk/~snijders/siena/) and Statnet (Snijders 2005, Snijders et al. 2010) are useful for determining whether changes in the network structure over time are due to attributes of nodes or the probability of structural change due to their starting structural position. Longitudinal social networks, though difficult to collect, are increasingly used to answer questions about what makes networks change and what those changes predict. Many modeling algorithms are available for free in the R development environment. EGO-CENTERED NETWORKS

Ego-centered networks, also called personal networks, examine the social context surrounding a focal respondent (egocentric networks refer to the analysis for every vertex or node of the adjacent nodes’ structure and composition within a whole network).

14_107-Bernard.indb 644

6/9/14 10:03 AM

20: Social Network Analysis

645

In contrast to whole networks, ego-centered networks are not typically bounded geographically or socially. The purpose of an ego-centered study is to operationalize individual social context as a set of variables that are used to predict or say something about the focal respondent. Respondents to an ego-centered network study are selected in much the same way as respondents would be selected for a survey. The selection of respondents, called egos, is usually dictated by the group to be studied. The group might consist of people who live in a particular place, such as a village, or certain types of people, such as the elderly. The objective is to say something about the group from which the respondents have been selected, and the rules governing survey sampling, such as targeting margins of error, will apply. Programs such as Egonet (http://sourceforge.net/projects/egonet/) are designed to collect and analyze ego-centered network data. There are generally four parts to an ego-centered network study. First are the questions one asks the respondent (ego) about themselves. These are typically the dependent variables, such as the number of children a respondent has or whether they are a smoker, and/or the control variables, such as age and ethnicity. Some network studies identify spurious relationships between social network variables and the dependent variables. That is, they may be related through a third variable, which is often easier to measure. Without control variables, you cannot identify the unique explanatory power of the network variables. The second step is to elicit the names of network members, called alters, from the respondent. Unlike whole network studies where the network members are typically known, with an ego-centered study the respondents list their network members. The questions used to get those names are called name generators. Name generators provide the boundary definition for the ego-centered network. There are many things to consider in designing name generators. The first is that the respondent is the only one who knows who is actually in their network, and unless alters are contacted (called alter-chasing) only respondents know if the list is correct. This introduces the possibility of error in the elicitation from one respondent to the next. The name generator should be worded carefully so this variability is minimized across respondents. Table 20.3 shows some name generators that have been used in the past. These name generators could result in very different lists of names. The number of people you have ever known will be a much larger list than those you currently talk to about important matters. Both will yield valid lists of names depending on the focus of the research. Another consideration is the bias associated with the way respondents recall names. People do not store names in memory in exactly the same way. This introduces the potential for bias in the way names are recalled. When given a very general name generator, most people will start listing their immediate family, although that will not be true of all respondents. Respondents may continue with extended family, or work colleagues, those they have seen recently, or even names that are similar to names that Table 20.3. Some Name Generators People you talk to about important matters People you can rely on for help with a personal problem People who you know by sight or by name that you have had contact with in the past two years People you have ever known

14_107-Bernard.indb 645

6/9/14 10:03 AM

646

Christopher McCarty and José Luis Molina

have already been listed. Elicitation bias is unavoidable. It is therefore advisable to word the name generator so that all respondents have the same bias. For example, one could ask respondents to list people beginning with those they have seen most recently. While this creates a biased network, at least all respondents have the same bias. Another design question is how many alters to elicit. This will be determined in part by the research topic, but also must be balanced by the potential respondent burden. Some researchers fix the number of alters at, say 20 or 30, while others allow respondents to have variable numbers of alters, using the number they list as a proxy for network size. There are other methods to measure network size, such as the network scale-up method (see Bernard et al. 2010). There are many reasons someone may stop listing alters—they may have listed all the alters, they may be fatigued, they may not remember more alters, or they may be uncooperative. Since it is not known why a respondent stops listing alters, it may be better to fix the number of alters for all respondents, requiring that they all expend the same effort. The next stage after alter elicitation are the questions about each alter; these are called name interpreters. The data from these questions will result in compositional variables that describe what types of people the social context is composed of. For example, it is common to ask respondents whether an alter is male or female since that may be a key predictor of a dependent variable. For each respondent, this will result in a compositional variable indicating the proportion of their network that is male and female. Continuous variables like age will result in an average age of each respondent’s alters. The questions are repeated for each alter listed. Ten questions about each alter and 20 alters will generate 200 questions. One must balance the number of alters and the number of name interpreters. There are some things respondents can be expected to know about their alters and others that are not reasonable to ask. When selecting name interpreters try to use questions that respondents could answer about all of their alters. Questions can often be modified from unreasonable to reasonable. For example, a respondent may not know the specific age of their friend but could estimate it within an age interval of 10 years. The last stage in the data collection process for an ego-centered network is the alter tie evaluation, which will result in a set of structural variables. In this stage, respondents evaluate whether their alters are tied in some way. The scale for the evaluation of these relations is narrower than with a whole network study since respondents are reporting on the way their network members are tied, not on anything about themselves. It is typical to use a three-point scale for tie evaluation—the two alters are not tied, they might be tied, or they are definitely tied. Alter tie evaluations are usually symmetric, meaning respondents are not asked to evaluate whether a relationship between two alters is asymmetric. Since the ties are symmetric, the respondent only has to respond to (N*(N−1))/2 pairs. For 10, alters they must evaluate (10*(10−1))/2 or 45 pairs. Figure 20.7 shows how the number of tie evaluations increases geometrically as the number of alters increases. Some argue that alter tie evaluations may not be accurate citing studies showing that respondents did not accurately report the interactions among members of a whole network to which they belonged (Bernard et al. 1984). However, there is a difference between reporting on ties among co-members of a whole network, all of whom the

14_107-Bernard.indb 646

6/9/14 10:03 AM

20: Social Network Analysis

647

Figure 20.7. Number of alter pair evaluations as alters are added.

respondent may not know, and reporting on the members of one’s own personal network, whom respondents know, by definition. Those who conduct ego-centered studies often contend that respondents can accurately report about the interactions among members of their networks (see McCarty et al. 2007). And most social network analysts believe the tie evaluations at least result in cognitive social structures; that is the social context the respondent believes exists, and that this has an effect on attitudes and behaviors (Krackhardt 1990). There are alternatives to having respondents evaluate each tie. McCarty and Killworth (2007) showed that eliciting a larger set of alters, and then sampling at least 25 of those alters, produces similar structural properties as asking for all tie evaluations among a larger set of alters. Some programs, such as Vennmaker (http://www.venn maker.com/en/), use a visual interface to collect network alters and ties. The advantage of a visual interface is that respondents may find it to be a more engaging format. With Vennmaker, the respondent’s network is represented as a set of concentric circles with the respondent in the center. The respondent drags icons representing men and women into the appropriate spot and connects them using icons for strong and weak ties. While the visual interface is appealing, it may introduce an experiment effect as respondents connect mostly those who are grouped closely together. After data collection, there are two ways in which personal network data are used. It is typical to use a visualization of the personal network as a cue to interview the respondent about the focus of the research (McCarty et al. 2007). This is illustrated with the personal network visualization in Figure 20.8 depicting the network of a second-generation West African in Barcelona, Spain. The respondent lived in the

14_107-Bernard.indb 647

6/9/14 10:03 AM

648

Christopher McCarty and José Luis Molina

Figure 20.8. Personal network visualizations of a second-generation West African.

same apartment with her Muslim family. The nodes representing the network alters are labeled by the country of residence, colored by the perceived skin color, sized by perceived closeness, and shaped by whether or not they smoke (squares are smokers and circles are non-smokers). Using these visualizations, the authors were able to ask the informant about the large number of isolates and the unconnected dyad. This, she said, accommodated her desire to smoke and to drink, which was not acceptable to others in her household. In other words, visualizations like these can be used as a platform to talk to respondents about their lives. The second way the data are used is to model the social network variables across respondents. For example, Lubbers et al. (2007) used the compositional and structural variables of personal networks to explain variability across respondents in ethnic identification. They used a cluster analysis to classify migrants into five network profiles on the basis of eight variables related to network composition (the percentage of Spanish alters, the percentage of migrants, the percentage of family members, average closeness and average frequency of contact with network members), and network structure (density, betweenness centralization, and number of cohesive subgroups), as well as a variable that combined composition and structure (homogeneity of cohesive subgroups). The resulting network profiles appeared to be differentially related to ethnic identifications. In particular, migrants with so-called dense family networks (networks in which network members—mostly family members and people from the country of origin—formed one dense cluster) tended to exhibit ethnic-exclusive self-

14_107-Bernard.indb 648

6/9/14 10:03 AM

20: Social Network Analysis

649

identification, whereas migrants with more heterogeneous personal networks tended to exhibit more plural definitions of belonging. COMBINATION OF WHOLE AND EGO-CENTERED NETWORKS

Some research can benefit from both the whole-network and ego-centered approaches. The benefit of the whole network approach is the detailed understanding of the social structure within a bounded group. The advantage of ego-centered networks is the inclusion of the social context that cuts across many groups. A common way to do both is to conduct a whole network study on, say, the members of a small village, and to use that network to identify respondents for an ego-centered study. For example, you may want to know what resources, external to the village, are available to those who are highly central within the whole network versus those who are marginalized or isolated. Another approach is combining a set of personal networks from respondents within a boundary in an attempt to create a whole network. This approach has the advantage of identifying critical nodes that may not be members of the bounded group but are commonly connected to it. For example, residents of a village may all mention an itinerant trader as an alter in their personal network, while the trader is not actually a member of the village (Mathews 2010). The overlapping network approach is useful for identifying broad structural properties of a whole network, but (depending on the sample size) it cannot reliably represent fine-grained properties. SOCIAL NETWORK THEORY

There are several theoretical concepts that are used by social network analysts. Some of them originated with the study of social networks; others originated elsewhere but are frequently used in social network analysis.

Social Capital The concept of social capital, originally formulated by Bourdieu (1980, 1986), has been developed by several authors such as Lin (1995, 1999, 2001), Putnam (2000), and Portes (1998). The last author describes social capital as “the ability of actors to secure benefits by virtue of membership in social networks or other social structures” (p. 5). The idea that social position may give someone access to information or resources they can then use to their benefit is a useful concept in many of the social sciences. Burt (2004) described the mechanisms by which someone can use social position to their advantage. People who occupy brokering positions, such as those with high betweenness centrality, are in a position to control the flow of information and to synthesize information from different sources into something new. In this sense, social capital leads to the potential for creativity. A contrasting view is that of Krackhardt (1999), whose notion of Simmelian ties is focused in part on conforming to conflicting norms. When applying social networks to a project, you should consider whether being in a brokering position between two or more groups is an advantage, in terms of information control, or a burden, in terms of the need to accommodate potentially competing interests. Some people may occupy a brokering position but may not be aware of it or not inclined to use it to their advantage. Network closure refers to the tendency for a network to become interconnected, particularly in small groups. Within small

14_107-Bernard.indb 649

6/9/14 10:03 AM

650

Christopher McCarty and José Luis Molina

networks, such as a village, there may be strong pressures for networks to become connected. Burt (2005) describes closure as an alternative form of social capital given increased trust from the interconnectedness that comes from small groups.

Diffusion Diffusion of cultural traits used to be a controversial theory in anthropology, although well accepted at some levels by archaeologists. These theories focused at the societal level looking for explanations of cultural traits distribution in wide geographical areas. Network analysts typically examine the diffusion of innovations at the individual level, looking for changes in behavior like the adoption of a given innovation (Rogers 2003). These ideas have been the foundation for many public health interventions (see Valente and Pumpuang 2007). Strong and Weak Ties Granovetter (1973) introduced the notion that people have strong ties to some people and weak ties to others. He showed that for some tasks, such as looking for a job, the weak tie may be better than the strong tie because the weak tie is a link to new opportunities. Strong ties tend to be redundant and connected, such as family and very close friends. These strong ties not only tend to recycle the same information, they may compete for the same opportunities. Weak ties can be thought of as brokerage across social situations. Small World In 1967, Stanley Milgram wrote an article for the first issue of Psychology Today about an experiment he conducted trying to measure the average number of links between any two randomly selected people in the United States. Most people are now familiar with the notion that any two people are linked on average by less than six intermediaries (six degrees of separation). Killworth et al. (1984) examined the nature of the choices people made in choosing intermediaries, showing that occupation and location explained most of these choices. In 1998, Watts and Strogatz showed mathematically that a relatively small number of nonrandom connections, such as through occupation and location, explains the short chains. Anthropologists may use the small world concept in understanding short and long connections between members of cultural groups. Scale-free Networks Following the article by Watts and Strogatz (1998), there was a surge of interest by physicists in social network problems. As the Internet led to social media sites such as MySpace and Facebook and a proliferation of websites, the approach of physics that lends itself to modeling of large amounts of data became a unique thread of research in the field of social networks. Borgatti et al. (2009) outline the difference between the approach of social scientists and that of physicists. Scale-free networks refers to the idea that the basic structure of a network is very much the same for both small and large networks; in other words, the structure is not determined by the size of the network. Some types of networks, such as website

14_107-Bernard.indb 650

6/9/14 10:03 AM

20: Social Network Analysis

651

connections, are scale free. While scale-free networks were first observed in citation networks, physicists have focused on them since the late 1990s. Anthropologists may find that some social phenomena they research are scale free.

Social Learning Social learning theory (Bandura 1977) is simply the idea that people learn from their social context; that is, they learn from others. The notion of social context is easily operationalized using social networks, both whole and ego-centered. This theory lends itself to studies that focus on information transmission and the social avenues and impediments to that transmission. POTENTIAL ISSUES WITH INSTITUTIONAL REVIEW BOARDS

Although social network analysis has been an established discipline for more than 80 years, the advent of social networking sites has dramatically increased the visibility and application of the social network approach. Each year, more disciplines and journals are publishing articles using social networks as a way to operationalize concepts. For decades, Institutional Review Boards (IRBs) approved many projects that they are now reconsidering (Borgatti and Molina 2005). Across universities, IRBs are not necessarily consistent in these evaluations. Perhaps the most important IRB-related issue as of this writing is the controversy over the rights of network members mentioned but not interviewed. This is far more an issue with ego-centered network studies than it is with whole network studies. This issue becomes particularly problematic when the study involves behaviors or conditions that are illegal or stigmatizing. For example, a social network study of migrants where respondents are asked to indicate whether their network alters were legal or not legal migrants would not be approved by some IRBs or research panels. The assumption is that those migrants whose immigration status was reported did not have an opportunity to consent and therefore their rights were violated. This is less of an issue with whole networks because the likelihood is higher that anyone whose behavior is reported is also a respondent. However, it is possible that a member of a whole network could fail to respond and then the same rules would potentially apply. It is possible, of course, to avoid asking any question that involves illegal or stigmatizing behavior. However, that would preclude using a network approach to understand controversial social problems. When possible, researchers can try to avoid real names, although some network techniques require them, particularly whole network studies. Some IRBs will be satisfied with more detailed consent forms or careful data encryption and data handling after data have been collected. LIMITATIONS OF SOCIAL NETWORKS

Studies using a social network approach or studying social networks have proliferated in the last decade. As of this writing, the Web of Science had 6,270 references to articles that contain the term social network. Since 2003, the number of articles has grown exponentially, with just over 200 in 2003 and over 1,000 in 2012. These references are spread across many disciplines, with the highest number from public health (677), sociology (508), psychiatry (483), and management (388). Anthropology had 208.

14_107-Bernard.indb 651

6/9/14 10:03 AM

652

Christopher McCarty and José Luis Molina

Assuming the trend continues, more researchers will apply social network analysis as a method or framework for collecting and analyzing their data. But social network analysis is not a panacea. Researchers are often attracted to it because it is trendy and because network visualizations tend to be compelling. Although it can be an appropriate approach, there is the danger that a relatively straightforward concept may be overcomplicated. Boissevain (1979) put this poignantly: “The battery of techniques with which social scientists have equipped themselves to answer the limited questions that network analysis can resolve produces overkill. Flies are killed with dynamite” (p. 393). While social network analysis is hardly dynamite, there are several reasons one should think twice before incorporating it into a research design. First, social network data are typically difficult to collect. One-mode, whole network data that rely on a survey typically involve high researcher burden as tie evaluations must be collected from as many members of the group as possible. When the group is relatively small and easily located, this is feasible. But with large groups, or groups that are dispersed, the logistics can get out of hand very quickly. With large networks, it becomes increasingly likely that some data will be missing. In contrast to whole networks, ego-centered network data tend to be high on respondent burden as respondents must answer many questions about the network alters they mention. Once network data are collected, they can be difficult to manage. While software packages such as UCINET and Egonet have made network data easier to work with, even these programs require a certain amount of expertise. In the case of whole network analysis packages such as UCINET, the data are sometimes not in a form that can be input without some manipulation. Like any type of data, quantitative or qualitative, one must be prepared for the commitment to learning the details and the peculiarities of social network data. Although readers and reviewers in some disciplines are increasingly becoming well informed about network analysis, most anthropologists are not familiar with it. This means that until network analysis becomes an established method in anthropology, considerable effort must be made to explain the data and analysis to reviewers and readers. The network visualizations that may attract users in the first place are often confusing to those who do not understand them. What may seem a random collection of dots and lines could come out in a variety of ways. For example, a network that is highly cohesive may lack disconnected groups or isolates. The contrast of cohesion versus disconnectedness may be a basis for explanation. Finally, it is not uncommon for researchers to try to apply the network approach when there is no network. It is possible to make a network out of many types of data that do not comprise an interacting set of actors. For example one could use a twomode approach to record different types of foods that a set of respondents eat. Network software can be used to create a network of the actors where the tie represents the foods they eat in common. Eating similar foods does not imply interaction. Despite these limitations, social network analysis has proven to be a useful tool in the social sciences. Anthropologists were among the early developers of network analysis, and the methods and theories of interaction that comprise network analysis are increasingly being used again by anthropologists in creative ways. Used correctly, social net-

14_107-Bernard.indb 652

6/9/14 10:03 AM

20: Social Network Analysis

653

work analysis captures something unique—the pattern of social relations. As the social network approach is applied in other disciplines, it will also establish a common set of variables for collaboration with researchers in both the natural and social sciences. NOTES

1. UCINET is a program developed by Stephen Borgatti, Martin Everett, and Linton Freeman and sold through the website www.analytictech.com. As of this writing, the price is $40 for students and $150 for faculty and government agencies. Pajek is a free program available at http://vlado.fmf.uni-lj.si/pub/networks/pajek/. UCINET tends to be easier to learn than Pajek. Pajek is more efficient in analyzing large datasets and developing network models. 2. Netdraw is a free program for social network visualization available from www.analytictech .com. It is included with the program UCINET. There is a tutorial to learn social network analysis using Netdraw at http://faculty.ucr.edu/~hanneman/nettext/. REFERENCES

Adler-Lomnitz, L. 1975. Cómo sobreviven los marginados. Madrid: Siglo XXI. Aral, S., L. Muchnika, and A. Sundararajana. 2009. Distinguishing influence-based contagion from homophily-driven diffusion in dynamic networks. PNAS 106: 21544–49. Bandura, A. 1977. Social learning theory. Englewood Cliffs, NJ: Prentice Hall. Barbosa de Almeida, M. W., and D. E. William. 2010. On the structure of Dravidian relationship systems, mathematical anthropology and cultural theory. An International Journal 3. www. mathematicalanthropology.org (accessed March 28, 2013). Barnes, J. 1954. Class and committees in a Norwegian island parish. Human Relations 7: 39–58. Berkes, F., and C. Folke, eds. 1998. Linking social and ecological systems: Management practices and social mechanisms for building resilience. Cambridge: Cambridge University Press. Berkman, L. F., and S. L. Syme. 1979. Social networks, host resistance, and mortality: A nine-year follow-up study of Alameda County residents. American Journal of Epidemiology 109: 186–204. Bernard, H. R., T. Hallett, A. Iovita et al. 2010. Counting hard-to-count populations: The network scale-up method for public health. Sexually Transmitted Infections 86: ii11–ii15. Bernard, H. R., E. C. Johnsen, P. D. Killworth et al. 1990. Comparing four different methods for measuring personal social networks. Social Networks 12: 179–215. Bernard, H. R., and P. D. Killworth. 1973. On the social structure of an ocean-going research vessel and other important things. Social Science Research 2: 145–84. Bernard, H. R., and P. D. Killworth. 1979. Why there are no social physics. Journal of the Steward Anthropological Society 11: 33–58. Bernard, H. R., P. D. Killworth, M. J. Evans et al. 1988. Studying social relations cross-culturally. Ethnology 27: 155–79. Bernard, H. R., P. D. Killworth, D. Kronenfeld, and L. Sailer. 1984. The problem of informant accuracy—The validity of retrospective data. Annual Review of Anthropology 13: 495–517. Boissevain, J. 1974. Friends of friends: Networks, manipulators and coalitions. London: Basil Blackwell. Boissevain, J. 1979. Network analysis: A reappraisal. Current Anthropology 20: 392–94. Boorman, S. A., and H. C. White. 1976. Social structure from multiple networks, II: Role structure. American Journal of Sociology 81: 1384–446. Borgatti, S. P. 2002. NetDraw software for network visualization. Lexington, KY: Analytic Technologies. Borgatti, S. P., M. G. Everett, and L. C. Freeman. 2002. UCINET for Windows: Software for Social Network Analysis. Harvard, MA: Analytic Technologies.

14_107-Bernard.indb 653

6/9/14 10:03 AM

654

Christopher McCarty and José Luis Molina

Borgatti, S. P., M. G. Everett, and J. C. Johnson. 2013. Analyzing social networks. Thousand Oaks, CA: Sage. Borgatti, S. P., A. Mehra, D. Brass, and L. G. Labianca. 2009. Network analysis in the social sciences. Science 323: 892–95 Borgatti, S. P., and J. L. Molina. 2005. Toward ethical guidelines for network research in organizations. Social Networks 27: 107–17. Bott, E. 1957. Family and social network. London: Tavistock Institute of Human Relations. Bourdieu, P. 1980. Le capital social: Notes provisoires. Actes de la Recherche en Sciences Sociales 3: 2–3. Bourdieu, P. 1986. The forms of capital. In Handbook of theory and research for the sociology of education, ed. J. G. Richardson, 241–58. New York: Greenwood Press. Burt, R. 2004. Structural holes and good ideas. American Journal of Sociology 110: 349–99. Burt, R. 2005. Brokerage and closure: An introduction to social capital. Oxford: Oxford University Press. Calvet, L., J. L. Molina, and V. Reyes-García. 2012. Seed exchange as an agrobiodiversity conservation mechanism. A case study in Vall Fosca, Catalan Pyrenees, Iberian Peninsula. Ecology and Society 17: 29. Christakis, M. D., and J. H. Fowler. 2009. Connected: The surprising power of our social networks and how they shape our lives. New York: Little, Brown and Company. Cohen, S., and T. A. Wills. 1985. Stress, social support, and the buffering hypothesis. Psychological Bulletin 98: 310–57. Coleman, J. S. 1961. The adolescent society: The social life of the teenager and its impact on education. New York: Free Press of Glencoe. Coleman, J. S., E. Katz, and H. Menzel. 1966. Medical innovation: A diffusion study. Indianapolis: The Bobbs-Merrill Company. Connelly, B. L., J. L. Johnson, L. Tihanyi, and A. E. Ellstrand. 2011. More than adopters: Competing influences in the interlocking directorate. Organization Science 22: 688–703. Costenbader, E., and T. W. Valente. 2003. The stability of centrality measures when networks are sampled. Social Networks 25: 283–307. de Nooy, W., A. Mrvar, and V. Batagelj. 2005. Exploratory social network analysis with Pajek. Cambridge: Cambridge University Press. Dunbar, R. 2008. Cognitive constraints on the structure and dynamics of social networks. Group Dynamic—Theory Research and Practice 12: 7–16. Ellen, R., P. Parkes, and A. Bicker. 2000. Indigenous environmental knowledge and its transformations: Critical anthropological perspectives. Amsterdam: Harwood Academic Publishers. Epstein, A. L. 1969. Gossip, norms and social networks. In Social networks in urban situations: Analyses of personal relationships in central African towns, ed. C. J. Mitchell, 117. Manchester, UK: Manchester University Press. Feld, S. 1981. The focused organization of social ties. American Journal of Sociology 86: 1015–35. Fischer, C. S. 1982. To dwell among friends: Personal networks in town and city. Chicago: University of Chicago Press. Freeman, L. C. 1999. Visualizing social networks. JOSS 1. http://www.cmu.edu/joss/content/ articles/volume1/Freeman.html (accessed March 15, 2013). Freeman, L. C. 2004. The development of social network analysis: A study in the sociology of science. Vancouver, Canada: Empirical Press. Gonçalves, B., N. Perra, and A. Vespignani. 2011. Modeling users’ activity on Twitter networks: Validation of Dunbar’s Number. PLoS ONE 6: e22656. doi:10.1371/journal.pone.0022656. Granovetter, M. 1973. The strength of weak ties. American Journal of Sociology 78: 1361–81.

14_107-Bernard.indb 654

6/9/14 10:03 AM

20: Social Network Analysis

655

Gruzd, A., B. Wellman, and Y. Takhteyev. 2011. Imagining Twitter as an imagined community, American Behavioral Scientist 55: 1294–318. Hage, P., and F. Harary. 1997. Island networks: Communication, kinship and classification structures in Oceania. Cambridge: Cambridge University Press. Hamberger, K. 2011. Matrimonial circuits in kinship networks: Calculation, enumeration and census. Social Networks 33: 113–28. Hanneman, R. A., and M. Riddle. 2005. Introduction to social network methods. Riverside, CA: University of California. http://faculty.ucr.edu/~hanneman/ (accessed March 28, 2013). Hopkins, A. 2011. Use of network centrality measures to explain individual levels of herbal remedy cultural competence among the Yucatec Maya in Tabi, Mexico. Field Methods 23: 307–28. Houseman, M., and D. R. White. 1996. Structures réticulaires de la pratique matrimoniale. L’Homme 139: 59–85. Jackson, M. O. 2008. Social and economic networks. Princeton, NJ: Princeton University Press. Johnsen, E. C. 1985. Network macrostructure models for the Davis-Leinhardt set of empirical sociomatrices. Social Networks 7: 203–24. Johnson, J. C. 1994. Anthropological contributions to the study of social networks: A review. In Advances in social network analysis: Research in the social and behavioral sciences, ed. S. Wasserman and J. Galaskiewicz, 113–51. Thousand Oaks, CA: Sage. Kadushin, C. 1968. Power, influence and social circles: A new methodology for studying opinion makers. American Sociological Review 33: 685–99. Kadushin, C. 1982. Social density and mental health. In Social structure and network analysis, ed. P. V. Marsden and N. Lin, 147–58. Beverly Hills, CA: Sage. Kadushin, C. 2012. Understanding social networks: Theories, concepts and findings. New York: Oxford University Press. Kapferer, B. 1972. Strategy and transaction in an African factory: African workers and Indian management in a Zambian town. Manchester, UK: Manchester University Press. Kasper, C., and B. Voelkl. 2009. A social network analysis of primate groups. Primates 50: 343–56. Killworth, P. D., and H. R. Bernard. 1976. Informant accuracy in social network data. Human Organization 35: 269–86. Killworth, P. D., and H. R. Bernard. 1979. Informant accuracy in social network data III: A comparison of triadic structure in behavioral and cognitive data. Social Networks 2: 19–46. Killworth, P. D., H. R. Bernard, and C. McCarty. 1984. Measuring patterns of acquaintance. Current Anthropology 25: 381–97. Kim, S., and E. H. Shin. 2002. A longitudinal analysis of globalization and regionalization in international trade: A social network approach. Social Forces 81: 445–71. Klovdahl, A. S. 1985. Social networks and the spread of infectious diseases: The AIDS example. Social Science and Medicine 21: 1203–16. Krackhardt, D. 1990. Assessing the political landscape: Structure, cognition, and power in organizations. Administrative Science Quarterly 35: 342–69. Krackhardt, D. 1999. The ties that torture: Simmelian tie analysis in organizations. Research in the Sociology of Organizations 16: 183–210. Laumann, E. 1973. The bonds of pluralism: The forms and substance of urban social networks. New York: Wiley. Latkin, C. A., V. Forman, A. Knowlton, and S. Sherman. 2003. Norms, social networks, and HIV-related risk behaviors among urban disadvantaged drug users. Social Science and Medicine 56: 465–76. Lewis, K., J. Kaufman, M. Gonzalez et al. 2008. Tastes, ties, and time: A new social network dataset using Facebook.com. Social Networks 30: 330–42.

14_107-Bernard.indb 655

6/9/14 10:03 AM

656

Christopher McCarty and José Luis Molina

Lin, N. 1995. Social resources: A theory of social capital. Revue française de sociologie 36: 685–704. Lin, N. 1999. Social networks and status attainment. Annual Review of Sociology 25: 467–87. Lin, N. 2001. Social capital: A theory of social structure and action. Cambridge. Cambridge University Press. Lorrain, F., and H. C. White. 1971. Structural equivalence of individuals in social networks. Journal of Mathematical Sociology 1: 49–80. Lubbers, M., J. L. Molina, and C. McCarty. 2007. Personal networks and ethnic identifications: The case of migrants in Spain. International Sociology 22: 720–40. Martin, C. L., R. A. Fabes, L. D. Hanish, and T. Holleristein. 2005. Social dynamics in the preschool. Developmental Review 25: 299–327. Mathews, M. C. 2010. Socio-economic change in the transition from patron-client to social movement networks in Brazilian Amazonia. University of Florida, ProQuest, UMI Dissertations Publishing, 3467581. McCarty, C., and P. D. Killworth. 2007. Impact of methods for reducing respondent burden on personal network structural measures social networks. Social Networks 29: 300–15. McCarty, C., J. L. Molina, C. Aguilar, and L. Rota. 2007. A comparison of social network mapping and personal network visualization. Field Methods 19: 145–62. Milgram, S. 1967. The small world problem. Psychology Today 1: 61–67. Mitchell, C. J. 1969. Social networks in urban situations: Analyses of personal relationships in central African towns. Manchester, UK: Manchester University Press. Mitchell, C. J. 1972. Foreword. In Strategy and transaction in an African factory: African workers and Indian management in a Zambian town, ed. B. Kapferer, vii–xvi. Manchester, UK: Manchester University Press. Molina, J. L., I. Maya-Jariego, and C. McCarty. 2014. Giving meaning to social networks: Methodology for conducting and analyzing interviews based on personal network visualizations. In Mixed methods social networks research: Design and application, ed. B. Holstein and S. Dominguez, 305–35. Cambridge: Cambridge University Press. Molina, J. L., S. Petermann, and A. Hertz. 2012. Defining and measuring transnational fields. MMG Working Papers Print. 16. Göttingen, Germany: Max Planck Institute for the Study of Religious and Cultural Diversity. http://www.mmg.mpg.de/publications/working-papers/ 2012/wp-12-16/ (accessed April 15, 2013). Moreno, J. L. 1934. Who shall survive? Foundations of sociometry. New York: Beacon House. Mouttapa, M., T. Valente, P. Gallaher et al. 2004. Social network predictors of bullying and victimization. Adolescence. 39: 315–35. Nadel, S. F. 1957. Teoría de la estructura social. Madrid: Guadarrama, Onnela, J. P., J. Saramaki, J. Hyvonen et al. 2007. Structure and tie strengths in mobile communication networks. PNAS 104: 7332–36. Padgett, J. F., and C. K. Ansell. 1993. Robust action and the rise of the Medici, 1400–1434. American Journal of Sociology 98: 1259–319. Portes, A. 1998. Social capital: Its origins and applications in modern sociology. Annual Review of Sociology 24: 1–24. Putnam, R. D. 2000. Bowling alone: The collapse and revival of American community. New York: Simon & Schuster. Reyes-García, V., J. L. Molina, J. Broesch et al. 2008. Do the aged and knowledgeable men enjoy more prestige? A test of predictions from the prestige-bias model of cultural transmission. Evolution and Human Behavior 29: 275–81. Rice, R. E. 1994. Relating electronic mail use and network structure to R&D work networks and performance. Journal of Management Information Systems 11: 9–20.

14_107-Bernard.indb 656

6/9/14 10:03 AM

20: Social Network Analysis

657

Richards, W., Jr., and G. A. Barnett, eds. 1993. Progress in communication sciences: Advances in communication network analysis. Norwood, NJ: Ablex. Robins, G. L., T. A. B. Snijders, and P. Wang et al. 2007. Recent developments in exponential random graph (p*) models for social networks. Social Networks 29: 192–215. Rogers, E. M. 2003. Diffusion of innovations, 5th ed. New York: Free Press. Rohilla Shalizi, C., and A. C. Thomas. 2011. Homophily and contagion are generically confounded in observational social network studies. Sociological Methods & Research 40: 211–39. Schweizer, T. 1997. Embeddedness of ethnographic cases: A social networks perspective. Current Anthropology 38: 739–60. Scott, J. 2000. Network analysis: A handbook, 2nd ed. Newbury Park, CA: Sage. Simmel, G. 1955. Conflict and the web of group-affiliations. New York: The Free Press. Snijders, T. A. B. 2005. Models for longitudinal network data. In Models and methods in social network analysis, ed. P. Carrington, J. Scott, S. Wasserman, 215–47. New York: Cambridge University Press. Snijders, T. A. B., G. G. van de Bunt, and C. E. G. Steglich. 2010. Introduction to actor-based models for network dynamics. Social Networks 32: 44–60. Srinivas, M. N., and A. Béteille. 1964. Networks in Indian social structure. Man 64: 212. Valente, T. W. 1997. Social network associations with contraceptive use among Cameroonian women in voluntary associations. Social Science & Medicine 45: 677–87. Valente, T. W. 2010. Social networks and health: Models, methods and applications. Oxford: Oxford University Press. Valente, T. W., and P. Pumpuang. 2007. Identifying opinion leaders to promote behavior change. Health Education & Behavior 34: 881–96. VanderWeele, T. J. 2011. Sensitivity analysis for contagion effects in social networks. Sociological Methods & Research 40: 240–55. Vertovec, S. 2009. Transnationalism. Key Ideas Series. New York: Routledge. Wasserman, S., and K. Faust. 1994. Social network analysis: Methods and applications. Cambridge: Cambridge University Press. Wasserman, S., and P. E. Pattison. 1996. Logit models and logistic regressions for social networks: I. An introduction to Markov graphs and p*. Psychometrika 61: 401–25. Wasserman, S., and G. Robins. 2005. An introduction to random graphs, dependence graphs, and p*. In Models and methods in social network analysis, ed. S. Carrington and S. Wasserman, 148–61. New York: Cambridge University Press, Watts, D. J., and S. H. Strogatz. 1998. Collective dynamics of “small-world” networks. Nature 393: 440–42. Wellman, B. 1979. The community question: The intimate networks of East Yorkers. American Journal of Sociology 84: 1201–31. White, D. R., V. Batagelj, and A. Mrvar. 1999. Anthropology: Analyzing large kinship and marriage networks with pgraph and Pajek. Social Science Computer Review 17: 245–74. White, D. R., and U. C. Johansen. 2005. Network analysis and ethnographic problems: Process models of a Turkish nomad clan. Plymouth, UK: Lexington Books. White, H. C., S. A. Boorman, and R. L. Breiger. 1976. Social structure from multiple networks, I: Blockmodels of roles and positions. American Journal of Sociology 81: 730–80. Wolfe, A. W. 1978. The rise of network thinking in anthropology. Social Networks 1: 53–64.

14_107-Bernard.indb 657

6/9/14 10:03 AM

14_107-Bernard.indb 658

6/9/14 10:03 AM

Part IV

APPLYING AND PRESENTING INFORMATION

14_107-Bernard.indb 659

6/9/14 10:03 AM

14_107-Bernard.indb 660

6/9/14 10:03 AM

CHAPTER 21

Theories and Methods in Applied Anthropology Robert T. Trotter, II, Jean J. Schensul, and Kristin M. Kostick Applied anthropologists commonly conduct research for two reasons. One is to produce research that has straightforward findings or implications that can be used for direct interventions or lead to recommendations for policy change (Pelto and Schensul 1987). The other is to test and improve anthropological theory through devising experiments in sociocultural interventions or policy changes (Van Willigen 2002; Wasson et al. 2012). Applied anthropologists study conditions ranging from health and disease prevention to educational innovation and instructional improvement, community economic development, business anthropology, and environmental protection. Most of their work is based in or concerned with culturally diverse communities and populations. Increasingly, research, interventions, and evaluations are conducted by partnerships that include anthropologists and representatives of the communities or countries in which the problem is addressed. The current movement toward increased community engagement in research (community-based participatory research [CBPR] and other forms) has very strong roots and representation in applied anthropology (Greenwood and Levin 2007; J. Schensul and S. Schensul 1992; Singer and Baer 2007). This combination of complex and diverse communities and subject matter in the context of plural cultural and political systems requires a robust and diverse set of theoretical and methodological options that can be applied under less-than-ideal research conditions. All research must be conducted with integrity, but two features of applied research place greater-than-usual responsibility on the researcher’s shoulders to assure compliance with both the letter and the spirit of professional ethics for applied research (Whiteford and Trotter 2008). First, the human, social, and ecological consequences of applied research are immediate, potentially significant, and sometimes critical to the life and survival of communities (Nastasi and Schensul 2005). Second, because the results are change oriented, they may be disruptive or threatening. Some researchers (e.g., Campbell and Stanley 1963; S. Schensul 1985) have said applied social science research can and should be the most creative and rigorous of all social science research so as to address these circumstances responsibly. The best approach to conducting ethical research in applied contexts is to carefully select the most appropriate combination of theory and methodological approaches that are consistent with local context and the intended outcomes of the research process.

661

14_107-Bernard.indb 661

6/9/14 10:03 AM

662

Robert T. Trotter, II, et al.

TYPES OF APPLIED APPROACHES

Applied anthropologists and other social scientists have delineated five types of applied research (see Ervin 2004; Podolefsky et al. 2008; S. Schensul and J. Schensul 1978; Van Willigen 2002). These are as follows.

Policy Research Policy research intends to assess the effects of a policy, to adapt or change it, or to generate new policies. In anthropology, it usually involves conducting ethnographic research and making suggestions for policy change through single events (press conferences or workshops). Less frequently, it involves describing the effects of implementing a set of policies on a target population and demonstrating the process of change as well as the need for policy change. In both approaches, the researcher speaks to the policymakers but is generally not actively involved either in the process of policymaking or in penetrating the locations, networks, or policy clusters within which policy is made (J. Schensul 1985). The recognition that those who produce scientific knowledge for policymakers must be involved in significant ways in the policymaking process has been at the forefront of thinking in anthropology for the past decade. It is only recently, however, that anthropologists have provided written examples of their direct involvement with policymakers at governmental and other levels (e.g., Shore et al. 2011). Evaluation Research Evaluation research is intended to improve or evaluate the efficacy or outcome of a program, project, or organization (Trotter 1996). In some instances, the anthropologist isn’t directly involved in developing or implementing the program and doesn’t have specific responsibility for translating research into program models or activities. However, there are growing exceptions to that general rule where anthropologists are responsible for corrective feedback in real time, or corrective interventions in the system, over time (Briody and Trotter 2008; Hahn and Inhorn 2009). Generally, the evaluation identifies cultural patterns, networks, processes, environmental conditions, system models, or other factors likely to help a program or be important in determining consistency of implementation and outcome. At times, the process includes participatory empowerment (Fetterman et al. 1993; Ovretveit 2002) or action research evaluation models, and the evaluation researcher plays a role in implementation and has experience in managing the class of programs targeted by the evaluation. In each case, the focus tends to be empirically based data collection within a targeted theoretical framework that compares ideal or benchmarked conditions with actual performance. Theories and methods that allow for a constant comparison approach tend to be favored in these endeavors. Cultural Intervention Research The applied anthropologist is directly involved in the development, conduct, and evaluation of a culturally based, theory-driven intervention. Here, the researcher conducts research that identifies cultural factors important in guiding interventions and uses the findings to generate or identify appropriate intervention theories. The creation and conduct of the intervention, based on cultural knowledge and theory development,

14_107-Bernard.indb 662

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

663

also falls to the anthropologist, who, like the participatory evaluator, must know a lot about methods of program implementation. Both multidisciplinary teamwork and community engagement is advisable in cultural intervention research. Examples of this type of approach can be found in social marketing research in the context of both localized and large-scale health promotion and disease prevention interventions (Bernard 2011; Grier and Bryant 2005; J. Schensul 1998a; J. Schensul et al. 2009; Stead et al. 2007; Van Willigen 2002; Weeks et al. 2009a).

Advocacy, Activist, or Action Research This is specifically directed toward identifying, critiquing, and addressing imbalances in allocation of power, economic resources, social status, material goods, and other desired social or economic elements in a community, society, or globally. Advocacy research may include evaluation, policy research, and research and development. The end result, however, is to increase, organize, and activate resistance in community groups, with unions, with groups representing underrepresented or excluded populations, and with those with limited (or perceived as limited) power to change their own conditions. Anthropologists engaged in advocacy research tend to do so from a liberal or critical perspective; thus, culturally based theories guiding advocacy research focus on structural barriers contributing to social inequities, including issues related to class and caste, power, gender, age, sexual preference, linguistic usage, and ethnic group identification. “Action research” is a now ubiquitous term in the social science literature; it can refer to iterative research leading to any action, rather than action guided by a critical perspective (see Stringer 1996). Advocacy research in anthropology stems from the concept of action research first suggested by Sol Tax (1960), and reiterated in the work of Stephen and Jean Schensul (1978; J. Schensul 2010). It also appears in other classic and contemporary work among anthropologists (e.g., Ervin 1996; Lester 2011; Singer and Baer 1995). In this chapter, we use the term “advocacy research.” To reinforce the critical perspective that undergirds activist research, S. Schensul et al. (this volume) have shifted to the term “transformational” anthropology, which emphasizes a reframing of an inequitable situation to address the structural and social factors that create it. Activist research, as defined and described by Hale (2007), is when an anthropologist forges alliances with groups and communities experiencing inequities or claiming their indigenous identities and heritage and conducts research in collaboration with them that furthers their political agendas and has a similar meaning. Participatory Action Research Participatory action research (PAR) involves several critical elements, including a long-term partnership with those directly affected by an issue, often because they are experiencing inequities in health, housing, cultural conservation, or political participation. The partnership involves continuous interaction of research with the action process through joint researcher/social actor data collection, analysis, reflection, and use of the research for social and social justice ends. In the other forms of research mentioned above, the means (research) leads to the end (an evaluation, a program, a policy change, etc.). In PAR, the means is the end, and the conduct of research is embedded in the process of introducing or generating change.

14_107-Bernard.indb 663

6/9/14 10:03 AM

664

Robert T. Trotter, II, et al.

PAR is, first and foremost, locally specific and intended to further local goals with local partners. It is also theoretically driven (see Berg et al. 2009) including framing by action oriented theorists such as Dale and Hyslop-Margison (2010), Freire (1995), Giroux (2010), and many others. Early work, such as that of Holmberg and colleagues in their classic intervention in the Cornell-Vicos experiment, merged scientific research with ongoing contributions to improving agriculture, education, housing, and other social domains (Holmberg 1954, 1958, 1966). Some committed to PAR believe that it should contribute to both the development of scientific knowledge on any given topic and make special efforts to publish their work, usually collaboratively. Others frame their research as contributing to various forms of political, social, and economic action and do not attempt to make the results, rigorous though they must be, known to the scientific community. Now, however, there are many scientific journals and publications devoted to describing and disseminating the results of PAR and it is a very widely accepted methodology for the ethical conduct of applied research. PAR is further discussed in S. Schensul et al. (this volume). Applied anthropologists selecting one or another of these orientations must consider with whom they wish to work, whether they have the skills or experience to conduct research and engage in practice, what their personal and professional values are, to what degree these values are rooted in particular theoretical frameworks, what position they occupy in the structure of the research setting, and where they wish to place themselves on the continuum that ranges from those attempting to be objective and value free to those who are integrated or embedded participants in the promotion of specific shared goals. “Positionality” and “reflexivity” are useful terms with respect to these decisions (Robertson 2002). Positionality refers to self-conscious reflection and recognition of the social position of the researcher in relation to the community and individuals of study (Jacobs-Huey 2002). Since the 1970s, anthropologists have written about their own social status and other critical dimensions (e.g., ethnicity, social race, class, formal education, appearance and style, age, country of origin, and so on), all of which can and do influence the ways in which they enter a community—their own or others’—and conduct research both initially and at any given time in the field experience (Harding 1987; LeCompte and J. Schensul 2010). Reflexivity refers to the researcher’s ability to reflect on their interactions, the ways in which they influence both interaction and ability to collect data and build partnerships, and their ability to recognize and modify their behaviors as best they can. Applied researchers may conduct any or all of the aforementioned forms of research, separately or in combination in the same field setting. Confusion can arise when these approaches serve contradictory purposes (e.g., are advocacy research and evaluation mutually exclusive?). It’s important to identify which approach is being used (and why) to avoid or resolve possible challenges to researcher identity, ensuring proper presentation of self in the research site. All of applied anthropology is subject to the constraints of time, local politics, and other contextual factors. Though we may try, we cannot control conditions in field situations as we can in clinic or laboratory settings. Ethnographic methods suit field research because they offer researchers and their partners a high degree of methodological flexibility to respond to new circumstances as

14_107-Bernard.indb 664

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

665

they arise, while also maintaining a defensible level of rigor in the research outcomes. Furthermore, ethnography, although guided by general theoretical principles, gives high value to inductive (emergent theory) or localized theory building. Local theory (in interplay with more general theories of change) that is controlled (developed, shared, and understood) by partners in change is far more likely to result in positive outcomes than theory imposed from above. This is not in opposition to experimental design, since we believe in and have successfully used experimental designs in field situations. Instead, we suggest that anthropologists have much to offer the field of applied social science and interdisciplinary research because we are not wedded to specific designs and instruments but to inquiry, exploration, and discovery that guides the most effective selection of theory, methods, and data-collection techniques for a given situation. Applied researchers must be excellent group facilitators as well as researchers and should be familiar with techniques for conducting research with groups and individuals to maximize the rigor of the research. BUILDING AN APPLIED SETTING

Applied research is embedded in a setting in which a problem has been identified and a group is present to address it. If these two conditions are not met, the research may be useful but it will rarely be used. An important part of method in applied anthropology involves paying close attention to the structures and relationships that affect the success or failure of the research mission. The use of ethnography generally distinguishes anthropologists from other social scientists. In traditional ethnography, the anthropologist is a student of the culture and the indigenous expert is the teacher. Applied anthropology calls for reciprocal learning and sharing of expertise in identifying a problem, defining a researchable question, conducting research, and using results, what Stull and Schensul (1987) call collaborative research and others label as partnerships. The applied researcher is involved in shaping theory, design, and data collection to fit the requirements of the field situation and those of the field partners. Common approaches to collaborative field research include consultations with bureaucracies, contract research or evaluations with governments or nonprofit organizations, and researcher-initiated collaborations with clinics, schools, community groups, or networks of nonprofit organizations (Harrison 2001; J. Schensul and S. Schensul 1992) as well both external and embedded applied research in business and industrial organizations (Jordan 2003, Briody et al. 2010). These partnerships are necessary for evaluating the implications of policy decisions or for developing and testing programmatic interventions. In addition to understanding traditional anthropological theory and methods, applications research demands skills and value orientations that are only now beginning to be taught systematically in anthropology graduate training programs. For example, to initiate a change-oriented program, it’s critical to identify key actors outside the university who are influential leaders and who can engage in community or institutional change, program development, or policy advocacy. These are people who are linked to service or policy systems—systems that also must be understood to introduce change efforts appropriately. Understanding the concept of program or intervention;

14_107-Bernard.indb 665

6/9/14 10:03 AM

666

Robert T. Trotter, II, et al.

being able to cross disciplinary boundaries in forging academic partnerships and in conceptualizing and writing intervention grants; designing and managing budgets; and administering research, cultural interventions, and interdisciplinary teams are all additional critically important skills (Lamphere 2004). Finally, team building and group facilitation skills are central to endeavors in complex community or institutional settings (Rappaport 2007; J. Schensul 1999). If researchers can’t quickly grasp the “politics” of the setting and negotiate relationships to promote and achieve common goals across sectors, partnerships can fall apart very quickly (Briody and Trotter 2008; J. Schensul et al. 1999). Increasingly, anthropologists do research and act to improve conditions for their own ethnic/social/racial or special interest group. These anthropologists speak as experienced social scientists as well as informed insiders and must contend with issues faced by any researcher conducting research in his or her own community. The issues faced by “insider researchers” are different from those encountered by researchers from the outside (Bartunek and Louis 1996). Insiders are already “tracked” in their communities, by gender, class, family reputation, affiliation, education level, and so on. They must juggle the demands of the outside world of funders, researchers, and policymakers with the demands of family, friends, and community politics. On the other hand, they are privy to insiders’ information, understand local language and references, and are more likely to recognize the utility of local social structures and networks and cultural beliefs in the development of interventions. As a result, they can more readily develop, test, and disseminate culturally appropriate research tools. Applied anthropology has a global as well as a local (i.e., U.S.-based) face. A full issue of the NAPA Bulletin reviewed applied anthropology in Asia (India and China), the Middle East (Egypt and Israel), the United Kingdom, Canada, Portugal, and other countries (Hill and Baba 2006). Latin American anthropologists working in countries such as Honduras, Mexico, Puerto Rico, and Argentina are involved in joint ethnography, research contributing to environmental advocacy and policy change, farmers’ and workers’ rights, and forensic work for NGOS (Field and Fox 2007; Guerrón-Montero 2004). Some of the articles in these publications are written by U.S. or Canadian anthropologists working with anthropological and other collaborators in other countries, and many are written by in-country anthropologists who are working with NGOs, voluntary organizations, and national institutions to conduct applied research. These publications all emphasize the importance of understanding multiple perspectives, addressing the challenges and constraints of local and national bureaucracies, and maintaining a creative stance in the development of theoretical frameworks and methods that respect local views, ideologies, and needs. BUILDING AND USING TESTABLE ETHNOGRAPHIC THEORY

Building and using strong, testable theory is the most crucial element for creating or selecting research methods in applied anthropology. Theory and methods are always bound together. The way theories are constructed and presented should suggest ways to test them. Theories imply directions in intervention, program development, or policy that will be acted on once the research is done or while it is being conducted. Testing theory in the field, through research and intervention, improves understanding of the

14_107-Bernard.indb 666

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

667

theory, the field situation, the cultural conditions to be modified or influenced, and human responses to both. The concept of midrange theory is especially attractive in applied anthropology. This concept has several meanings. Pelto and Pelto (1978) refer to it as predictive generalizations arising from the data and linked to broader theoretical approaches or paradigms such as structural, functional, ecological, or systems theories or critical, materialist, or evolutionary theories. We agree with this definition but refine it by referring to the idea that midrange theory is anthropological or cultural theory that can be tested and subsequently applied within an empirical context (Trotter 1997). Midrange cultural theories are attempts to identify the important patterns of thought or behavior in specific domains of a culture—patterns that are representative of an identified group of people in a designated context (Trotter 1997). In other words, midrange theories are locally situated. A classic example includes identifying the decision-making processes farmers use to decide whether to plant subsistence or cash crops (Barlett 1977). Midrange theoretical models describe, explain, and/or predict what is going on in one or more cultural and behavioral domains in a specific local environment. Such models are generated from prior knowledge and field experience, are tested in the field, and are continually refined (Glaser and Strauss 1967; Strauss and Corbin 2008; Trotter 1995, 1997). The cultural “frame” (i.e., the lens through which culture is viewed and defined) may be cognitive, behavioral, structural, or critical. The choice of cultural frame influences (or is influenced by) the selection of a problem and theory. It also influences the selection of methods of sampling, data collection and data analysis. And it influences interpretation and utilization of the results of the research. Some applied researchers use a combination of theoretical approaches. Midrange, descriptive cultural theories may predict cultural choice and point to general intervention theories or actions to be taken, but they don’t always specify how (in detail) the intervention will be structured. For example, network theories (see below) suggest several directions for intervention: diffusion of information using opinion leaders (Weeks et al. 2009a), working with bounded networks to influence “peer norms,” or mobilizing social supports for individuals through their ego-centered networks (Trotter et al. 1995b, 1995c, 1996). But those theories may not specify exactly how to work with opinion leaders. They may not specify all the approaches that are useful in diffusing information (regardless of the diffusion agent) or exactly what to do with social support networks to strengthen, modify, or eliminate selected behaviors. So, to bring about changes in individuals, systems, and policies, applied anthropologists often consider intervention theories from other disciplines in addition to those implied by cultural research. MIDRANGE THEORY DEVELOPMENT IN APPLIED ANTHROPOLOGY

When conducting research, anthropologists generally focus on cultural theories—that is, theories that predict patterns among groups rather than of single individuals. Here, we discuss several areas of successful cultural theory development that can be used to select ethnographic and other qualitative research methods shown to have positive implications for change.

14_107-Bernard.indb 667

6/9/14 10:03 AM

668

Robert T. Trotter, II, et al.

Sociocognitive Theories A number of theories about cultural beliefs and thought processes have been derived from the cognitive sciences (including psychology and psychological anthropology) to investigate the psychological aspects of cultural dynamics. A subset of these theories is concerned with links between cognition and behavior. The midrange cognitive behavioral theories used most commonly in anthropology include cultural models approaches and systematic approaches to assessing cultural beliefs (J. Schensul 1998b). The underlying assumption in these approaches is that culture is a mental phenomenon, consisting of identifiable conceptual domains that are shared and can be identified, analyzed, and potentially changed using theoretically congruent processes that change cognitive–behavioral constructs at a cultural level. In an applied context, this approach assumes that group change in behavior occurs through changes in cultural beliefs, attitudes, perceived norms, and concepts. These theories lend themselves to social marketing, to communications approaches to change, and to approaching change through individual learning and changes in beliefs or knowledge. The following theoretical models should not be thought of as mutually exclusive. They are often successfully used in combination to construct effective programs or intervention studies. CULTURAL MODELS AND CONSENSUS THEORIES

Cultural models theories are based on a view of culture as a set of modal cultural beliefs and norms (Romney et al. 1986) and lead to methods that explore the degree of consensus and variation across individuals in groups as well as individual expertise or knowledge of a given cultural domain (Weller and Romney 1988). Examples of cultural models theories in applied research can be found in the use of cultural congruency (J. Schensul et al. 1993; Trotter 1991; Trotter and Potter 1993) and consonance models (Dressler et al. 2005) to explore health beliefs, behaviors, and emic models of risk, contagion, or trauma (Kleinman 1980; Kleinman and Benson 2006; Quinn and Holland 1987). Researchers operating from this paradigm also acknowledge that the link between beliefs and behaviors is not always straightforward and that beliefs and norms are differentially internalized and motivated across individuals within cultures (D’Andrade et al. 1972; Spiro 1987; Vygotsky 1986), but can lead to interesting explorations of cultural decision modeling in practical cases (Gladwin 1989; Ryan and Bernard 2006). Some examples of how the cultural models approach has been used in applied settings include identifying variation in health beliefs across patients and providers to reduce cognitive incongruity and promote shared understandings that can help remove general or patient-specific barriers to quality care (Chavez et al. 1995; Garro and Mattingly 2000). The method for testing the theory melds ethnographic survey questions with a formal mathematical model based on approaches used by psychometricians in test construction and is influenced by signal detection theory and latent structural analysis procedures (Romney et al. 1986). Some anthropologists involved in work on HIV/AIDS use consensus theory to identify core versus peripheral values, where core values are those that most people agree on and peripheral values are those about which there is less agreement. The identification of significant variation within a group can

14_107-Bernard.indb 668

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

669

be instrumental in identifying beliefs and norms within a community that represent targets of positive change through social marketing and/or cognitive–behavioral interventions (e.g., Kostick et al. 2011). SOCIAL CONSTRUCTION THEORIES

Social construction theories argue that cultural knowledge, norms, skills, and behaviors are co-constructed through a negotiated group process in specific cultural contexts. According to this perspective, interpersonal exchange is essential to the development of individual cognitions and behavior and to the evolution and transmission of culture (see, e.g., Bearison 1982; Berger and Luckman 1966; Rogoff 1990; Vygotsky 1978, 1986 [1934]; Wertsch 1991). This approach suggests that new ways of thinking and behavior develop initially during social interactions in which more experienced or knowledgeable individuals mediate the person–environment interaction. With repeated exchanges in similar contexts, new ideas and behavior become internalized. Interpersonal relationships provide the context for reinforcing shared beliefs and behaviors, enhancing perceptions of competence, and encouraging persistence of group norms. This theoretical framework lends itself to the use of ethnographic research methods involving symbolic interactionism methods, network research methods, group elicitation techniques for negotiating as well as documenting existing and changing group norms, and audiovisual recording methods for recording and understanding group processes. The approach can be used in group interventions, as illustrated by the work of Berg, Schensul, and Nastasi in the development and implementation of prevention curricula for adults (J. Schensul et al. 2008), and for teens in and out of school based on the co-construction of knowledge through collective research (Berg et al. 2009; Nastasi et al. 1998); it’s linked to network theory and research since networks are naturally bounded groupings within which group interventions can be conducted (Trotter et al. 1996). CULTURAL SCRIPTING THEORIES

Scripting theories assume that patterns of conduct are locally situated and socially rooted, learned, and change over time. Cultural scripts are selectively used, modified, and adapted as people make choices in their lives and implement their understandings in interpersonal scripts with friends and partners. According to this theory, the scripts that people develop include a meshing of what both the cultural setting and the individual define as the cultural domain. The theory promotes the notion that both individuals and public institutions can be innovative in changing the ways in which a cultural domain is represented. For example, middle-class suburban children’s birthday parties are scripted in a particular way—including features such as balloons, paid entertainment, acceptable gift categories, appropriate clothes, acceptable foods, and so on. Sufficient innovation—such as a decision to avoid giving gifts or to favor family parties—can change the cultural scripts in a society (Castillo and Geer 1993). As Gagnon and Simon and others show, sexual relationships and rituals are scripted as well (Gagnon and Simon 1987; J. Schensul 1998b), shaping partners’ perceptions and understandings of their interaction and sexual negotiation, their behaviors, and their internalizations of responses to those behaviors.

14_107-Bernard.indb 669

6/9/14 10:03 AM

670

Robert T. Trotter, II, et al.

Script theory is closely related to network theory since scripts as event sequences are socially negotiated through networks or groups (Parker and Gagnon 1995). Often, a combination of social marketing (societal level), diffusion theory and influence or power (network level), and cognitive behavioral interventions (individual or dyad level) are necessary to bring about desired changes in cultural scripting. One of the strengths of combining these approaches is that they are all methodologically compatible with the same set of research methods (e.g., semi-structured [theory based] indepth interviews, contextual/pathway analysis interviews, grounded theory metaphor analysis, and cultural models interviews).

Contextual Research Anthropological midrange theories have helped establish the importance of cultural contexts and the organization and structure of human systems (Kozulin et al. 2003). Contextual research addresses the cultural environment or social conditions that have an impact on daily living (e.g., environmental, political, economic, and social contexts of group beliefs, norms, and behaviors). This research derives from theories of social identity, kinship, and social network analysis and the impact of social and cultural structures on human behavior. Theoretical models include diffusion theory approaches to cultural change and innovation, which address how knowledge is differentially or strategically spread throughout social networks, theories of organizational control and behavior, and theories on dynamics of social networks and the small-world phenomenon (Watts 1999). Other context-specific theoretical models come from community participation research: gender, race, and power analysis and studies of cultural diffusion, cultural resistance, and cross-cultural conflict. Below, we address ecological theories, social network theories, and critical theories in guiding interventions. ECOLOGICAL THEORIES

Ecological theories link cultural conditions to the context of humans within an ecological or political framework and are a third area of development and exploration for applied anthropology. The models derived from ecological theory are multifactoral and may include interaction with physical or bioenvironmental characteristics. In addition to consideration of multiple “independent variable domains” (i.e., selected elements of larger systems) influencing individual or group behavior, two additional factors are important in ecologically driven investigation: the local nature of the investigation and the notion of adaptation, which assumes that individuals and groups engage in continuous adjustment to environmental circumstances. Most prevention research now uses an ecological framework, identifying and attempting to address or rectify barriers to change at any level (e.g., the family) in interaction with other levels (e.g., the health care system, individual constraints, school problems, etc.; see Bronfenbrenner 1977; Dryfoos 1990). The midrange theories tested within this paradigm include barriers to change research (referring to environmental factors impeding change or access), cultural congruency models (which attribute results to differences in beliefs and practices between those seeking and those delivering services), human–biological interactions research, and comparative cultural models research.

14_107-Bernard.indb 670

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

671

One example of midrange theory combined with observational methods in a cultural ecological context is a series of studies, supported by the National Institute on Drug Abuse, of needle sharing and needle hygiene practices. Part of HIV risk-reduction efforts, these studies focus on context-specific uses of injection equipment among drug users in the United States. Early descriptions (Clatts 1994; Koester 1994; Singer et al. 1991) explore the meaning and the processes of injection drug use and needle sharing and the public health consequences of drug paraphernalia laws that restrict the possession of syringes that might be used for drug abuse. Later studies (Bourgois 1995; Needle et al. 1995) explore the consequences of needle hygiene and needle sharing at the micro-environmental level. One applied approach to complex systems is multilevel intervention, a term that refers to the conduct of interventions at multiple social/structural levels simultaneously to maximize effect as Schensul and Trickett describe in a collection of such articles on increasing influenza vaccine uptake (J. Schensul et al. 2009), reducing substance use risk (Berg et al. 2009), and preventing HIV infection among active drug users (Weeks et al. 2009b). Another is dynamic systems modeling (Briody et al. 2010). Although much of this research was originally applied in health related research, recently there have been successful technology transfer applications of this type of research to business and industrial contexts (Jordan 2010). SOCIAL NETWORK THEORIES

Social network theories have evolved over the past 40 years within a number of research contexts germane to applied research (Galaskiewicz and Wasserman 1993; Johnson 1994; Wasserman and Faust 1994). Some of the broader midrange theories associated with these approaches are personal networks theory, social network structural theory, social support theory, and viewpoint theory. Network theory has been used in studies of family systems and adaptation (Bott 1957; Cross 1990), in diffusion studies concerned with the flow of innovation, information, or infection in populations (Kostick et al. 2011; Weeks et al. 2009a), and in studies testing the efficacy of group interventions in natural groups or networks (Nastasi et al. 1997; J. Schensul et al. 1997; Trotter et al. 1996). Several primary methods (or methodological sets) have been used in conducting these studies, including ethnographic network mapping (Trotter et al. 1995b; Trotter et al. 1996), ego-centered network surveys (Johnson et al. 1995; Trotter et al. 1995a), and full relational network analysis (McGrady et al. 1995; Needle et al. 1995; Trotter et al. 1995b). More recently, network analysis has provided an interesting theoretical and methodological model in designing partnerships and collaborative ventures for corporations and universities as well as governments and nonprofit organizations (Briody and Trotter 2008; Trotter et al. 2008; Valente 1995). CRITICAL THEORY

Critical theory in anthropology tends to use a systems approach, but its “value frame” substitutes resistance for adaptation (Freire 1995; Giroux 1981a, 1981b). It calls for examining cultural behaviors at the local level in the context of the political economy of national systems in a global system dominated by nationalistic, capitalist, or other forms of hegemonic control over information and economic and human

14_107-Bernard.indb 671

6/9/14 10:03 AM

672

Robert T. Trotter, II, et al.

resources. Singer and Baer (1995), Hill (1991), and others summarize and integrate cognitive/cultural and behavioral domains into their critical theoretical framework. They have identified the importance of gender, race, ethnicity, and identity in the context of anthropological theory. Other anthropologists and cultural theorists address these areas in greater detail (DeVos and Romanucci-Ross 1995; Heath and McLaughlin 1993; Jordan and Weedon 1995; Morgen 1993; Singer et al. 2006), using a combination of postmodern and critical or Marxist theory. These authors generally view race and ethnicity as socially constructed, with localized meanings influenced by definitions of power and authority that are local manifestations of national or international systems. They argue against static definitions of racial, ethnic, or gender identity, suggesting that these are contested territories and, as such, do not lend themselves to acculturation, gender, or ethnic affiliation scaling techniques unless the tools of measurement are locally situated, constructed, and validated. Critical race, critical Latino, and critical feminist theory have been applied by educational anthropologists to examine structural factors contributing to inequitable distribution or allocation of legal decision making and disposition of cases, inequitable access to health care, and instructional pedagogy (Ginwright and Cammarota 2007; Ladson-Billings 2009; Singer 1989). Although some of these macro theories have been transformed into applied midrange theory (Johnsen 1985; Johnsen et al. 1995), others await testing both as models for applied interventions and for analysis as applicable theoretical models. APPROACHES TO SAMPLING IN APPLIED ANTHROPOLOGY

Applied research can best be conducted when there is an appropriate group of cultural experts who are willing to share their knowledge with the researcher. Midrange theories can only be tested where they are linked to systematic techniques that identify an appropriate representative selection of cultural experts (sampling techniques and key informant selection processes). This is a methodological condition that has been too long ignored in applied anthropological work. Applied projects must be designed to create the highest level of confidence in the research results. To provide this confidence, quantitative social sciences have most commonly favored probabilistic (random) sampling techniques that allow for statistical analysis of the data collected. These techniques work well when the universe from which the sample is to be drawn can be identified and where everyone in a population (a school, a town, a country) has an equal chance of being chosen to express their viewpoint. It does not work for qualitative approaches (cf. Trotter and Medina Mora 2000; Trotter 2012), where other conditions apply (J. Schensul and LeCompte 20l0/2012). For ethnographic and general qualitative research purposes, it is frequently scientifically inappropriate to draw simple random samples or to use random sampling procedures without using other techniques that are guaranteed to produce a qualitatively representative sample of information about the culture, not just a sample of people representing the group. For example, random sampling in multiethnic neighborhoods, when the target of study is a specific ethnic group, may fail to produce a representative sample of members of that ethnic group. Instead, cultural experts (i.e., individuals who are nominated as having culturally representative knowledge in their domain of expertise) must

14_107-Bernard.indb 672

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

673

be identified through ethnographic informant selection procedures (Johnson 1990) or cluster or stratified random sampling techniques, or must follow recent developments in systematic qualitative sampling (Trotter and Medina Mora 2000; Trotter 2012) that provide strong defenses of purely qualitative sampling for applied anthropology. Probabilistic (quantitative) sampling techniques are designed to be used in projects where statistical analysis is the central analysis strategy. However, there are numerous conditions in which randomized or systematic quantitative sampling and quantitative surveying cannot be met. There are other conditions where probabilistic sampling may be contra-indicated because they do not meet the standards for systematic ethnographic research or the research questions may call for the discovery of cultural patterns or range of variation at the cultural (group) but not at the individual level (Morse et al. 2002). Alternative approaches to sampling must often be used because the target population is hidden, rare, or difficult to find (Luborsky and Rubenstein 1995; J. Schensul et al. 1999). Any systematic approach to research has to specifically tie the theoretical framing of the research to the analytical strategy that will be used to analyze and interpret the data. The following sections provide options to probabilistic sampling in applied ethnographic research contexts.

Systematic Sampling Procedures Because we want to identify cultural domains or variations in cultural patterning, much of the ethnographic sampling process focuses on cultural expertise (Trotter and Medina-Mora 2000, Trotter 2012). When we are exploring cultural knowledge, it’s important to talk to individuals who are carefully selected for their expertise in that area, rather than being randomly selected from the general population. To explore cultural domains, or for cultural consensus purposes, saturation sampling (Polkinghorne 2005; Sandelowski 1986) is often preferable and sufficient (i.e., interviewing a succession of individuals to the point where no new information is obtained). Saturation sampling depends on having sufficient information about the social setting to be able to identify key informants who represent the widest possible or anticipated range of views on the topic under investigation. For example, if the topic is micro-economic enterprises for women, key informants should include women involved in such enterprises, trainers, policymakers, representatives of lending agencies, design experts, educators, and community opinion leaders. In bounded settings such as small communities, schools, or residential buildings, it is possible to conduct a census in which everyone or almost everyone is interviewed. This is called universal interviewing and it depends on the assumption of a bounded cultural system or social group. Saturation, redundancy, and universal interviewing frequently require multiple interviews with individual informants rather than relying on single interviewing. Some advocates of qualitative research believe that a single interview is sufficient to capture comprehensive information from individual informants about a cultural domain especially at the cultural level. We don’t agree, since single interviews have problems in full saturation and validity; conducting repeated interviews with the same individuals builds trust and reflective capacity as well as producing consistency of account. In addition, saturation sampling does not apply solely to individual respondents. Sampling units can be specific events,

14_107-Bernard.indb 673

6/9/14 10:03 AM

674

Robert T. Trotter, II, et al.

event sequences, classroom activities, or observations at scheduled time points. The most important consideration is not whether the sampling unit is an individual or some other unit in time and space, but on what basis the sampling unit is determined in relation to overall research design and methods. One of the common alternative forms of universal interviewing is interviewing the entire population of consensus experts for a particular cultural domain (i.e., interviewing anyone who is identified by the community as an expert in a particular area of culture). This approach has been the norm in small community research projects involving prolonged stays and repeated exposures to most individuals in that context. It is also used when the number of people (cultural experts) in a study setting is small enough so that each individual can be interviewed/observed at least once and a portion can be interviewed multiple times. Interviewing the whole population on a particular topic is an alternative for ensuring a representative sample and eliminates the need for complicated random or selective sampling designs. A second form of saturation sampling is interviewing to “sufficient redundancy.” This involves reviewing interviews and determining that there is very little variation in response, requiring a small number of interviews. Alternatively, where issues are complex with wide divergence in viewpoints, researchers may interview until there are obvious diminishing returns on new thematic discoveries with each subsequent interview (Polkinghorne 2005). We would not assume that the full scope of responses has been covered in a multiethnic or multigenerational community, for example, unless we were sure that our sampling procedure covered all known or suspected sources of significant variation on the topic. What we cannot know is the range of variation among individuals in the population in terms of how they relate to that cultural domain. Answering that question falls within the scope of probability sampling. The classic approaches to ethnographic sampling are well explained by Johnson (1990), who thoroughly explores the similarities and differences between probabilistic sampling used in surveys and experimental designs compared with the purposive sampling strategies necessary for successful qualitative research.

Nominated Expert Sampling Nominated expert sampling is a classic ethnographic or qualitative approach to exploring culture and social meanings in various populations, communities, and cultural groups. It is theoretically supported by a consensus theory approach to sampling and sample size (see Romney et al. 1986). The sample recruitment process is to identify all of the consensus experts (i.e., those nominated by multiple other individuals in a community) who have the most extensive expertise in a specific area of social or cultural knowledge. The gold standard is to recruit the entire expert sample to provide a saturation level of information about the targeted research topic. Since experts tend to agree about the vast majority of their subject area, and provide virtually the entire explanation of the variability in expert views (they know each other’s agreements and disagreements), this provides an in-depth investigation of a topic that is qualitatively valid, reliable, and culturally generalizable. J. Schensul et al. (1999) suggest that a sample size of between 12 and 20, if properly selected, is sufficient to provide a valid and reliable explanation for most cultural phenomena.

14_107-Bernard.indb 674

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

675

Snowball Sampling and Respondent-Driven Sampling In snowball sampling, each person interviewed connects the researcher to the next person or persons, based on a designated set of criteria. The result is the continuous accrual of related research respondents. It is an important instance of chain analysis (Bieleman et al. 1993; Diaz et al. 1992). This technique allows the researcher to build a sample of individuals with one or more common characteristics within a large known or unknown universe of individuals, all of whom may not, as a group, share the behavior or cultural element in question. As an example, it has been used to study drug subcultures such as injection drug users in the Northeast (Mosack et al. 2005) or cocaine users in the Netherlands (Bieleman et al. 1993). In many of these studies, drug users served as case-finding agents, introducing to the researchers to a growing sample of other users. A special case of snowball sampling, respondent-driven sampling (Heckathorn 1997, 2002; Scott 2005) is now in wide use for creating samples in hidden or hard-toreach populations. One of the strengths of these forms of nominated sampling is that they can be designed (Snijders 1992) to be statistically sound approaches for estimating large population characteristics and comparing those conditions to conditions derived from probabilistic/random samples. Targeted Sampling Targeted sampling is designed for situations where it is difficult to identify the universe of units from which to draw a sample (e.g., commercial sex workers, homeless youth, undocumented household workers, or school dropouts). In these examples, key parameters—including the need to remain hidden—prevent the use of traditional random sampling procedures. In such cases, targeted sampling (Watters and Biernacki 1989) is an appropriate substitute. Targeted sampling is a systematic technique for creating a proxy sampling framework that ensures that the major divisions or categories of the population being studied are systematically sampled in theoretically correct portions. It uses all of the available secondary data relating to the population to create geographically focused targeted sampling areas. These data may include health information, social service data, and ethnographic knowledge of the population, observations, or any secondary data sources that describe some important segment of the target population. New advances in respondent-driven sampling have been amalgamated with basic targeted sampling strategies to advance appropriate sampling in hidden populations (Heckathorn 2002; Scott 2005). METHODS IN APPLIED ANTHROPOLOGY

Anthropological research typically uses a variety of data collection techniques—including observations, interviews, focused life stories, discussion groups, the analysis of social networks, decision modeling, projective techniques, and household surveys—to gain detailed knowledge about cultural contexts, patterns of social behaviors, sequences of events, and cultural norms or beliefs. There are many resources on methods of ethnographic data collection and analysis. The classics include Pelto and Pelto’s (1978) pioneering work, Anthropological Research: The Structure of Inquiry; Bernard’s (2011)

14_107-Bernard.indb 675

6/9/14 10:03 AM

676

Robert T. Trotter, II, et al.

Research Methods in Anthropology; the two-volume series by Werner and Schoepfle (1987) entitled Systematic Fieldwork; Weller and Romney’s (1988) Systematic Data Collection; Strauss’s (1985) Qualitative Analysis for Social Scientists; and Strauss and Corbin’s (1990) Basics of Qualitative Research. These works contain descriptions of research design and methods for participant observation and advanced ethnographic data collection. They are complemented by Miles and Huberman’s (2002) Qualitative Data Analysis. Many specific ethnographic methods—decision tree modeling, cognitive methods (pile sorts, triad tests, etc.), the long interview, the life history—are described in a continually expanding series of monographs published by Sage Publications. The most in-depth general anthropological methods publications are Schensul and LeCompte’s Ethnographer’s Toolkit (2010/2012), and Bernard’s Research Methods in Anthropology (2011). The following sections identify the methods commonly used in applied anthropology, especially educational, medical, business, and urban anthropology.

Analysis of Culturally Defined Cognitive Systems Cognitive anthropologists have developed methods that allow us to explore how people think about and locate meaning in the world around them. As we noted earlier, it is frequently important to consider cognition and behavior in applied contexts since the relationship between them is not always predictable. Change agents often recognize the need to simultaneously influence behavior by changing attitudes, beliefs, norms, values, motivations, and intentions. There is a well-developed set of basic cognitive anthropology techniques that have been used in cross-cultural research. In this section, we discuss cognitive research methods that: (1) assist in determining the content and limits of cultural domains; (2) help us analyze the structural elements of cultural domains; (3) portray a domain from a consensual framework; and (4) identify key informants and domains or population sectors for future research (Kostick et al. 2011). Cultural Models Approaches The cultural models methods of Quinn and Holland (1987) and Kleinman (1980; Kleinman and Benson 2006) are solid starting points for cross-cultural cognitive research. These approaches provide systematic questions to investigate broad cultural domains, such as models of health conditions, policy issues, and values systems. Other ethnographic cognitive methods include systematically administered, semi-structured, and open-ended (qualitative or ethnographic) interviews analyzed through hierarchical coding and pattern recognition of themes and conceptual linkages (Weller and Romney 1988; Werner and Schoepfle 1987) and systematic data collection techniques (sometimes referred to as elicitation techniques) such as pile sorts and triad tests, borrowed from cognitive psychology and linguistics (Bernard 2011; J. Schensul and Lecompte 2010/2012). Determining the Content and Limits of Cultural Domains Free listing is the technique most commonly used to begin the exploration of cognitive domains. The basic format is to ask a set of respondents to list and describe all the

14_107-Bernard.indb 676

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

677

things that are part of a particular domain. The ethnographer records and probes unexpected or unfamiliar responses in detail (including new words and phrases or words used in new ways) since these labels provide a window into unknown concepts, beliefs, or behaviors. Free lists provide information and local vernacular that can be used in culturally specific questionnaire construction, written educational materials, or behavioral exercises that are being constructed to meet intervention or health education goals. They also allow us to differentiate between key subdivisions in the populations since the domains can differ significantly by gender, ethnicity, age, and cultural expertise. This gives researchers the ability to assess intra- and intercultural variation within the same geographic region, across the nation, or around the world. Some of the more sophisticated uses of free-listing data treat these nominal or categorical data as variables that can be used in statistical procedures to provide more extensive explorations of the relationships among informants or among the elements in a cultural domain (see Bernard 2011; Weller and Romney 1988). Techniques similar to free listings, such as exploratory open-ended questions, Spradley’s domain analysis techniques (1979), or sentence-completion processes (D’Andrade et al. 1972) can also be analyzed using the approaches described for free listings. These rapid-scanning techniques can be used as an individual exercise in a face-to-face interview or in group settings (as a form of focus group). Empiricists tend to use the individual interviewing technique so as to avoid contamination by other informants, while constructivists often use the group interview (including the focus group) to enable observation of negotiated meaning.

Techniques to Define and Analyze the Structural Relationships among Elements in a Cultural Domain Research methods in cognitive anthropology that allow a researcher to explore the relationships among the elements of a cultural domain include pile sorts (Weller and Romney 1988), triads tests (Weller and Romney 1988), and sentence frame techniques (D’Andrade et al. 1972; Weller and Romney 1988). Each of these techniques begins where free listings leave off. They start with the elements of a well-defined cultural domain (explored through free listings). The researcher explores the relationships among the key elements of that domain by asking informants to make judgments about the similarities and differences of the items in the domain to one another. One such technique is a pile sort, a rapid assessment technique that uses visual aids to let informants create either free or constrained (predefined) classifications of elements within a cultural domain. The most common method is to place pictures, real objects, written labels, or combinations of the three on cards. Each card represents one element in the domain being studied. The researcher asks the informant to classify all the elements by stacking the cards into piles defined by one or more common elements. The final groupings represent each individual’s classification system for items in the domain. Weller (this volume) describes the pile sort method in detail. Handwerker and Borgatti (this volume) describe multidimensional scaling and cluster analysis, two of the methods most commonly used in analyzing pile sort data. Weller and Romney (1988) and Bernard (2011) show how these techniques can be integrated into ethnographic research. The

14_107-Bernard.indb 677

6/9/14 10:03 AM

678

Robert T. Trotter, II, et al.

most commonly used computer programs for the analysis of pile sort data are ANTHROPAC (Borgatti 1990), UCINET (Borgatti et al. 2006), and both SPSS and Excel.

Consensus Modeling In consensus modeling, an ethnographer can identify a consensus-based description of a cultural domain while simultaneously assessing individual informants’ expertise in that domain (Romney et al. 1986) based on the assumption that “cultural truth” and informant accuracy are derived from a probabilistic model of culture. Investigators need to know the accuracy of the information they receive in self-reports from informants, and consensus theory provides one type of answer to these questions. Consensus theory is designed to work with a set of questions, all in the same format, all on the same topic, and all at the same level of difficulty. The goal, then, is to estimate the best set of culturally appropriate answers to the questions. The formal model (Romney et al. 1986) can accommodate categorical-type responses: single words or short-phrase responses to open-ended questions, or closed-ended multiple choice responses (including true/false or yes/no). An informal version of the model can accommodate interval or fully ranked data. For an example of the use of the formal model with yes/no responses, see Weller et al. (1993); for the informal model with ranked responses, see Chavez et al. (1995). Analysis of Social Structures Anthropologists and other social scientists have long been interested in the effects of social structure on human survival and social interaction. The anthropological literature is filled with information about kinship organization, voluntary associations, and formal organizations found in cultures around the world. Modern network analysis is a set of techniques for expanding our knowledge of the effects and dynamics of human social organization. These techniques are used in the study of kin-based and nonkinbased networks (see Edwards [2010], Marsden [1990], and Scott [1991] for reviews of methods and analytical issues in network analysis). Increasingly, applied anthropological research has involved the examination of informal and formal human networks. Ethnographic research in this area commonly focuses on three kinds of networks: specialized interactive networks (e.g., sex or drug use/risk networks), ego-centered networks, and full relational networks. Each involves different methods of data collection. Research questions drive the choice of which type of network to investigate and which type of data to collect. Ethnographic Data on Specialized Networks Since social networks are the basis for social activity in a community or institutional setting, one effective method for identifying local social networks is through ethnographic interviewing. Interviewers ask respondents to identify clusters, networks, cliques, or other kinds of groups in which individuals are related to one another (see Bott 1957) or are connected by some form of cultural interaction. The characteristics of the networks defined through ethnography can be used to create a typology or classification of the types of social relationships that exist in a culture and the groupings by size, class, gender, ethnicity, income, family, or other demographic characteristics that they represent.

14_107-Bernard.indb 678

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

679

Curtis et al. (1995) used ethnography to define initial networks in their study of urban street-level drug markets. Ethnography was used to describe the activities of network members and to contextualize networks in the street drug economy. They followed up with a formal survey to confirm social ties among network members to assist the development of street based drug interventions. Another applied example of this kind of research in an industrial/business context involved the ethnographic and social network analysis of industry–university research partnerships (see Briody and Trotter 2008; Trotter et al. 2008 for models of successful partnership interactions). A summary of qualitative network approaches and comparative qualitative methods that are useful for applied projects can be found in Fischer (2011).

Ego-Centered Network Data Ego-centered network analysis is based on an individual’s definition of the individuals connected to him or her by specified social relationships which are assessed as network attributes (Knoke and Yang 2008). The attributes of ego-centered networks (size, gender, and ethnic composition, etc.) and characteristics of those networks themselves (density, intensity, etc.) can be incorporated into “typical” network profiles, which can then be analyzed in association with other psychosocial variables (Wasserman and Faust 1994; Watts 1999). Ego-centered networks can provide the basis for determining specific influences on ego, which can then be used in interventions. The social support literature examines ego’s networks to manage chronic health problems as well to understand risk-taking behavior and to provide direction for drug and HIV intervention programs (Latkin et al. 1995). Full Network (Relational) Data Ethnographic and ego-centered network approaches provide valuable baseline data for intervention strategies (Trotter et al. 1996), but they don’t always provide detailed information about the type, strength, or direction of relationships within networks. This type of data emerges from the analysis of reciprocal relationships among all members of a network. Scott (1991) describes macro-network data collection, analysis, and use. The collection of data on full or macro-networks is costly and time consuming, so there is continuing interest in the development of techniques to approximate full networks from data on partial networks. Klovdahl (1989), Klovdahl et al. (1994), and McGrady et al. (1995) discuss procedures, advantages, and disadvantages of sampling in constructing macro-networks. THICKER DESCRIPTION

Applied anthropologists use many other approaches for collecting data. Here, we review several approaches, including sociogeographic mapping, group interviews, ethnographic surveys, and rapid assessment.

Sociogeographic Mapping Anthropologists have always used maps in field research in part, at least, because early fieldwork was often conducted in places where there were no maps. It was important to bound communities and demarcate residential and other structural units in

14_107-Bernard.indb 679

6/9/14 10:03 AM

680

Robert T. Trotter, II, et al.

relation to one another. Mapping “the community”—whether a classroom, organization, neighborhood, or invasion—is still highly recommended. The process of making a community map helps researchers draw a household sample, generate hypotheses about social relationships among households and between households and other social units, and observe changes over time, especially with respect to household/land and other environmental use patterns. Computerized mapping programs and national and state GIS (geographic information systems) allow for relatively quick mapping of virtually any data across space and over time. Geographic mapping of social networks by residence of network members in relation to primary points of interaction in the community (defined through ethnographic observation) can be used to frame the location of interventions based on natural patterns of spatial use. As an example, the Institute for Community Research was able to show patterns of intra-neighborhood, inter-neighborhood, and intercommunity mobility for each of Hartford’s neighborhoods by mapping residential locations over time. The demographic data, portrayed visually, were immediately usable by educational policymakers for school-based planning (ICR 1993). GIS applications are useful for describing the arrangement of social variables in geographic space, for hypothesis testing, and for eliciting cognitive responses to research-driven questions reflected in such data. A number of emerging software resources allow researchers to overlay social and cultural information on traditional geographical maps and variables, which provides for some very interesting interactive analytical capabilities using spatial statistics in conjunction with common demographic analysis (J. Schensul et al. 1999).

Group Interviews Group interviews yield text data for coding and analysis, which can be treated quantitatively and qualitatively. Group interviews can be used for many purposes—for example, to collect information on a cultural domain, to develop listings for pile sorts, to identify the range of variation in opinions or attitudes on a set of topics, to collect simple numerical data on reported experiences, or to react to the results of previously collected data. We use the term “focused” to refer to formal and informal interviews that are intended to gather ethnographic information on any topic that lends itself to group discussion. More recently, focus groups have been used to study knowledge, attitudes, and beliefs in many social situations. One disadvantage of group interviews is that they are limited to topics that people are willing to discuss in public. Topics considered personal or intimate should be avoided in group interviews or depersonalized. However, focus group interviews are useful for orienting yourself to a new field of study, for generating hypotheses based on informants’ insights, for evaluating different research sites or study populations, for developing individual questions for interview schedules and questionnaires, and for getting participants’ interpretations of results from earlier studies (Khan et al. 1991). They can produce a lot of data in a short time from a larger number of people than would be possible by interviewing key informants. They tend to produce good “natural language discourse,” which allows the researcher to learn the communication patterns in the community rapidly. Krueger and Casey (2000) and Onwuegbuzie et al. (2009) offer new methodological considerations for optimizing information flow in group interviews.

14_107-Bernard.indb 680

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

681

The focus group technique appeared in the 1930s as an alternative to direct interviews and became popular as a method for qualitative research in marketing. Technically, focus groups are used to obtain audience response to designs, products, interventions, or data dissemination. Anthropologists are more likely to use more informal, open-ended, and exploratory approaches to group interviews, including interviews with groups in the field, than to use formal group interviews to test ideas, interventions, and products.

Ethnographic Surveys Ethnographic surveys are often an integral part of strong mixed methods approaches to applied research, and have a couple of advantages over generic surveys applied to special groups (Creswell 2009). First, ethnographic surveys are based on prior experience in a specific field situation. They may incorporate instruments or questions drawn from other studies, including nationally validated instruments, but their strength is their validity in relation to local culture and the construction and testing of midrange theory. Thus, ethnographic surveys measure constructs known to be relevant to, or understood by, the study population. Second, they are most commonly administered in a face-to-face interview (preferred because it is more intimate). Self-administration is not possible with nonliterate respondents or to those unfamiliar with answering written questions. Rapid Ethnographic Assessment Rapid ethnographic assessment techniques have been developed for situations where there is a strong need for ethnographic data but little time to conduct a full ethnography. These “rapid scanning” techniques are also called rapid ethnographic procedures, rapid rural appraisals, focused ethnographic studies, and brief ethnographies (Handwerker 2001; Scrimshaw 1992; Scrimshaw and Gleason 1992). There have been a number of critiques of the original rapid assessment approaches, which have led to substantial improvements in the basic methodological design of rapid assessment, response, and evaluation designs (Needle et al. 2000; Trotter et al. 2001; Trotter and Singer 2005). Most of these techniques share the following characteristics: 1. They are narrowly focused—for example, one disease category, or one cultural domain. 2. They are problem oriented—intended to provide time-sensitive evidence to address a current issue. 3. They are participatory; local partners include potential users, who, at the same time, can provide ethnographic insights otherwise obtained through more time-intensive participant observation and in-depth interviewing. 4. They provide techniques for rapid sampling of representative sectors. 5. They use small sample sizes and fit community contexts and conditions. 6. They do not pursue intracultural complexity or range of variation; instead, they focus on cultural patterning and on gross differences across sectors of populations and service providers. 7. They use a systems perspective, making sure to collect information from all relevant sectors of the community. 8. They use cognitive techniques to identify and assess cultural domains. 9. They generally do not (but can) make use of quantitative sampling or survey techniques.

14_107-Bernard.indb 681

6/9/14 10:03 AM

682

Robert T. Trotter, II, et al.

Typical rapid ethnographic assessments are modest in cost and duration. The assessments include mapping, brief participant observation in targeted cultural domains and spaces, free listing and pile sorting or other systematic elicitation methods, key informant interviewing, and group interviewing (Trotter et al. 2001). Rapid assessment and focused ethnography require extensive prior knowledge of the culture in question as well as a prior determination of all of the important sectors contributing to the problem, from which researchers can draw representative samples of key informants and focus group respondents. This determination can be made either by ethnographers familiar with the setting or by the interdisciplinary/intersectoral team responsible for the study and its uses. Challenges include the need to develop an accurate understanding of the problem and its context in a relatively short and cost-effective period of time, the need for systems for transforming the data into satisfactory solutions, and the need to produce socioculturally acceptable solutions (Needle et al. 2000). These three requirements characterize much of applied research, but the protocols for the conduct and utilization of brief ethnographies are still not widely known and accepted. Thus, both researchers and contractors/clients take risks when using brief ethnographies for programs, and especially for policy-related purposes. RESPONDENT ACCURACY—VALIDITY AND RELIABILITY

Informant accuracy is a critical issue for designing and conducting ethnographic research. In survey research, data reliability and validity depend on the consistency of self-report data and studies in which survey responses are checked against sources of information that are known to be correct (laboratory tests, health records, or mechanically measured data). The validity of the responses of ethnographic informants i.e., cultural experts) is assessed by other criteria (Bernard 2011; Bernard et al. 1984; Kirk and Miller 1986). Ethnographic field research depends on developing close personal relationships with members of a community over time. It emphasizes the rapport established between the researcher and the respondent. Multiple interviews with the same individual, as well as the increasing intimacy of the ethnographer–informant relationship, is expected to produce increasingly accurate information, although in some cases cultural values interfere with this process (Blimes 1975; Nachman 1984). Field research offers the potential for repeat interviews with the same respondent during the study period. These interviews, both formal and informal, are opportunities to look for narrative inconsistencies, recheck and verify data, and clarify previous statements. Ethnographic data are heavily based on individual perception, memory, and self-report through life histories, cultural process interviews (Pelto and Pelto 1978), narratives, or stories (Florio 1997) as well as elicitations (J. Schensul and LeCompte 2010/2012). Each approach raises challenges to questions about the accuracy of recall and the veracity of individual informants. Individuals vary in their level of expertise and in their ability to accurately recall information about the things that have happened to them. Some are highly accurate in describing unique events; others are more accurate in describing repeated events. Some informants have narrowly defined or specialized expertise; others are knowledgeable about a range of cultural domains but their depth of knowledge on any single topic is limited. Sometimes it is important to

14_107-Bernard.indb 682

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

683

interview “special” people in a culture; at other times, it is best to talk to “typical” or representative samples of people. (Bernard et al. 1980, 1982; Freeman and Romney 1987; Romney et al. 1986; Weller 1984). We believe that the most effective way to ensure reliability and validity of ethnographic data is to obtain comparable, confirmatory data from multiple sources at different points in time, and through the use of multiple methods. This is the process of triangulation. Many investigators, however, now consider ethnographic self-reporting as a form of narrative or storytelling in which the individual interviewed is attempting to convey a particular impression or image to the researcher (Marcus and Fischer 1986). The story itself must be situated historically, contextually, and in the life of the storyteller to understand it as data. Extensive and informative discussions of qualitative reliability, replicability, validity, and generalizability have been introduced in recent publications on systematic qualitative research that are also valuable to applied anthropological methods (Morse et al. 2002). ETHNOGRAPHIC FIELD TEAMS AND CROSS-SITE/CROSS-NATIONAL APPLIED RESEARCH

Ethnographic teamwork requires proper and constant management of team members and the data they are collecting. Some of the best examples for ethnographic team development come from multisite medical anthropology projects for the Centers for Disease Control and Prevention and the National Institutes of Health (Gust and MacQueen 2008). These projects demonstrate that proper management calls for attention to comparability of interview and observation skills across interviewers, careful group construction of coding systems related to the theoretical framework of the study and the field situation, regular monitoring and feedback with respect to field notes, and attention to entering and coding of data. Furthermore, since good field research involves interaction with the data and both deepening and expanding of text codes, continuous analysis of incoming data with the field team is important. Ethnographic field teams encounter situations comparable to cross-national ethnographic research when team members work in settings marked by differences in ethnic culture. In both instances, meaning systems, contexts, and social interactions may vary, reducing the comparability of coding systems. Investigators who engage in cross-site or cross-national studies need to pay very close attention to the construction of comparable coding categories across settings and to the possibility that some phenomena may be unique to each setting (MacQueen et al. 2001). One solution to the problem of cross-site comparability is to assume that common research methods will produce unique cultural responses in each setting that point to differences as well as possible similarities in approach. Most of these projects also utilize ethnographic computer data management and analysis software, which has been available for some time now to assist in these processes and to give teams of researchers the ability to systematize the collection, coding, data retrieval, and basic analysis of text data (MacQueen et al. 1999; McLellan et al. 2003). The most popular current programs are NVivo, NUD*IST, and ATLAS-ti. However, new ones are added frequently, old ones change or are eliminated, and there is a consist evolution of available software packages that should be evaluated for each new applied ethnographic project.

14_107-Bernard.indb 683

6/9/14 10:03 AM

684

Robert T. Trotter, II, et al.

SUMMARY AND CONCLUSIONS

We have presented an overview of the theories, informant selection processes, and research methods used in applied anthropology. Though our examples are based primarily on applied research conducted in the United States, several are drawn from other countries, and the methodology we outline is relevant in a global setting. Many of the methods are common to all cultural anthropology, but some, like rapid assessment, consensus theory modeling, cultural models, social network analysis, and systems theory are more commonly used in applied anthropology. All applied methods must have a strong theoretical foundation that relates to the purpose of an applied project, to the theory of cultural meaning that is intended to guide the project, and to the plan for change being attempted. Applied research methods start with this theory and evolve through the interaction between expressed needs in the field, the literature on the issue, and the deepening research experience of the anthropologists and their partners in the field. The evolution of the theory, by definition, involves the collaboration of “partners in change,” for without partners for whom the designated direction of change is centrally meaningful, the research will remain on the shelf. We have argued for consideration of the appropriate selection of individuals who will be able to access data critical to the success of the project. Sometimes such individuals come from the community wishing to initiate cultural change; sometimes they are from the same designated group but from another geographic location; more often than not, they are other anthropologists of different backgrounds who are committed to sharing their research and social skills to improve the quality of life in a community. Each of these individuals can serve an applied research project well. Once the initial stages of project development have occurred, a multifaceted set of research and development methods driven by clearly articulated theory can be put into place. These methods are intended to do two things simultaneously: (1) produce sound data using rigorous research methods that are convincing and can guide change efforts; and (2) maintain close working relationships with partner communities committed to the change process. Both are necessary. The means should always be consistent with the desired ends to avoid the contradictions that can so easily destroy a project even before it begins. Whenever possible, there should be a full partnership of researchers and interventionists or change agents in all aspects of the research. This is important to ensure full use of the results and to ensure that people who know best how to make and act on decisions to bring about the desired change are also intimately familiar with the research that is intended to guide it. Finally, as applied anthropologists, we should never forget that in addition to methodological flexibility, the greatest strength of our field is that it provides ethnographers with the methods and tools to understand culturally based needs, values, perceptions, beliefs, knowledge, models, and reasons for behavior—and to use these, preferably in partnership with those whose issues are of concern, for designing programs of change. Even with the best intentions of all partners to change, it is only with the use of sound research methodology that a change effort is likely to result in long-term success.

14_107-Bernard.indb 684

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

685

REFERENCES

Barlett, P. F. 1977. The structure of decision making in Paso. American Ethnologist 4: 285–308. Bartunek, J. M., and M. R. Louis. 1996. Insider/outsider team research: Qualitative research methods #40. Thousand Oaks, CA: Sage. Bearison, D. J. 1982. New directions in studies of social interaction and cognitive growth. In Social cognitive development in context, ed. F. C. Serafica, 199–221. New York: Guilford Press. Berg, M., E. Coman, and J. J. Schensul. 2009. Youth action research for prevention: A multi-level intervention designed to increase efficacy and empowerment among urban youth. American Journal of Community Psychology 43: 345–59. Berger, P. L., and T. Luckman. 1966. The social construction of reality: A treatise in the sociology of knowledge. Garden City, NJ: Doubleday. Bernard, H. R. 2011. Research methods in anthropology, 5th ed. Thousand Oaks, CA: Sage. Bernard, H. R., P. D. Killworth, and L. Sailer. 1980. Informant accuracy in social network data IV: A comparison of clique-level structure in behavioral and cognitive network data. Social Networks 2: 191–218. Bernard, H. R., P. D. Killworth, and L. Sailer. 1982. Informant accuracy in social-network data V. An experimental attempt to predict actual communication from recall data. Social Science Research 11: 30–66. Bernard, H. R., P. Killworth, and L. Sailer. 1984. The problem of informant accuracy: The validity of retrospective data. Annual Review of Anthropology 13: 495–517. Bieleman, B., A. Diaz, G. Merlo, and C. H. Kaplan. 1993. Lines across Europe: Nature and extent of cocaine use in Barcelona, Rotterdam and Turin. Amsterdam: Academic Publishing Division, Swets & Zeitkinger. Blimes, J. 1975. Misinformation in verbal accounts. Some fundamental considerations. Man 10: 60–71. Borgatti, S. 1990. ANTHROPAC 3.2. Columbia: Analytic Technologies. Borgatti, S., M. Everett, and L. Freeman. 2006. UCINET for Windows: Software for Social Network Analysis. Harvard, MA: Analytic Technologies. Bott, E. 1957. Family and social network: roles, norms and external relationships in ordinary urban families. London: Tavistock. Bourgois, P. 1995. In search of respect: Selling crack in El Barrio. New York: Cambridge University Press. Briody, E. K., and R. T. Trotter, II, eds. 2008. Partnering for performance: Collaboration and culture from the inside out. New York: Roman & Littlefield. Briody, E. K., R. T. Trotter, II, and T. L. Meerwarth. 2010. Transforming culture: Creating a better manufacturing organization. New York: Palgrave-Macmillan. Bronfenbrenner, U. 1977. Toward an experimental ecology of human development. American Psychologist 32: 513–31. Campbell, D. T., and J. C. Stanley. 1963. Experimental and quasi-experimental designs for research. Boston: Houghton Mifflin. Castillo, C. O., and J. H. Geer. 1993. Ambiguous stimuli: Sex in the eye of the beholder. Archives of Sexual behavior 22: 131–43. Chavez, L. R., F. A. Hubbell, J. M. McMullin et al. 1995. Structure and meaning in models of breast and cervical-cancer risk-factors: A comparison of perceptions among Latinas, Anglo women, and physicians. Medical Anthropology Quarterly 9: 40–74. Clatts, M. C. 1994. All the king’s horses and all the king’s men: Some personal reflections on ten years of AIDS ethnography. Human Organization 53: 93–95.

14_107-Bernard.indb 685

6/9/14 10:03 AM

686

Robert T. Trotter, II, et al.

Creswell, J. W. 2009. Research design: Qualitative, quantitative, and mixed methods approaches, 3rd ed. Thousand Oaks, CA: Sage. Cross, W. J., Jr. 1990. Race and ethnicity: Effects on social networks. In Extending families: The social networks of parents and their children, ed. M. Cochran, D. Larner, L. Riley et al., 67–85. New York: Cambridge University Press. Curtis, R., S. R. Friedman, A. Neaigus et al. 1995. Street-level drug markets: Network structure and HIV risk. Social Networks 17: 229–49. Dale, J. A., and E. J. Hyslop-Margison. 2010. Paulo Freire: Teaching for freedom and transformation: The philosophical influences on the work of Paulo Freire. New York: Springer. D’Andrade, R., N. R. Quinn, S. B. Nerlove, and A. K. Romney. 1972. Categories of disease in American-English and Mexican-Spanish. In Multidimensional scaling, Vol. 2, ed. A. K. Romney, R. N. Shepard, and S. B. Nerlove, 11–54. New York: Seminar Press. DeVos, G., and L. Romanucci-Ross. 1995. Ethnic identity: A psychocultural perspective. In Ethnic identity: Creation, conflict and accommodation, ed. L. Romanucci-Ross and G. DeVos, 349–79. Walnut Creek, CA: AltaMira. Diaz, A. M. Barruti, and C. Doncel. 1992. The lines of success? Study on the nature and extent of cocaine use in Barcelona. Barcelona: Laboratory of Sociology, ICESB. Dressler, W. W., C. D. Borges, M. C. Balieiro, and J. E. dos Santos. 2005. Measuring cultural consonance: examples with special reference to measurement theory in anthropology. Field Methods 17: 331–55. Dryfoos, J. G. 1990. Adolescents at risk: Prevalence and prevention. New York: Oxford University Press. Edwards, G. 2010. Mixed-method approaches to social network analysis. ESRC National Centre for Research Methods Review paper. University of Manchester National Centre for Research Methods NCRM/015. Ervin, A. M. 1996. Collaborative and participatory research in urban social planning and restructuring: Anthropological experiences from a medium-sized Canadian city. Human Organization 55: 324–33. Ervin, A. M. 2004. Applied anthropology: Tools and perspectives for contemporary practice. New York. Pearson/Allyn & Bacon. Fetterman, D. M., L. Romanucci-Ross, and G. DeVos. 1993. Speaking the language of power: Communication, collaboration and advocacy (translating ethnography into action). Washington, DC: Falmer Press. Field, L., and R. G. Fox, eds. 2007. Anthropology put to work. New York: Berg. Fischer, M. 2010. Social network analysis and qualitative comparative analysis: Their mutual benefit for the explanation of policy network structures. Methodological Innovations Online 6: 27–51. DOI: 10.4256/mio.2010.0034. Florio, S. 1997. To tell a new story: Reinventing narratives of culture, identity and education. Anthropology and Education Quarterly 28: 152–62. Freeman, L. C., and A. K. Romney. 1987. Words, deeds, and social structure: A preliminary study of the reliability of informants. Human Organization 46: 330–34. Freire, P. 1995. Pedagogy of the oppressed. New York: Continuum. Gagnon, J. H., and W. Simon. 1987. The scripting of oral-genital sexual conduct. Archives of Sexual Behavior 16: 1–25. Galaskiewicz, J., and S. Wasserman. 1993. Social network analysis: Concepts, methodology, and directions for the 1990s. Sociological Methods & Research 22: 3–22. Garro, L., and C. Mattingly, eds. 2000. Narrative and the cultural construction of illness and healing. Berkeley: University of California Press.

14_107-Bernard.indb 686

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

687

Ginwright, S., and J. Cammarota. 2007. Youth activism in the urban community: Learning critical civic praxis within community organizations. International Journal of Qualitative Studies in Education 20: 693–710. Giroux, H. A. 1981a. Dialectics and the development of curriculum theory. Ideology, culture, and the process of schooling. Philadelphia: Temple University Press. Giroux, H. A. 1981b. Paulo Freire’s approach to radical educational theory and practice. In Ideology, culture, and the process of schooling, ed. H. A. Giroux, 127–42. Philadelphia: Temple University Press. Giroux, H. A. 2010. Rethinking education as the practice of freedom: Paulo Freire and the promise of critical pedagogy. Policy Futures in Education 8: 715–21. Gladwin, C. H. 1989. Ethnographic decision modeling. Newbury Park, CA: Sage. Glaser, B. G., and A. L. Strauss. 1967. The discovery of grounded theory. Chicago: Aldine. Greenwood, D., and M. Levin. 2007. Introduction to action research. Thousand Oaks, CA: Sage. Grier, S., and C. A. Bryant. 2005. Social marketing in public health. Annual Review of Public Health 26: 319–39. Guerrón-Montero, C. 2002. Practicing anthropology in Latin America. Practicing Anthropology 24: 2–4. Gust, G., and K. M. MacQueen. 2008. Handbook for team-based qualitative research. Lanham, MD: AltaMira. Hahn, R. A., and M. C. Inhorn. 2009. Anthropology and public health: Bridging differences in culture and society, 2nd ed. New York: Oxford University Press. Hale, C. R. 2007. In praise of “reckless minds”: Making a case for activist anthropology. In Anthropology put to work, ed. L. Field and R. G. Fox, 103–28. New York: Berg. Handwerker, W. P. 2001. Quick ethnography. Walnut Creek, CA: AltaMira. Harding, S., ed. 1987. Feminism and methodology: Social sciences issues. Bloomington: Indiana University Press. Harrison, B. 2001. Collaborative programs in indigenous communities: From fieldwork to practice. Walnut Creek, CA: AltaMira. Heath, S. B., and M. W. McLaughlin, eds. 1993. Identity and inner-city youth: Beyond ethnicity and gender. New York: Teachers College Press. Heckathorn, D. D. 1997. Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems 44: 174–99. Heckathorn, D. D. 2002. Respondent-driven sampling II: Deriving valid population estimates from chain-referral samples of hidden populations. Social Problems 49: 11–34. Hill, C. E., ed. 1991. Training manual in applied medical anthropology. Special Publications No. 27. Washington, DC: American Anthropological Association. Hill, C. E., and M. L. Baba. 2006. The globalization of anthropology. NAPA Bulletin No. 25. New York: Wiley-Blackwell. Holmberg, A. 1954. Participant intervention in the field. Human Organization 14: 23–26. Holmberg, A. 1958. The research and development approach to the study of culture change. Human Organization 17: 12–16. Holmberg, A. 1966. Vicos: Metodo y práctica de antropología aplicada. Investigaciones Sociales, Serie: Monografías Andinas, No. 5. Lima, Editorial Estudios Andinos, S.A. Institute for Community Research (ICR). 1993. Rapid sociodemographic assessment Project: 13 neighbourhood profiles. Hartford: ICR. Jacobs-Huey, L. 2002. The natives are gazing and talking back: Reviewing the problematics of positionality, voice, and accountability among “Native” anthropologists. American Anthropologist 104: 791–804.

14_107-Bernard.indb 687

6/9/14 10:03 AM

688

Robert T. Trotter, II, et al.

Johnsen, E. C. 1985. Network macrostructure models for the Davis-Leinhardt set of empirical sociomatrices. Social Networks 7: 203–24. Johnsen, E. C, H. R. Bernard, P. D. Killworth et al. 1995. A social network approach to corroborating the number of AIDS/HIV+ victims in the U.S. Social Networks 17: 167–87. Johnson, J. C. 1990. Selecting ethnographic informants. Qualitative research methods, Vol. 22. Beverly Hills, CA: Sage. Johnson, J. C. 1994. Anthropological contributions to the study of social networks: A review. In Advances in social network analysis, ed. S. Wasserman and J. Galaskiewicz, 132–48. Beverly Hills, CA: Sage. Johnson, J. C., E. Schade, and S. C. Weller. 1995. Truth or dare?: Sexual networks, friendship networks, and risk behavior in an informal gay group. Connections 18: 73–88. Jordan, A. T. 2003. Business anthropology. Prospect Heights, IL: Waveland. Jordan, A. T. 2010. The importance of business anthropology: Its unique contributions. International Journal of Business Anthropology 1: 15–25. Jordan, G., and C. Weedon. 1995. Cultural politics: Class, gender, race and the postmodern world (esp. Part III). Cambridge: Basil Blackwell. Khan, M. E, M. Anker, B. C. Patel et al. 1991. The use of focus groups in social and behavioral research: Some methodological issues. World Health Statistical Quarterly 44: 145–49. Kirk, J., and M. L. Miller. 1986. Reliability and validity in qualitative research. Beverly Hills, CA: Sage. Kleinman, A. 1980. Patients and healers in the context of culture. Berkeley: University of California Press. Kleinman, A., and P. Benson. 2006. Anthropology in the clinic: The problem of cultural competency and how to fix it. PLoS Medicine 3: e294. Klovdahl, A. S. 1989. Urban social networks: Some methodological problems and possibilities. In The small world, ed. M. Kochen, 176–210. Norwood, NJ: Ablex. Klovdahl, A. S., J. J. Potterat, D. E. Woodhouse et al. 1994. Social networks and infectious disease: The Colorado Springs Study. Social Science and Medicine 38: 79–88. Knoke, D., and S. Yang. 2008. Social network analysis, 2nd ed. Thousand Oaks, CA. Sage. Koester, S. 1994. Coping, running, and paraphernalia laws: Context and high risk behavior. Human Organization 53: 278–95. Kostick, K. M., S. L. Schensul, R. Singh et al. 2011. A methodology for building culture and gender norms into intervention: An example from Mumbai, India. Social Science & Medicine 72: 1630–38. Kozulin, A., B. Gindis, V. S. Ageyev, and S. M. Miller, eds. 2003. Vygotsky’s educational theory in cultural context. Cambridge: Cambridge University Press. Krueger, R. A., and M. A. Casey. 2000. Focus groups: A practical guide for applied research. Thousand Oaks, CA: Sage. Ladson-Billings, G. 2009. Race still matters: Critical race theory in education. In Routledge international handbook of education, ed. M. Apple, W. Au, and L. A. Gandin, 110–22. New York: Routledge. Lamphere, L. 2004. The convergence of applied, practicing and public anthropology in the 21st century. Human Organization 63: 431–43. Latkin, C., W. Mandell, M. Oziemkowska et al. 1995. Using social network analysis to study patterns of drug use among urban drug users at high risk for HIV/AIDS. Drug and Alcohol Dependence 38: 1–9. LeCompte, M., and J. Schensul. 2010. Designing and conducting ethnographic research: An introduction. Lanham, MD: AltaMira.

14_107-Bernard.indb 688

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

689

Lester, R. J. 2011. How do I code for black fingernail polish? Finding the missing adolescent in managed mental health care. Ethos 39: 481–96. doi: 10.1111/j.1548-1352.2011.01210.x. Luborsky, M. R., and R. L. Rubinstein. 1995. Sampling in qualitative research: Rationale, issues, and methods. Research on Aging 17: 89–114. MacQueen, K. M., E. McLellan, K. Kay, and B. Milstein. 1999. Codebook development for teambased qualitative analysis. Cultural Anthropology Methods 10(2): 31–36. MacQueen, K. M., E. McLellan, D. S. Metzger et al. 2001. What is community? An evidence-based definition for participatory public health. American Journal of Public Health 91: 1929–38. Marcus, G. E., and M. M. J. Fischer. 1986. Anthropology as cultural critique. Chicago: University of Chicago Press. Marsden, P. V. 1990. Network data and measurement. Annual Review of Sociology 16: 435–63. McGrady, F. A., C. Marrow, G. Myers et al. 1995. A note on implementation of a random-walk design to study adolescent social networks. Social Networks 17: 251–55. McLellan, E., K. M. MacQueen, and J. L. Neidig. 2003. Beyond the qualitative interview: Data preparation and transcription. Field Methods 15: 63–84. Miles, M. B., and A. M. Huberman. 2002. The qualitative researcher’s companion. Thousand Oaks, CA: Sage. Morgen, S. 1993. Gender and anthropology: Critical reviews for research and teaching. Washington, DC: American Anthropological Association. Morse, J. M., M. Barrett, M. Mayan et al. 2002. Verification strategies for establishing reliability and validity in qualitative research. International Journal of Qualitative Methods 1: 1–19. Mosack, K. E., M. Abbott, M. Singer et al. 2005. If I didn’t have HIV, I’d be dead now: Illness narratives of drug users living with HIV/AIDS. Qualitative Health Research 15: 586–605. Nachman, S. R. 1984. Lies my informants told me. Journal of Anthropological Research 40: 536–55. Nastasi, B. K., J. J. Schensul, M. A. De Silva et al. 1998. Community-based sexual risk prevention program for Sri Lankan youth: Influencing sexual-risk decision making. International Quarterly of Community Health Education 18: 139. Nastasi, B. K., J. J. Schensul, C. Varjas, and C. S. Yoganathan. 1997. Community-based HIV/ AIDS prevention program for Sri Lankan youth. Unpublished manuscript. Nastasi, B. K., and S. L. Schensul. 2005. Contributions of qualitative research to the validity of intervention research. Journal of School Psychology 43: 177–95. Needle, R. H., S. L. Coyle, S. G. Genser, and R. T. Trotter, II. 1995. Introduction: The social network paradigm. In Social networks, drug abuse, and HIV transmission, ed. R. H. Needle, S. L. Coyle, S. G. Genser, and R. T. Trotter, 1–2. NIDA Research Monograph 151. Rockville, MD: USDHHS. Needle, R. H., E. Goosby, C. Bates et al. 2000. Crisis response team initiative: A guide for conducting community based rapid assessment, response, and evaluation. Washington, DC: Department of Health and Human Services. Onwuegbuzie, A. J., W. B. Dickinson, N. L. Leech, and A. G. Zoran. 2009. Toward more rigor in focus group research: A new framework for collecting and analyzing focus group data. International Journal of Qualitative Methods 8: 1–21. Ovretveit, J. 2002. Action evaluation of health programs and changes. Abigdon, UK: Radcliffe Medical Press. Parker, R. G., and J. H. Gagnon. 1995. Conceiving sexuality: Approaches to sex research in a postmodern world. New York: Routledge. Pelto, P. J., and G. H. Pelto. 1978. Anthropological research: The structure of inquiry, 2nd ed. Cambridge: Cambridge University Press.

14_107-Bernard.indb 689

6/9/14 10:03 AM

690

Robert T. Trotter, II, et al.

Pelto, P. J., and J. J. Schensul. 1987. Toward a framework for policy research in anthropology. In Applied anthropology in America, ed. E. Eddy and W. L. Partridge, 505–28. New York: Columbia University Press. Podolefsky, A., P. Brown, and S. Lacy. 2008. Applying anthropology: An introductory reader, 9th ed. New York: McGraw Hill. Polkinghorne, D. E. 2005. Language and meaning: Data collection in qualitative research. Journal of Counseling Psychology 52: 137–45. Quinn, N., and D. Holland. 1987. Cultural models in language and thought. New York: Cambridge University Press. Rappaport, J. 2007. Anthropological collaborations in Columbia. In Anthropology put to work, ed. L. Field and R. G. Fox, 21–43. Oxford: Berg. Robertson, J. 2002. Reflexivity redux: A pithy polemic on “positionality.” Anthropological Quarterly 75: 785–92. Rogoff, B. 1990. Apprenticeship in thinking: Cognitive development in social context. New York: Oxford University Press. Romney, A. K., S. C. Weller, and W. H. Batchelder. 1986. Culture as consensus: A theory of cultural and informant accuracy. American Anthropologist 88: 313–38. Ryan, G. W., and H. R. Bernard. 2006. Testing an ethnographic decision tree model on a national sample: Recycling beverage cans. Human Organization 65: 103–14. Sandelowski, M. 1986. The problem of rigor in qualitative research. Advances in Nursing Science 8: 27–37. Schensul, J. J. 1985. Systems consistency in field research, dissemination and social change. In Collaborative research and social policy: Special issue of American Behavioral Scientist, ed. J. J. Schensul and G. Stern, 29: 186–204. Schensul, J. J. 1998a. Community-based intervention with urban youth. Miniseries edited by B. K. Nastasi. Journal of School Psychology 2. Schensul, J. J. 1998b. Learning about sexual meaning and decision-making from urban youth. In Cross-cultural perspectives of women’s sexual decision making: Implications for sexual health protection at the community level. Special issue of the International Quarterly of Community Health Education, ed. M. I. Torres and M. Weeks, 29–48. Schensul, J. J. 1999. Organizing community research partnerships in the struggle against AIDS. Health Education & Behavior 26: 266–83. Schensul, J. J. 2010. Engaged universities, community-based research organizations and third sector science in a global system. Human Organization 69: 307–20. Schensul, J. J., and M. J. Berg, 2004. Core elements of participatory action research for educational empowerment and risk prevention with urban youth. Practicing Anthropology 26: 5–9. Schensul, J. J., M. J. Berg, and K. M. Williamson. 2008. Challenging hegemonies: Advancing collaboration in community-based participatory action research. Collaborative Anthropologies 1: 102–37. Schensul, J. J., and M. LeCompte. 2010/2012. The ethnographer’s toolkit, 2nd ed. Lanham, MD: Altamira. Schensul, J. J., K. Radda, E. Coman, and E. Vazquez. 2009. Multi-level intervention to prevent influenza infections in older low income and minority adults. American Journal of Community Psychology 43: 313–29. Schensul, J. J., and S. L. Schensul. 1992. Collaborative research: Methods of inquiry for social change. The handbook of qualitative research in education. New York: Academic Press. Schensul, J. J, M. Torres, and T. Wetle. 1993. The Latino Alzheimers education project: Training curriculum. Hartford: Institute for Community Research.

14_107-Bernard.indb 690

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

691

Schensul, J. J., K. S. Truscott, S. Sydlo, and D. Robinson. 1997. The National Teen Action Research Center: Concept paper. Hartford: Institute for Community Research. Schensul, J. J., M. Weeks, and M. Singer. 1999. Building research partnerships. Researcher Roles and Research Partnerships 6: 85–164. Schensul, S. 1985. Science, theory, and application in anthropology. The American Behavioral Scientist 29: 164–85. Scott, G. 2005. Ethnographic assessment of injection drug use in Chicago: A foundation for behavioral surveillance; January. Atlanta: Centers for Disease Control and Prevention. Scott, J. 1991. Social network analysis: A handbook. Thousand Oaks, CA: Sage. Scrimshaw, N. S., and G. R. Gleason, eds. 1992. RAP: Rapid assessment procedures: Qualitative methodologies for planning and evaluation of health-related programs. Boston: International Nutritional Foundation for Developing Countries (INDFC). Scrimshaw, S. C. M. 1992. The adaptation of anthropological methodologies to rapid assessment of nutrition and primary health care. In RAP: Rapid assessment procedures: Qualitative methodologies for planning and evaluation of health-related programs, ed. N. S. Scrimshaw and G. R. Gleason, 25–38. Boston: International Nutritional Foundation for Developing Countries (INDFC). Shore, C., S. Wright, and D. Pero, eds. 2011. Policy worlds: Anthropology and the analysis of contemporary power. New York: Berghahn Books. Singer, M. 1989. The coming of age of critical medical anthropology. Social Science & Medicine 28: 1193–203. Singer, M., and H. Baer. 1995. Critical medical anthropology. Amityville, NY: Baywood Publishing Co. Singer, M., and H. Baer. 2007. Introducing medical anthropology: A discipline in action. Lanham, MD: AltiMira. Singer, M. C., P. I. Erickson, L. Badiane et al. 2006. Syndemics, sex and the city: understanding sexually transmitted diseases in social and cultural context. Social Science & Medicine 63: 2010–21. Singer, M., R. Irizarry, and J. J. Schensul. 1991. Needle access as an AIDS prevention strategy for IV drug users: A research perspective. Human Organization 50: 142–53. Snijders, T. A. B. 1992. Estimating on the basis of snowball samples: How to weight? Bulletin de Methodologie Sociologique 36: 59–70. Spiro, M. 1987. Culture and human nature. In Culture and human nature: Theoretical papers of Melford E. Spiro, ed. B. Kilborne and L. L. Langness, 145–60. Chicago: University of Chicago Press. Spradley, J. P. 1979. The ethnographic interview. New York: Holt, Rinehart and Winston. Stead, M., R. Gordon, K. Angus, and L. McDermott. 2007. A systematic review of social marketing effectiveness. Health Education 107: 126–91. Strauss, A. L. 1985. Qualitative analysis for social scientists. Cambridge: Cambridge University Press. Strauss, A. L., and J. Corbin. 2008. Basics of qualitative research: Techniques and procedures for developing grounded theory, 3rd ed. Newbury Park, CA: Sage. Stringer, E. T. 1996. Action research: A handbook for practitioners. Newbury Park, CA: Sage. Stull, D., and J. J. Schensul. 1987. Collaborative research and social change. Boulder: Westview. Tax, S. 1960. Action anthropology. In Documentary history of the Fox Project. Department of Anthropology, Chicago: University of Chicago Press. Trotter, R. T., II. 1991. Ethnographic research methods for applied medical anthropology. In Training manual in applied medical anthropology, ed. C. E. Hill, 180–212. American Anthropological Association Special Publications No. 27. Washington, DC: American Anthropological Association.

14_107-Bernard.indb 691

6/9/14 10:03 AM

692

Robert T. Trotter, II, et al.

Trotter, R. T., II. 1995. Drug use, AIDS, and ethnography: Advanced ethnographic research methods exploring the HIV epidemic. In Qualitative methods in drug abuse and HIV research, pp. 38–64. NIDA Monograph Series No. 157. Washington, DC: National Institute on Drug Abuse, NIH. Trotter, R. T., II. 1996. Communication and community participation in program evaluation processes. In Advanced methodological issues in culturally competent evaluation for substance abuse prevention, ed. A.-H. Bayer, F. L. Brisbane, and A. Ramirez, 241–66. CSAP Cultural Competence Series (No. 6). DHHS Publication No. (SMA) 96-3110. Trotter, R. T., II. 1997. Anthropological midrange theories in mental health research: Selected theory, methods and systematic approaches to at-risk populations. Ethos 25: 259–74. Trotter, R. T., II, 2012. Qualitative research sample design and sample size: Resolving and unresolved issues and inferential imperatives. Preventative Medicine. http://dx.doi.org/10.1016 /j.ypmed.2012.07.003. Trotter, R. T., II, J. A. Baldwin, and A. M. Bowen. 1995a. Network structure and proxy network measures of HIV, drug and incarceration risks for active drug users. Connections 18: 89–104. Trotter, R. T., II, A. M. Bowen, J. A. Baldwin, and L. J. Price. 1996. The efficacy of network-based HIV/AIDS risk reduction programs in midsized towns in the United States. Journal of Drug Issues 26: 591–606. Trotter, R. T., II, A. M. Bowen, and J. M. Potter. 1995b. Network models for HIV outreach and prevention programs of drug users. In Social networks, drug abuse, and HIV transmission, ed. R. H. Needle, S. L. Coyle, S. G. Genser, and R. T. Trotter, II, 144–80. NIDA Research Monograph 151. USDHHS. Rockville, MD: National Institute on Drug Abuse. Trotter, R. T., II, E. K. Briody, G. H. Sengir, and T. L. Meerwarth. 2008. The life cycle of collaborative partnerships: Evolutionary structure in industry-university research networks. Connections 28: 40–58. Trotter, R. T., II, and M. E. Medina Mora. 2000. Qualitative methods. In Guide to drug abuse epidemiology, pp. 93–123. Geneva: World Health Organization. Trotter, R. T., II, R. H. Needle, E. Goosby et al. 2001. A methodological model for rapid assessment, response and evaluation: The RARE Program in Public Health. Field Methods 13: 137–59. Trotter, R. T., II, and J. M. Potter. 1993. Pile sorts, a cognitive anthropological model of drug and AIDS risks for Navajo teenagers: Assessment of a new evaluation tool. Drugs and Society 7: 23–39. Trotter, R. T., II, R. B. Rothenberg, and S. Coyle. 1995c. Drug abuse and HIV prevention research: Expanding paradigms and network contributions to risk reduction. Connections 18: 29–46. Trotter, R. T., II, and M. Singer. 2005. Rapid assessment strategies for public health: Promise and problems. In Community intervention and AIDS, ed. E. Trickett and W. Pequeqnat, 130–52. Oxford: Oxford University Press. Valente, T. W. 1995 Network models of the diffusion of innovations. Computational & Mathematical Organization Theory 2: 163–64. Van Willigen, J. 2002. Applied anthropology: An introduction, 3rd ed. Westport, CT: Bergin and Garvey. Vygotsky, L. S. 1978. Mind in society: The development of higher psychological processes. Cambridge, MA: Harvard University Press. Vygotsky, L. S. 1986 [1934]. Thought and language, rev. ed. Cambridge: MIT Press. Wallerstein, N. 2012. Paulo Freire in the north: Interdisciplinary approaches to empowerment education. Trabalhos em Lingüística Aplicada 14(1).

14_107-Bernard.indb 692

6/9/14 10:03 AM

21: Theories and Methods in Applied Anthropology

693

Wasserman, S., and K. Faust. 1994. Social network analysis: Methods and applications. Cambridge: Cambridge University Press. Wasson, C., M. O. Butler, and J. Copeland-Carson. 2012. Applying anthropology in the global village. Walnut Creek, CA: Left Coast Press, Inc. Watters, J. K., and P. Biernacki. 1989. Targeted sampling: Options for the study of hidden populations. Social Problems 36: 17–18. Watts, D. J. 1999. Networks, dynamics, and the small-world phenomenon. American Journal of Sociology 105: 493–527. Weeks, M. R., M. Convey, J. Dickson-Gomez et al. 2009a. Changing drug users’ risk environments: Peer health advocates as multi-level community change agents. American Journal of Community Psychology 43: 330–44. Weeks, M. R., J. Li, J. Dickson-Gomez et al. 2009b. Outcomes of a peer HIV prevention program with injection drug and crack users: The risk avoidance partnership. Substance Use & Misuse 44: 253–81. Weller, S. C. 1984. Consistency and consensus among informants: Disease concepts in a rural Mexican town. American Anthropologist 86: 966–75. Weller, S. C., L. M. Pachter, R. T. Trotter, II et al. 1993. Empacho in four Latino groups: A study of intra‐and inter‐cultural variation in beliefs. Medical Anthropology 15: 109–36. Weller, S. C., and A. K. Romney. 1988. Systematic data collection. Sage University Paper Series on Qualitative Research Methods, Vol. 10. Beverly Hills, CA: Sage. Werner, O., and G. M. Schoepfle. 1987. Systematic fieldwork: Ethnographic analysis and data management, Vol. 2. Beverly Hills, CA: Sage. Wertsch, J. V. 1991. Voices of the mind: A sociocultural approach to mediated action. Cambridge, MA: Harvard University Press. Whiteford, L., and R. T. Trotter, II. 2008. Ethics in anthropological research and practice. Boston: Wadsworth.

14_107-Bernard.indb 693

6/9/14 10:03 AM

14_107-Bernard.indb 694

6/9/14 10:03 AM

CHAPTER 22

Presenting Anthropology to Diverse Audiences Conrad Phillip Kottak

In this chapter I assume I’m mainly addressing cultural anthropologists and that many or most of my readers will be academics and graduate students. I draw on a career that goes back to 1962, when I was an undergraduate anthropology major at Columbia College of Columbia University and had my first field experience in Arembepe, Bahia, Brazil. I got my Ph.D. from Columbia in 1966, did postdoctoral fieldwork in Madagascar, and in January 1968—direct from Madagascar—joined the faculty of the University of Michigan, where I taught through my retirement in 2011. My career in anthropology has allowed me to address, through writing and speaking, many different audiences. Since 1981, in addition to academic anthropology, I have done applied anthropology, working as a consultant to organizations including USAID and, especially, the World Bank. All this experience is the basis for what follows. I offer several observations and suggestions, not all of which I’ve followed consistently. My aim is to suggest ways of writing and speaking that are appropriate to contexts into which cultural anthropologists enter regularly. I’ve never worked in a museum, made a film, or collected music in the field. I have ideas about dos and don’ts in ethnographic film-making and museum displays, but I won’t say much about those here, leaving such comments for others more experienced in those areas than I am. WRITING ANTHROPOLOGY

Cultural anthropology is primarily a “book field”: Most cultural anthropologists are expected to write books as well as articles. Anthropologists mainly write three kinds of books: academic, trade, and textbooks. If a sociocultural anthropologist wants to get tenure at a major university, he or she needs to have an academic book published (along with peer-reviewed articles). Other subfields of anthropology are or have been “article fields.” That means one can get tenure with a series of articles in respected, refereed journals in which manuscripts are submitted independently (rather than being solicited) and sent out for peer review. Linguistic anthropology used to be an article field but has become more of a book field. Biological anthropology remains mainly an article field.

Professional (Academic) Books Ideally, your first book should be an academic book, demonstrating your special knowledge and skills at research, analysis, and interpretation. Often, the first book is a 695

14_107-Bernard.indb 695

6/9/14 10:03 AM

696

Conrad Phillip Kottak

revision of a doctoral dissertation. When graduate students defend their dissertations, their committee members usually suggest ways of turning the dissertation into a book and/or a series of articles (for more on turning a dissertation into a book, see Germano [2005] and Luey [2004]). After the recent Ph.D. has done some work incorporating the committee’s suggestions, he or she usually submits the revised manuscript to a university press or a monographs series. Such a series may have one or more anthropologists who function as acquiring editors. They may read the manuscript and suggest further revision before sending it out for review. Acquiring editors who are not anthropologists often request a prospectus and sample chapter(s) rather than the entire manuscript. They send these to one or two trusted reviewers for a quick response. If favorable, the manuscript is sent to two or three external reviewers who read the work, recommend for or against publication, and suggest further revisions. Usually, the manuscript’s fate rests with those readers. Anthropologists are notable for their divergent views about merit. (This is true among manuscript and book reviewers as well as for grant proposal reviews—see below.) If there are generally positive reviews, the author and editor work together on a revision plan (which may be minor), and the book eventually goes into production. This process usually takes at least two years following the dissertation defense. Normally, a second book, based on a different research project or body of work, is needed (along with articles) for promotion to the rank of professor. These “promotion books and articles” need to be aimed at an academic audience and to have gone through peer review. Only after these hurdles are passed can the anthropologist consider the other forms of writing discussed below. In addition to revised dissertations, most academic (“scholarly”) books (aka books that few will read) typically are submitted to a university press. (A few contemporary presses, including AltaMira and Left Coast Press, publish anthropology books intermediate between textbooks and university press titles.) Academic books are supposed to make some kind of contribution to “theory,” a term that means very different things to different cultural anthropologists. By definition, theory may refer to: (1) “a coherent group of general propositions used as principles of explanation for a class of phenomena”; or (2) “the branch of a science or art that deals with its principles or methods, as distinguished from its practice” (Stein 1982, 1362). The second sense is the basis of the distinction between “theoretical” (academic) and “applied” (practicing) anthropology. With respect to the first meaning, “ethnological theory” once encompassed an established set of “isms”: evolutionism, diffusionism, historical particularism, functionalism, structuralism, etc. For a significant number of contemporary sociocultural anthropologists, however, theory often resembles the name-dropping characteristic of literary criticism. Current norms in cultural anthropology facilitate the ritual invocation, across several academic fields, of “the usual suspects”—a too-limited set of voguish names, works, and words. In cultural anthropology, such fashionable words include “constructing,” “representing,” “contesting,” “resisting,” “privileging,” “grounded,” “theorized,” “interlocutor,” “discourse,” “dialogic,” “hegemonic,” and “subaltern.” Because you are writing for insiders, you must, to some extent, show that you share the norms and knowledge of the culture of academic anthropology.

14_107-Bernard.indb 696

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

697

Much academic writing is bad writing (Sword 2012). Deplorably, some professors reward it and perpetuate it among their students. I have worked with graduate students who entered our program as excellent writers, but whose prose was subsequently contaminated by the purveyors of jargon, name dropping, and complications that may obscure either simple or complex thoughts. I warn graduate students about this hazard and urge them to resist. Students should be urged to “eschew obfuscation.” I like to think that, in the long run, good writing is rewarded with readers and respect. Although an academic book may appeal mainly to specialists (such as geographic area specialists), it will reach a wider academic audience if you address theoretical, methodological, historical, or comparative issues beyond the specific case and area. In general, an anthropological work should place a case study in a broader context through comparison with material from other areas and/or by showing its relation to theory. A killing review in a major journal might say “This book is of interest only to area specialists” (although this is exactly the audience you want to reach if you are publishing in a journal like Oceania). Academic books aimed at anthropologists should have something to say to as many anthropologists as possible.

Trade Books A trade book is likely to be read by an educated general public rather than only by academics or students. Such a book has a chance to appear on the New York Times weekly best seller list. Jared Diamond (an ornithologist who often writes about anthropological issues) frequently appears on this list. (See Richardson [1990] for an account of a trade book author’s experience.) Many anthropologists have written successful trade books, including prominent cultural anthropologists such as Margaret Mead, Ruth Benedict, Marvin Harris, and Oscar Lewis. A literary agent can be useful in negotiating trade book contracts—but not so much for other kinds of books, such as textbooks. Since Mead’s death, no cultural anthropologist has dominated the trade market as she did. One reason for Mead’s fame was her use of multimedia—books, magazines, films, television, radio, and public lectures. Another reason was her unique ability to bring lessons from other cultures to bear on discussions of social issues and cultural change in the United States. The range of topics she dealt with also attracted attention: sex, gender, adolescence, sex education, child rearing, and the generation gap. Since Mead’s death, biological anthropologists have been more successful in trade publishing than cultural anthropologists have. Examples include Richard Leakey, Donald Johanson, and especially Jane Goodall—another master of multimedia. Can a book written for a college course also attract a trade audience? Usually, no: You have to identify and write for a specific audience (e.g., college students, educated nonspecialists, lay readers). A trade book (e.g., Marvin Harris’s Cows, Pigs, Wars, and Witches [1974] or Cannibals and Kings [1977]) may find a college audience, but a book written for college students rarely crosses over to the trade market. Initially, I tried to sell my book Assault on Paradise (Kottak 2006) as a trade book, but this study of social change in a Brazilian fishing community was too case specific and too anthropological for a trade audience. It had too much data about issues that weren’t of overwhelming interest to the general public, even the educated public. After

14_107-Bernard.indb 697

6/9/14 10:03 AM

698

Conrad Phillip Kottak

being rejected by a trade publisher, Assault on Paradise was published by the Random House College Division. It all worked out: Assault on Paradise has sold better than many trade books. As a readable case study of social change and globalization at the local level, it has been assigned in an array of college courses because of its case material illustrating such anthropological topics as race, class, economics, kinship and marriage, religion, culture and personality, and social change. Like other popular case studies (also known as “supplemental texts”) used in introductory courses, such as Nisa (Shostak 1981) and Yanomamo (Chagnon 2013), Assault departs somewhat from traditional ethnographic writing, using such literary conventions as a cast of characters that includes the ethnographer(s), first-person narration, and occasional dialogue. Publishers of major textbooks in anthropology also make available a series of short ethnographic case studies to use as supplemental texts. Cengage (formerly Wadsworth) maintains a series of titles in its Case Studies in Anthropology series, including recent editions of popular books by Napoleon Chagnon on the Yanomamo (6th ed., 2013) and Richard B. Lee on the Dobe Ju/hoansi (4th ed., 2013). McGraw-Hill offers Assault on Paradise (4th ed., 2006) and Bruce Knauft’s The Gebusi (3rd ed., 2012) as well as Holly Peters-Golden’s Culture Sketches (6th ed., 2012). McGraw-Hill also has an arrangement with Waveland Press that makes several popular case studies available for use with McGraw-Hill texts.

Writing Ethnographies Early ethnographers often viewed their fieldwork and writing as salvage ethnography—studying, documenting, and recording cultural diversity threatened by Westernization. Such early ethnographic accounts as Bronislaw Malinowski’s classic Argonauts of the Western Pacific (1961 [1922]) were similar to earlier traveler and explorer accounts in describing the writer’s discovery of unknown people and places. The scientific aims of early ethnographies, however, set them apart from accounts by explorers, amateurs, and novelists such as Joseph Conrad, another Pole writing in English, to whose prose Malinowski’s is sometimes compared. The style that dominated “classic” ethnographies was ethnographic realism. The writer’s goal was to present an accurate, objective, scientific account of a different way of life, written by someone who knew it first-hand. This knowledge came from an “ethnographic adventure” involving immersion in an alien language and culture. Ethnographers derived their authority—both as scientists and as voices of “the native” or “the other”—from this personal research experience. Malinowski’s ethnographies reflected his belief that aspects of culture are linked and intertwined. Beginning by describing a Trobriand sailing expedition, Malinowski then followed the links between that entry point and other areas of the culture, such as magic, religion, myths, kinship, and trade. Today’s ethnographies tend to be less inclusive and holistic, focusing on particular topics, such as kinship or religion, and more concerned with theory than with capturing the primordial essence of an endangered other. Recent trends include focusing on the art and craft of ethnographic writing (Narayan 2012; van Maanen 2011), along with questioning its traditional goals, methods, and styles (Brettell 1996; Clifford 1982, 1988; Marcus and Cushman 1982; Marcus and Fischer 1999; Waterston and Vesperi 2009). Marcus and Fischer argue that experimen-

14_107-Bernard.indb 698

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

699

tation in ethnographic writing is necessary because all peoples and cultures must be “rediscovered . . . in changing historical circumstances” (1999, 24). Some contemporary cultural anthropologists see ethnographic writing as an art as much as a science. Such a view might regard an ethnographic account as allowing the ethnographer, as mediator, to communicate information from the “natives” to readers. Some experimental ethnographies are dialogic, presenting ethnography as a dialogue between the anthropologist and one or more informants (e.g., Behar 1993; Dwyer 1982). These works draw attention to ways in which ethnographers, and by extension their readers, communicate with other cultures. However, some such ethnographies have been criticized for spending too much time talking about the anthropologist and too little time describing the local people and their culture. The dialogic ethnography is one genre within a larger experimental category—reflexive ethnography. Here the ethnographer puts his or her personal feelings and reactions to the field situation right in the text. Experimental writing strategies are prominent in reflexive accounts. The ethnographer may adopt some of the conventions of the novel, including first-person narration, conversations, dialogues, and humor. Contemporary ethnographies usually recognize that cultures constantly change and that an ethnographic account applies to a particular moment. A current trend in ethnography is to focus on the ways in which cultural ideas serve political and economic interests (see Brettell 1996; Wolf 1992). Another approach is to describe— sometimes through a life history approach—how various particular “natives” participate in broader historical, political, and economic processes (Abu-Lughod 1993; Shostak 1981). For more on the relation between epistemology and writing, see the chapter by Schnegg in this volume. Anthropologists have had varying degrees of success in using these varied formats. Experimental writing is not, however, recommended for dissertation authors or for fledgling writers in general.

Textbooks Not everyone can write a textbook. Writing a general (four-field) anthropology text requires a background in all four subfields, and a decreasing number of departments are providing such training. Another option is for a general anthropology text to have coauthors with different subdisciplinary backgrounds. Such a book, however, may lack a uniform, coherent voice. The most effective textbooks, I believe, emerge from their author’s personal commitment to undergraduate teaching. The first edition of my textbook Anthropology: Appreciating Human Diversity, which covers four-field anthropology, was published in 1974. That book, now in its fifteenth edition, led to spin-off texts—Cultural Anthropology: Appreciating Cultural Diversity (also in its fifteenth edition), Mirror for Humanity: A Concise Introduction to Cultural Anthropology (ninth edition), Window on Humanity: A Concise Introduction to Anthropology (sixth edition), and, most recently, Culture (with Lisa Gezon, second edition). Other successful general anthropology texts also have at least a cultural anthropology text as a spin-off. Each edition entails a prepublication review process. Before revision begins, the current edition is reviewed by several actual and potential users, who make suggestions for

14_107-Bernard.indb 699

6/9/14 10:03 AM

700

Conrad Phillip Kottak

revision. They usually propose adding something. Sometimes new chapters are reviewed as a textbook is being written or revised. How does one use and respond to such reviews? I pay special attention to recurrent comments and criticisms, those made by more than one reviewer. I also accept suggestions I agree with, even if made by a single reviewer. I generally reject suggestions with which I disagree if they come from a single reviewer (unless my editor convinces me the reviewer is onto something). Typically, reviewers who are specialists want more detail than an introductory textbook should include. If you follow their suggestions, your book can become too advanced—over the head of the introductory reader, who, along with his or her professor, is your target audience. Anthropology has been changing fairly rapidly, and, as new material and topics are added, texts tend to get too long. If the book is not to become unwieldy, you must cut as well as add; this means constantly rethinking what is fundamental. Doing a successful text means writing for a known target audience—usually firstand second-year college students who are taking an introductory course. Fields with multiple texts include general anthropology, cultural anthropology, anthropological archaeology, and biological anthropology. Some smaller fields with successful, multiedition texts include gender, race and ethnicity, sociolinguistics, medical anthropology, human evolution, human genetics, and Old and New World prehistory. The textbook author should aim for the reading level of the target audience. Writing programs like Grammatik, now part of some word processors, enable you to check reading (grade) level; they also make suggestions for improving readability, for example, by shortening sentences and eliminating obscure words and passive constructions. In principle, students are the target audience, but more influential are the professors who decide which texts to assign. Instructors choose textbooks they like and feel comfortable using. Good textbooks (no matter the course) have most of the following features: up-to-date scholarship, accessible prose, interesting writing (texts can be low level but boring), visual appeal, and illustrations that themselves have pedagogical (not just decorative) value. Interesting boxes and various kinds of study aids (numbered chapter summaries, study questions, self-tests, suggested reading) are found in most texts. Today, all kinds of Internet resources are routinely available with a major text, and many texts come in electronic formats and can be rented rather than purchased. As additional supplements, the readable case study (e.g., Nisa, Yanomamo, Assault on Paradise) and the article anthology (e.g., Angeloni 2011; Podolefsky et al. 2012; Spradley and McCurdy 2012) also are standard parts of introductory curricula in general and cultural anthropology. Most contemporary anthropology teachers continue to present the field as a mirror for humanity: By looking at people in other cultures, we can better understand ourselves. The centrality of discussions of race, language, culture, and mind is part of the Boasian foundation of American anthropology. Marvin Harris once commented at an annual meeting of the American Anthropological Association that his (and my) introductory textbooks were rooted in the introductory anthropology course we both took (almost a generation apart) at Columbia University. This course can be traced back to Franz Boas—in Harris’s case through Charles Wagley, his teacher; in mine, through

14_107-Bernard.indb 700

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

701

Lambros Comitas. The framework of that course (and its transformations at Berkeley, Florida, Pennsylvania, and Yale) has remained remarkably constant and continues as the framework for most textbooks in introductory general, four-field anthropology. Although particular instructors want new and different features and use textbooks differently, the basic course has remained remarkably stable. The organization and content of the most popular textbooks reveal a basic notion of what anthropology is about. Experimental textbooks in general anthropology usually don’t do very well. To stray too far from the familiar is to be overinnovative and to risk poor sales. Constant updating is fine, particularly inclusion of new fossil and archaeological finds or discussions of “hot topics”—but only when those topics are clearly related to anthropology’s traditional subject matter. New—innovative—chapters should be added with care—instructors may be miffed when a chapter is added that they don’t see as part of the core. (There is more variation in the organization and content of introductory courses in cultural anthropology, reflecting different schools of thought and university traditions.) ARTICLES

Refereed Journals Junior faculty seeking tenure should publish in refereed (peer-reviewed) journals. The major ones in which cultural anthropologists publish include American Ethnologist, American Anthropologist, Ethnology, Journal of Anthropological Research, Current Anthropology, Cultural Anthropology, Human Organization, Human Ecology, and the Journal of the Royal Anthropological Institute, which has been titled both JRAI and Man. Other potential venues include area studies journals such as the Journal of Southeast Asian Studies or Latin American Research Review. All these journals share the canons of academic books. They offer variable speed of publication because an article manuscript will normally go out to three unpaid readers who have little incentive (other than intrinsic interest) to read it. Usually, the readers’ reports, if generally favorable, will require at least minor revisions. The editor’s specialty can make him or her more or less receptive to your article. If the editor is in the same field as you and likes your article, he or she may pay less attention to readers, prod them to speed up the review, and decide to publish even if the reviews are mixed. If reviewers are not prodded, it can take longer to get an article out than a book. Norms have changed across the field of anthropology, but more so in certain venues than in others. If you aim to publish in a particular journal, you should determine its current conventions and make sure your subject, treatment, language, and style are right for that journal. A playful title and trendy language are common, for example, in recent issues of the American Ethnologist or Cultural Anthropology. The first three articles in the August 2011 issue of American Ethnologist have these titles: “Indigenizing the City and the Future of Maori Culture: The Construction of Community in Auckland as Representation, Experience, and Self-making”; “Agonistic Intimacy and Moral Aspiration in Popular Hinduism: A Study in the Political Theology of the Neighbor”; and “Sketching Knowledge: Quandaries in Mimetic Reproduction of Pueblo Ritual.” All three illustrate the now almost ubiquitous format for an article or paper title in cultural anthropology in which a main title is followed by a postcolonic phrase. This format will not work in other subfields. An article entitled “Empowering Homo erectus:

14_107-Bernard.indb 701

6/9/14 10:03 AM

702

Conrad Phillip Kottak

The Dialogics of Oppression and Resistance in a Lower Paleolithic Population” probably won’t fly at the American Journal of Physical Anthropology. Indeed, the norms in cultural anthropology have changed in recent decades. The first three articles in the August 1984 issue of the American Ethnologist, for instance, had more straightforward, less playful-literary titles. They were “The Maintenance of Peasant Coffee Production in a Peruvian Valley,” “The Horns of the Mediterraneanist Dilemma,” and “The Political Use of Sande Ideology and Symbolism.” Neither the colon nor any current buzzword adorned any title in that issue, and the titles were shorter and clearer. Although the pattern of Title-colon-Subtitle has become the rule in cultural anthropology, there is a danger, particularly now that Googling has become the norm in searching, that “cute,” nonsubstantive titles may hide the work from those most interested in its actual content.

Collections Many collections of articles originate in conferences and can be uneven. Others are solicited, like the chapters in this book. For contributors, publishing in an edited volume like this one usually entails less independent effort (someone you know asks you to contribute) than when you submit a journal article; it wasn’t your original idea to begin with. Chapters in books carry less prestige for promotion, although they usually are refereed, as the chapters in this book have been. Popular Articles Cultural anthropologists haven’t been very successful recently in writing for a large audience of educated nonspecialists. There are few venues for popular anthropology. It’s rare to find a piece by an anthropologist in the New York Times Magazine, The New Yorker, Harper’s, or Atlantic Monthly. Anthropologists do publish occasionally in Natural History, Psychology Today, Discover, and The New York Review of Books. Since Margaret Mead, however, cultural anthropology has produced no “popularizer” comparable to invertebrate paleontologist and evolutionist Stephen Jay Gould. His columns for Natural History were collected to yield paperback trade books. (Given the paucity of anthropologist popularizers, the Anthropology in Media award of the American Anthropological Association often goes to nonanthropologists, including Gould, Ursula K. Le Guin, and John Noble Wilford.) One strategy for popular writing (which was Margaret Mead’s forte) is to use an anthropological approach or perspective derived from studying other cultures to cast light on aspects of contemporary U.S. culture such as child care, fast food, the media, or sports. Another strategy is to focus on a problem affecting a traditional anthropological population or locale (endangered peoples or primates, or rainforests)—but avoid whining. Suggest solutions, so that readers can get involved, rather than merely throw up their hands in despair. The journal Anthropology Now, which has been published since 2009 (three times a year by Paradigm Publishers), tries to publish well-written, accessible articles by anthropologists writing about current problems and issues likely to be of general interest within the field. Its audience, however, consists mainly of anthropologists. Accessibility and general interest across the subfields also are charac-

14_107-Bernard.indb 702

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

703

teristic of articles published in General Anthropology, the twice-yearly bulletin of the General Anthropology Division of the American Anthropological Association. GRANT PROPOSALS

Grant proposals are vehicles for addressing stingy and highly critical colleagues. I’ve served on the NSF Cultural Anthropology panel and have received grants from NSF, NIMH, and NASA (for satellite imagery and on-the-ground research in areas of increasing deforestation in Madagascar). Based on this experience, I conclude that any grant proposal should answer four key questions: 1. 2. 3. 4.

Why this topic/problem? Why this place? Why this person or team? How will the study be done?

Why This Topic/Problem? Funds are limited; reviewers and panelists determine which projects are worthy of funding. The grant writer must convince them that the topic is important—more so than other topics being proposed for funding at the same time. Why This Place? The topic may be important, but why study it in the place proposed? Some topics make sense in certain locales but not others; the grant writer must show the connection. For example, it was easy for me to justify research on the cultural impact of television in Brazil because, at the time of my study, that country had the world’s most watched commercial television network, with national penetration. It wouldn’t have made much sense for me to study television in Madagascar, where sets and transmissions reach a limited audience—mainly urban. Why This Person or Team? With the proliferation of anthropologists, it’s not uncommon to come across two people who want to study similar topics in the same country. Especially in that event, but really in any case, it’s important to identify the special qualifications of the person(s) proposing to do this research in this place. Relevant here are such background factors as: a history of residing in the country, language training or fluency, prior or pilot research in the proposed project locale, experience there or elsewhere with the topic or method, and the like. How Will the Study Be Done? Grant proposals in cultural anthropology are notoriously weak in their formulation and discussion of methods. Grant writers should take the methods section very seriously, especially sampling design. Books such as Anthropological Research by Pertti and Gretel Pelto (1978) and H. Russell Bernard’s Research Methods in Anthropology (5th ed., 2011) have fostered the growth of anthropology methods courses in colleges and universities. (Annual methods camps and various workshops supported by NSF have

14_107-Bernard.indb 703

6/9/14 10:03 AM

704

Conrad Phillip Kottak

also enhanced our methodological training.) However, I’ve heard and read comments by grant program officers and panelists that anthropologists still need to give more serious attention to sampling design. Too many applicants say they’re planning to use network sampling or snowball sampling rather than one of the random sampling procedures described in the Pelto and Bernard books. Random sampling designs support greater inference about populations and are used more widely in the other social sciences, but nonrandom samples are also appropriate in some cases (see the chapter by Guest in this volume). The reader of a grant proposal (or of an article or book) should know certain things after reading the abstract and a few pages: What’s the topic/problem? What’s the research plan? What’s going to be tested and how? Why is this research important? Where and when will it happen? Is the person proposing it qualified to do it? How will he or she do it? How will the data be collected, analyzed, and stored? What are its broader impacts and significance? To what dimensions of theory will it contribute? It’s amazing how much wheel spinning a reader may encounter in the first few pages of a grant proposal. Get to the point at once; don’t waste the reader’s time with too much background at the start—bring it in later when it’s relevant. (And don’t irritate the reader’s eyes with a small font or right justification—harder to read.) Systematically addressing key questions about significance, feasibility, and the research plan is especially important when submitting proposals to the more scientifically oriented funding sources, including NSF and NIH. Knowing your audience means paying attention to the names and missions of particular agencies, such as the National Science Foundation. Although not all agencies have scientific mandates, you always have to justify your topic, setting, and qualifications and say what you propose to do and how, even if you don’t plan to use scientific methods (Winslow 2007). As mentioned previously, anthropologists who review grant proposals often diverge in their views. When I served on the NSF cultural anthropology panel, I was surprised by the variability in reviewers’ scores and comments about the same proposal. For some proposals, scores ranged from excellent (5) to poor (1), although this range is not the norm. If at all possible (and often it is impossible), try to identify panelists likely to review your proposal. Some organizations make a list of members of review committees publicly available. Especially if you’re writing within their areas of expertise, be sure to cite them and to do it correctly (see also Krathwohl and Smith [2005] on preparing dissertation proposals). SOCIAL MEDIA AND CYBER COMMUNICATION

Imagine life without Twitter, Facebook, texting, and for some, Apple’s Siri, an inconsistently reliable “personal assistant.” These contemporary channels and means of communication allow people to stay in almost constant contact with others. Facebook, created in 2004, enables people to send public posts and private messages, share photos, organize events, and remember birthdays. Twitter, launched in 2006, was developed as a way of sending text messages to multiple people. People now use it to keep track of their friends and celebrity favorites (e.g., Ashton Kutcher, Sarah Palin, Barack Obama). People emit “tweets” to update people who “follow” them.

14_107-Bernard.indb 704

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

705

The complete world of cyberspace, including the blogosphere, constantly grows richer in the resources and communication opportunities available to anthropologists. Some of the most widely read anthropological blogs include: Savage Minds, a group blog http://savageminds.org Living Anthropologically, by Jason Antrosio http://livinganthropologically.org Neuroanthropology, by Greg Downey and Daniel Lende http://blogs.plos.org/neuroanthropology/ Also see this detailed list of anthropology blogs, as updated for 2013: http://anthropologyreport.com/anthropology-blogs-2013-list/

Anthropologists participate as well in various listservs and networking groups (e.g., LinkedIn). A bit of Googling on your part will take you to anthropologists’ personal websites as well as to research project websites. Specific communicative practices develop for different media audiences. One equalizing feature of cyberspeak (the range of styles and conventions people use when they write messages in cyberspace—or send text messages) is informality compared with print. Language is neither as fixed nor as precise as it is in writing for print. Cyberwriters and texters aren’t as concerned with typos as print writers are. Users of newer iPhones know that Siri sometimes corrects what you intend to say—and that her mistakes can be truly hilarious. Texting, email, and social media use their own abbreviations, such as BTW, IMHO, LOL, and WTF. (Use Google to find the meanings of these abbreviations, if they are unknown to you.) There also r single letters that stand 4 words and diacritical symbols (including emoticons) used to express various degrees of pleasure or displeasure. One lesson I learned as department chair was to use email with care. If you get a message that is at all problematic, don’t answer it right away, except perhaps to say: “I’ll get back to you on this.” Email usually can’t be undone, once a message is sent. Think over your response, ideally for a day. Email makes your writing seem blunter than you may want it to be. You may want to soften it with “pleases,” exclamation points, or (-: smiley faces. Again, know your audience—don’t write family members or good friends using the same impersonal style you use for university correspondence. Long email messages may be okay between family and friends, but can be irritating in professional email. (BTW: Also avoid leaving long voice mail and answering machine messages. And always repeat your phone number or say it slowly.) PUBLIC SPEAKING

For any kind of presentation, know your audience and address it. Before you give a “job talk,” for example, find out all you can about what is being sought and make sure your talk addresses that need. Memorize the names and interests of the likely members

14_107-Bernard.indb 705

6/9/14 10:03 AM

706

Conrad Phillip Kottak

of your audience. Talk to members of the search committee about topics and issues of interest to them. It’s usually not worth your time to apply for positions you don’t fit. If an ad is asking for an Africanist and you work in the Middle East, don’t apply. If the ad says Africanist, give your talk about Africa, not some other place you’ve worked. And if the ad says economic anthropology, don’t give a talk about religion. (It’s truly amazing how many people make these mistakes.) Anthropologists “give papers” at meetings. Far from giving their papers, however, many anthropologists make you work hard to receive them; most papers are written to be read by the eyes, not heard by the ears. Cultural anthropologists too often read aloud, rather than talking directly, to their audience. Biological anthropologists and archaeologists tend to use PowerPoint or other presentation software, and many use it effectively—with simple text (very few, short points per slide), photos, videos, and other visuals, for their presentations (Duarte 2008). Realizing that read papers are boring and hard to follow, I made a vow several years ago never again to read a paper at an AAA meeting. (Although my vow was culturally inappropriate, it hasn’t yet caused me grief. However, departing from tradition can be riskier for someone with less lecturing experience or at a more junior level.) I’ve kept my vow; I talk instead. I’m sure my audience hears, follows, understands, and remembers what I say better than they would if I read aloud from a prepared text. I also use this approach—talking with no notes or minimal notes—in my teaching. This lecture style requires a good memory, but memories can be trained. How do you lecture without a text or notes? Gradually reduce the memory aids you take to class; you’ll get the hang of it eventually. And there’s always PowerPoint, which has its own problems if not used properly (Duarte 2008). Anthropologists who care about reaching their audience use different techniques to make their presentations at meetings and conferences interesting. My colleague Ruth Behar, a gifted teacher and author who has mastered trade book writing and poetry, among other genres, tells me she does read her papers aloud but has learned how to read them dramatically. Anthropologists really do need to pay a lot more attention to their public speaking. It’s embarrassing to see a novice anthropologist (and even many senior ones) rush through a too-long text that has been written for the eyes rather than for the ears and that can’t possibly fit into the allotted time slot. Even when we’re not “giving” (i.e., inflicting) papers, most anthropologists probably spend more time presenting their field through talking than through writing. If you’re like me, you find yourself constantly explaining the nature, breadth, and value of anthropology in new settings and to diverse audiences—friends, relatives, neighbors, students, administrators, colleagues in other departments; the people “in the field” whom we study; and in settings and organizations where people have little correct information about the range of anthropological knowledge, methods, perspectives, and general expertise.

Teaching Anthropology Here is a key to successful teaching that I learned the hard way: Address the audience you have rather than the one you’d like. I failed to do this when I started teaching at the University of Michigan. I arrived in Ann Arbor in January 1968, having just

14_107-Bernard.indb 706

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

707

spent 15 months in Madagascar, following 6 months in Paris, and—before that— having been a graduate student (with no teaching experience) at Columbia University. The department chair assigned me to teach introductory anthropology, which I began inappropriately and inauspiciously—by discussing issues that had been important to me in graduate school. Many of my topics were way over the heads of the 300 or so primarily first-year students in my class. I wrote out my lectures and read them aloud, not very dramatically. I didn’t know how to be a lecturer or a teacher. In recompense, students would greet my entrance to the auditorium with hisses and boos. Often that semester, I regretted having turned down a more research-oriented position I’d been offered at the University of Illinois. I did recover. I had the good fortune to teach two much smaller classes (economic anthropology and culture and personality) during the ensuing spring–summer session. Each met twice a week for seven weeks; I didn’t have time to overprepare or to write out my lectures. I learned to talk to students and make sure they were getting the main points in the readings. In the fall, I transferred those lessons (talk to students, don’t overprepare or read aloud) to the introductory course, and my evaluations improved. As the world and our ways of understanding it change, new courses proliferate (e.g., “Ethnography as a Form of Writing,” “The Culture of Postmodernity”), replacing old ones with antiquated titles like “Peoples and Cultures of the Soviet Union” or “The Mind of Primitive Man.” Today’s students are more diverse (e.g., in age and ethnic background) than ever, yet they have grown up in a mass-mediated world and most are more comfortable with modern technology than their teachers are. For these students, and to explicate the culture(s) of the contemporary world, we teach new courses and use new teaching styles, methods, and tools. For example, one of my favorite recent courses is “Television, Society, and Culture.” The objectives of this class, intended mainly for seniors in anthropology, American culture, and communication studies, are: 1. to develop the ability to analyze television (content and impact) using perspectives from cultural anthropology; 2. to consider the social/cultural context of (a) creation, (b) distribution, reception, and (d) impact of TV messages; and 3. to apply the anthropological perspective within a single culture (e.g., the United States) and across cultures (e.g., Brazil and the United States).

Most of us use modern technology in our teaching (Pacansky-Brock 2013). Access to TV, the Internet, and video equipment (by students and in the classroom) is indispensable to “Television, Society, and Culture.” Students produce their own videos and clips for projects analyzing aspects of television in a cultural context.

College Administrators and Academic Colleagues During the ten years I served as chair of my department (1996–2006), I was constantly “selling” anthropology to others. Administrators know the reputations, strengths, and weaknesses of departments in their units, but are often unaware of changes in academic fields outside their own, especially a stereotyped field such as anthropology. “Been on any digs lately” lurks in the minds of most people who hear

14_107-Bernard.indb 707

6/9/14 10:03 AM

708

Conrad Phillip Kottak

the word anthropology. It’s important for administrators to be aware of anthropology’s increasing focus on contemporary issues and societies, including the United States. It’s useful for anthropologists to demonstrate the field’s relevance to such issues as immigration and diversity based on gender, race, ethnicity, and sexual orientation. Anthropology’s contemporary relevance isn’t well known outside the field, which continues to have an antiquarian reputation. Our field’s increasing focus on problems and issues reflecting social and cultural diversity adds value to our discipline within the modern college or university, which has an increasingly diverse student body. One key challenge is ensuring that anthropology’s long-time “ownership” of diversity isn’t forgotten as colleges and universities increasingly institute requirements for courses on diversity and multiculturalism. At the University of Michigan, I’ve noticed that many of the people who have been most active in advocating a diversity requirement know little about anthropology. We need to get our perspectives and cumulative knowledge across to them. Anthropologists at Michigan worked hard to ensure that introductory anthropology and other appropriate anthropology courses were included when the college instituted a new diversity (race and ethnicity) requirement in the 1990s. Many academics who advocate a diversity requirement only want other American voices represented, not voices from around the world—voices anthropologists have been heeding for years. Ideas about anthropology held by colleagues in other disciplines tend to be favorable but misinformed. Many still think that anthropology deals exclusively with non-Western societies, that it is necessarily interpretive rather than explanatory and qualitative rather than quantitative. It’s easy to show the links between anthropology and many other fields: for example, social work (applied anthropology), natural resources and the environment (ecological anthropology), business (anthropology’s global focus), and economic development (development anthropology). I’ve often heard a person in some unit, agency, or organization remark “People tell me we need an anthropologist.” The person making the remark doesn’t actually know what anthropologists do. Most people are unaware of anthropology’s breadth and the diversity of its subject matter. Most equate anthropology and archaeology. The more sophisticated think international. It’s not widely known that anthropologists work on social issues and topics related to public policy in the United States. Anthropologists also must confront (and combat) a series of naive impressions about culture, ethnicity, and multiculturalism (Kottak and Kozaitis 2006). Many Americans (even academics and intellectuals) rely on a Disneyesque “Small World” model of cultural diversity. I’m referring here to an attraction at Florida’s Walt Disney World and California’s Disneyland in which visitors take a boat ride through an exhibit consisting of hundreds of “ethnically correct” dolls from all over the world. The dolls move rhythmically while broadcasting an insidiously infectious ditty “It’s a small world after all.” The exhibit is a relentlessly commercial melding of globalization, multiculturalism, and postmodernism. Many nonanthropologists envision other cultures in terms of curious customs and colorful adornments—clothing, jewelry, hairstyles, dancing—like the dolls at Disney World. This bias toward the exotic and the superstructural even shows up in anthropo-

14_107-Bernard.indb 708

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

709

logical films. Many ethnographic films start off with music, often drum beats: “Bonga, bonga, bonga, bonga. Here in (supply place name), the people are very religious.” The (usually unintended) message is that people in nonindustrial societies spend most of their time wearing colorful clothes, singing, dancing, and practicing religious rituals. This feeds into the current commercial manipulation of culture as exotic superstructure—costumes, jewels, crystals, lost arks, temples of doom, mysterious, inexplicable, and ultimately unserious—to be appreciated, appropriated, and sold in ethnic fairs and quaint boutiques. In their ethnographic accounts—whether spoken, written, or visual—anthropologists can do more to spread the word that culture is something that ordinary people live each day.

The Media Be careful talking to the press. Reporters often get it wrong, and your spur-of-themoment comments, especially the more controversial ones, are likely to find their way into print as sound bites. On the other hand, if you’re boring or prosaic, you probably won’t be quoted. Surprisingly, one of my most carefully done and accurately reported interviews (by phone) was for the National Enquirer. Its grabber title was “Anthropologist Tells—Why Going to McDonald’s Is Like Going to Church.” Reporters, columnists, and TV people are after juicy tidbits; be especially careful what you say on television. Press interviews can become time consuming; you may eventually have to make choices about which venues are worth your time. And you may learn to avoid reporters and columnists who want to pick your brain for ideas they eventually present as their own—without mentioning you. People in the Field (Local People, Informants) Grant writing forces us to be selective and specific; we must identify a problem and develop hypotheses and a research design. The trend toward specialization diverts us from old-style holistic ethnography. We have to narrow our focus—even though it may have been breadth that drew us to anthropology to begin with. What happens, however, when we arrive in the field and discover that local people don’t have much interest in what we have chosen to study? (This problem can be precluded by a preliminary field trip to the area.) People in the field may work hard to direct us away from the issue we were funded to study and try to push us in another direction. Also, government officials can deprive us of the research clearance we need to pursue certain subjects. Brazilians reacted very differently to successive studies I did in their country. Three projects I directed during the 1980s and 1990s entailed multi-site team research investigating (1) the impact of television (1983–87); (2) the emergence of ecological awareness and environmental risk perception (1991–95); and (3) participatory development in the Northeast (1993–94). Issues that are of substantial interest (and considered highly fundable) in North America may generate much less enthusiasm elsewhere and vice versa. I had an easier time getting funding for the ecological awareness project than for the study of television. (Both were funded by NSF, among other agencies, as has been a new [2013] project on the evolution of media impact in Brazil). But Brazilians at all our field sites liked the television project much more than the ecological awareness project, which bored them. Finally, rural people in northeastern Brazil were willing

14_107-Bernard.indb 709

6/9/14 10:03 AM

710

Conrad Phillip Kottak

to cooperate in our World Bank–funded participatory development project—mainly because they thought they might get something out of it. The fact that the people we study appreciate some research topics more than others affects the direction and scope of our fieldwork. My first research project in Madagascar eventually reflected the interests of my Betsileo informants (in history and kinship) as much as my own initial focus (on economy, ecology, and political organization). The research preferences of local people may be especially evident in places that have been sites of team research and field schools, such as Arembepe, Bahia, Brazil (Kottak 2006), which has had several projects and researchers. Arembepe has been an anthropological field site since 1962, when it was one of several sites (in Brazil, Peru, Mexico, and Ecuador) chosen for undergraduate research within the ColumbiaCornell-Harvard-Illinois Summer Field Studies Program in Anthropology. Several researchers, including me, Isabel Wagley Kottak, David Epstein, Marvin Harris, Niles Eldredge, Janice Perlman, Shepard Forman, Carl Withers, Joseph Kotta, and Peter Gorlin, did summer fieldwork there or participated in the program as field director or assistant director. I was there most often in 1962, 1964, and 1965 and have done longitudinal research in Arembepe ever since (see Kottak 2006), most recently in 2012—50 years after my first visit. Several of my students also have worked in Arembepe. Doug Jones, doing a project that Arembepeiros liked a lot, weighed people and took and showed photographs as he investigated standards of physical attractiveness. Janet Dunn examined the role of various modernization variables, including television, in influencing family planning. The late Christopher O’Leary first studied Protestant conversion, then dietary change in Arembepe. The current NSF-funded project on the evolution of media impact in Brazil is proceeding in Arembepe as I write, along with fieldwork at four other sites in that country. The key lesson here is that you have to sell your anthropology in the field as well as back home. As with development projects (see below), local people are more likely to cooperate with research projects they like (and that are culturally appropriate) than with those that don’t interest them. Another factor that influences our field experience is our sponsorship—the agency or agencies that are funding us. Fulbright fellowships, for example, provide contacts with U.S. embassy and USIS officials, which can provide logistical support, especially in national capitals. The anthropologist working as an independent researcher will have a different status and role (and probably more explaining to do in the field) than the anthropologist working for an organization like USAID or the World Bank. When one works for “the Bank,” cooperation by government officials and interested local people is virtually guaranteed (at least the appearance of it is)—because people hope to gain from what the anthropologist is investigating. The benefits to be derived from participating in an independent research project often are unclear to our hosts—so the anthropologist has more selling to do here, both with government officials and local people. Independent researchers have to justify their missions to host country officials at various levels and to host country academics who can recommend necessary affiliations. One key obligation of the independent researcher is to remember such people—those who have facilitated our research. Host

14_107-Bernard.indb 710

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

711

country academics and students can become our future research collaborators and coauthors. We often have better access to funding sources, which may one day facilitate their research, incomes, or academic careers. Fieldwork raises another issue that returns us to the topic of writing for an audience. In the field, the researcher may want to employ an interview schedule, the cultural appropriateness of which must be determined through pilot research and pretesting. In 1984, we used Arembepe to pretest a basic interview schedule that was eventually used at six other Brazilian field sites (with some necessary local tinkering at each site). One result of pretesting was our discovery that some questions could be discarded because they produced little or no variation. (However, such questions can still be valuable for comparison across communities or cultures, or over time, because variation may develop.) Pretesting also showed which questions respondents did not understand and which ones were too complicated. For example, we rejected the advice of an American researcher to use seven-point ordinal scales because rural Brazilians were not accustomed to dealing with such fine distinctions. We used three- or four-category scales instead. The manipulation of a seven-point ordinal scale requires a degree of exposure to print media and testing that exists in contemporary North America but not in rural Brazil. Again, the issue is cultural appropriateness. PRESENTING APPLIED ANTHROPOLOGY

I see applied anthropology (also called practicing anthropology) as the use of anthropological data, perspectives, theory, and methods to identify, assess, and solve problems that have a social or cultural dimension. The presentation of applied anthropology proceeds in multiple contexts. Anthropologists serve as investigators, social commentators, problem solvers, policymakers, advisers, and evaluators. We express our policy views in meetings, books, journals, and technical reports, through participation in social and political movements, and through professional organizations, such as the Society for Applied Anthropology and the National Association of Practicing Anthropologists. An increasing number of anthropologists work for groups that promote, manage, and assess programs aimed at influencing human social conditions. The scope of applied anthropology includes change and development abroad and social problems and policies in North America. Practicing anthropologists work (regularly or occasionally) for nonacademic clients: governments, nongovernmental organizations (NGOs), tribal and ethnic associations, interest groups, businesses, corporations, and social service and educational agencies. As Karin Tice (1997) observes, practicing anthropologists have to teach their field constantly, because they work in so many contexts in which sociocultural anthropology is unfamiliar. But we also learn in such settings. Many of us never imagined as graduate students or young academics that we would eventually do applied anthropology. But the world has changed, more rapidly and radically than we imagined, and all our field sites have been affected by global forces. Anthropologists are asked to consult about the direction of change in nations, regions, even communities where they originally did academic research. Anthropologists accept consultancies for varied reasons: (1) they allow us to return to the field without the paperwork and uncertainty of the grant application process;

14_107-Bernard.indb 711

6/9/14 10:03 AM

712

Conrad Phillip Kottak

(2) they let us play a role in advocating, designing, implementing, and evaluating policies affecting people we care about; (3) the linkages and transformations in the modern world system are intrinsically interesting; and (4) we can use the money. Applied anthropologists gain personal experience with real-world issues, agencies, and operations. They get to know policymakers, agents, and innovators. While consulting, I have simultaneously used my eyes and ears, as a trained ethnographer, to study people, organizations, and institutions involved in social change. Sometimes, especially working full time for such agencies, anthropologists will feel frustrated. We routinely argue perspectives and positions unfamiliar to, or challenged by, associates. We have to fit into results-driven, deadline-oriented organizations and confront goals we may regard as unwise. As advocates, we argue for our views, but they don’t always prevail. My own applied anthropology mostly has been development anthropology. I’ve consulted as a social analyst in several settings related to economic development. Again, I’ll draw on my experience to indicate some of the varied roles an anthropologist can play in development organizations.

Presenting Anthropology at the World Bank I began my work in development anthropology with an assignment for the World Bank in Madagascar in 1981. My work for the Bank started with a phone call from John Malone, representing the Operation Evaluations Department (OED), charged with evaluating completed projects. A development anthropologist who had read my book on Madagascar (Kottak 1980) had referred Malone to me. Malone described to me a large-scale irrigation project that had gone awry, in the Lake Alaotra region of north-central Madagascar. He mentioned that an accountant often used by the Bank had been sent to Madagascar to evaluate the project, but he had spent his time talking to officials in the national capital, rather than visiting the actual project site, although he had heard there were social problems there. I suggested some likely reasons for those problems. By the time the phone call was over, I had been invited to go to Madagascar as a short-time consultant for the Bank. After a briefing in Washington, I traveled to Madagascar for three weeks in February 1981. Following a procedure I would recommend to any anthropologist in my position, I also arranged to have a local social scientist, Jean-Aimé Rakotoarisoa, whom I knew well from my earlier work in the country, appointed as my coinvestigator. (Rakotoarisoa later served as my counterpart on a USAID project in Madagascar for which I served as social soundness analyst in 1990). He and I (eventually joined by OED’s Malone) traveled to the Lake Alaotra region and spent several days consulting with a series of villagers affected by the project. We discovered some interesting and egregious errors that could have been prevented if anthropologists had been involved in planning and implementing the project and not just called in to find out what had gone wrong after the project was over. My report on the Lake Alaotra project led to other opportunities, as John Malone recommended me to Michael Cernea, the Bank’s long-time senior sociologist, who retired in 1996 but who has done more than anyone to ensure that social analysts are employed by the Bank and that social analysis is an important part of project plan-

14_107-Bernard.indb 712

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

713

ning. In 1983, Michael and John asked me to examine Bank files for a comparative study of the extent and role of social analysis in project planning and implementation, and the relation of social analysis input to the sociocultural appropriateness and outcome of the project. For that 1983–84 study, I reviewed materials on 68 completed rural development projects from all over the world. My task included an assessment of the social and cultural factors that had affected those projects, most of which had been designed during the 1960s and early 1970s, when planners had little experience using sociocultural expertise for development planning. Most of the projects I examined demonstrated a tendency to stress technical and financial factors and to neglect social issues. One of the key findings was that social analysis aimed at identifying and designing socioculturally compatible projects paid off financially. Among the 68 rural development projects studied, the average economic rate of return for culturally compatible projects exceeded 19%, much higher than the comparable rate for incompatible projects, less than 9%. (Sociocultural compatibility was evaluated independently of economic rate of return, to avoid any tendency to identify projects as culturally incompatible once they were known to be economic failures.) I’ve written about this comparative study for several different audiences. In addition to my 1984 report for the Bank, I contributed a chapter (Kottak 1985) in Cernea’s edited volume Putting People First (Cernea 1991). Those publications made the findings available to “the development community.” Because I also wanted to reach anthropologists, I wrote an essay for the American Anthropologist (Kottak 1990) that highlighted the main findings of the study. This strategy of publishing in different genres for different audiences can be useful for many applied anthropologists. Reports written for development agencies have little or no value for tenure and promotion unless and until they are turned into academic studies and published as books or as articles in refereed journals or chapters in refereed books. There is one problem, however, with such publication. Some consultant contracts have a clause banning external dissemination without approval. (I strike out this clause before I sign a contract, but doing so may not always be possible.) David Mandelbaum (1963, 10) argued long ago that studies in applied anthropology “show how theoretical concepts are deployed empirically and how the empirical data feed back into the development of theory.” My own experience leaves me with no doubt that theory and practice are intertwined. The fieldwork and comparison I’ve done in applied anthropology consistently have fed back on my academic work, providing opportunities, practice, classroom material, and research topics for me and my students. Applied anthropology has offered me, mainly an academic anthropologist, new and varied opportunities to do research on change. For me, a key role of the applied anthropologist is to facilitate “bottom up” social change (see also chapters by Schensul et al. and Trotter et al. in this volume). In my work in development anthropology, I’ve tried to ascertain “locally based demand” by visiting a series of communities and promoting wide public consultation, sometimes organized through town meetings. The presentation and implementation of development anthropology itself remains somewhat problematic because of the variety of backgrounds and lack of uniform training among anthropologists. A World Bank official once remarked to me “When

14_107-Bernard.indb 713

6/9/14 10:03 AM

714

Conrad Phillip Kottak

you hire an economist you know what you’re getting; when you hire an anthropologist you have no idea.” Outsiders may perceive anthropologists as being hopelessly idiosyncratic. “If one anthropologist leaves,” said the same official, “there’s no way of telling whether the next one will use the same approach or something completely different.” More consistent training of applied anthropologists in academic settings, including major universities that lack applied anthropology foci per se, may be one solution.

Reports for Development Agencies Documents produced for development organizations go through a long review process during which they may be modified severely—not only to improve them, but also to protect feelings, interests, and reputations—of agency officials and nationals in host countries. The review process varies by agency. At the World Bank, multiple audiences typically hear and read reports, especially when a specific country is involved. Analysts must be sensitive about people’s turfs and egos, especially in a highly bureaucratized organization. Consultants have to learn the culture of the Bank, and that culture is resistant to change. Bank officials like to hire people who have worked for the Bank before and thus are familiar with procedures, formats, and expectations. Reports should read like other Bank reports. Paragraphs are numbered; zippy prose is taboo. Passive constructions are excellent devices for obscuring responsibility. The canon of cultural appropriateness remains in effect in presenting anthropology within development organizations, businesses, or wherever. If you are to be successful as a consultant, you need to avoid overinnovation. You must know, and not deviate too far from, cultural expectations. It is said that World Bank staff spend so much time traveling and attending meetings (and they do) that they must hire short-term consultants to read the reports they hire short-term consultants to write. Development agencies (like businesses) require reports to be prefaced by some kind of executive summary or top-line report, which lists the main findings and conclusions. Often this is all that staff reads. The bulk of the reports are shelved as eventual consultant fodder. (Consulting for the World Bank in Recife, Brazil, in 1993, I was delighted to find a copy of Cernea’s Putting People First [1985] on a prominent shelf in the Bank office library—until I discovered it was wrapped in its original cellophane.) Presenting Anthropology at USAID As an anthropologist, it should be second nature for you to start by assessing the culture of whatever organization you are working in, long or short term. You must determine and know your audience(s) and address it (them). As an employer, USAID is more problematic than the World Bank. Its personnel lack the international diversity and financial expertise that are strong features of the Bank and it doesn’t pay as well. (Salaries and consultant fees are pegged to U.S. congressional salaries.) At the Bank, but especially at USAID in my experience, you not only have to know your audience, you also have to keep it. One of the most troubling characteristics of development organizations is their personnel instability. People come and go, rotating in and out of “missions.” This means that, even if you have a long-term commitment to the country, area, or project, you often have to start over, with someone new. You may have convinced some official

14_107-Bernard.indb 714

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

715

that social analysis is valuable and that a project should unfold in a certain way. Then someone else arrives from Washington, and the process must start again. Anthropologists who have worked in Madagascar for several decades, for example, have seen a succession of USAID mission heads and officers, at all ranks, rotate in and out. The new directors arrive with ideas of their own, often knowing little or nothing about social science—but sometimes with prior positive or negative experiences with anthropologists. So again you reinvent the wheel; you resell the role of anthropology, either in general or for a specific project. Note that this instability applies not only to the major development agencies, but also to the NGOs that increasingly are charged with project implementation. Your relationships must be constantly forged anew. And you also may have to persuade politicians and elected representatives about project soundness and proper implementation strategy. BE TRUE TO YOUR ANTHROPOLOGY

Working mainly among nonanthropologists in an organization, how does one remain an anthropologist, rather than becoming simply another organization man or woman? There’s a real danger that the need to communicate with others within that organization will distance you from an anthropological perspective. The discourse, behavior, and modus operandi of the mix of professionals making up a team tend to converge toward an organizational norm. Anthropologists should remain mindful of the special potential contribution of their field. In our globalizing world, anthropology offers a people-centered vision of social change, while also providing powerful ways of understanding how the world actually works. Lessons derived from the study of human diversity in time and space can and should be applied to benefit humanity. Remembering and appreciating anthropology’s key goals and lessons can help guide us as we present anthropology to diverse audiences. REFERENCES

Abu-Lughod, L. 1993. Writing women’s worlds: Bedouin stories. Berkeley: University of California Press. Angeloni, E. ed. 2011. Annual editions, anthropology, 12/13. New York: McGraw-Hill. Behar, R. 1993. Translated woman: Crossing the border with Esperanza’s story. Boston: Beacon Press. Bernard, H. R. 2011. Research methods in anthropology, qualitative and quantitative approaches, 5th ed. Walnut Creek, CA: Altamira. Brettell, C. B., ed. 1996. When they read what we write: The politics of ethnography. Westport, CT: Bergin and Garvey. Cernea, M. M., ed. 1991. Putting people first: Sociological variables in rural development, 2nd ed. New York: Oxford. Chagnon, N. 2013. Yanomamo, 6th ed. Belmont, CA: Wadsworth Cengage. Clifford, J. 1982. Person and myth: Maurice Leenhardt in the Melanesian world. Berkeley: University of California Press. Clifford, J. 1988. The predicament of culture: Twentieth-century ethnography, literature, and art. Cambridge, MA: Harvard University Press. Duarte, N. 2008. slide:ology: The art and science of creating great presentations. Sebastopol, CA: O’Reilly Media.

14_107-Bernard.indb 715

6/9/14 10:03 AM

716

Conrad Phillip Kottak

Dwyer, K. 1982. Moroccan dialogues: Anthropology in question. Baltimore: Johns Hopkins University Press. Germano, W. 2005. From dissertation to book. Chicago: University of Chicago Press. Harris, M. 1974. Cows, pigs, wars, and witches: The riddles of culture. New York: Random House. Harris, M. 1977. Cannibals and kings: The origins of cultures. New York: Random House. Knauft, B. M. 2012. The Gebusi: Lives transformed in a rainforest world, 3rd ed. New York: McGraw-Hill. Kottak, C. P. 1980. The past in the present: history, ecology, and social organization in Highland Madagascar. Ann Arbor: University of Michigan Press. Kottak, C. P. 1985. When people don’t come first: Some lessons from completed projects. In Putting people first: Sociological variables in rural development, ed. M. M. Cernea, 325–56. New York: Oxford. Kottak, C. P. 1990. Culture and “economic development.” American Anthropologist 93: 723–31. Kottak, C. P. 2006. Assault on paradise: Social change in a Brazilian village, 4th ed. New York: McGraw-Hill. Kottak, C. P., and K. A. Kozaitis. 2012. On being different: Diversity and multiculturalism in the North American mainstream. New York: McGraw-Hill. Krathwohl, D. R., and N. L. Smith. 2005. How to prepare a dissertation proposal: Suggestions for students in education and the social and behavioral sciences. Syracuse, NY: Syracuse University Press. Lee, R. B. 2013. The Dobe Ju/hoansi, 4th ed. Belmont, CA: Wadsworth Cengage. Luey, B., ed. 2004. Revising your dissertation: Advice from leading editors. Berkeley: University of California Press. Malinowski, B. 1961 [1922]. Argonauts of the Western Pacific. New York: Dutton. Mandelbaum, D. G. 1963. The transmission of anthropological culture. In The teaching of anthropology, ed. D. G. Mandelbaum, G. W. Lasker, and E. M. Albert, 1–21. Berkeley: University of California Press. Marcus, G. E., and D. Cushman. 1982. Ethnographies as texts. Annual Review of Anthropology 11: 25–69. Marcus, G. E., and M. M. J. Fischer. 1999. Anthropology as cultural critique: An experimental moment in the human sciences, 2nd ed. Chicago: University of Chicago Press. Narayan, K. 2012. Alive in the writing: Crafting ethnography in the company of Chekhov. Chicago: University of Chicago Press. Pacansky-Brock, M. 2013. Best practices for teaching with emerging technologies. New York: Routledge. Pelto, P., and G. Pelto. 1978. Anthropological research: The structure of inquiry. New York: Cambridge University Press. Peters-Golden, H. 2012. Culture sketches: Case studies in anthropology. New York: McGrawHill. Podolefsky, A., P. J. Brown, and S. M. Lacy. 2012. Applying anthropology: An introductory reader. New York: McGraw-Hill. Richardson, L. 1990. Writing strategies: Reaching diverse audiences. Qualitative Research Methods Series, no. 21. Newbury Park, CA: Sage. Shostak, M. 1981. Nisa: The life and words of a !Kung woman. Cambridge, MA: Harvard University Press. Spradley J., and D. W. McCurdy, eds. 2012. Conformity and conflict: Readings in cultural anthropology, 14th ed. Upper Saddle River, NJ: Pearson. Stein, J., gen. ed. 1982. Random House college dictionary, rev. ed. New York: Random House.

14_107-Bernard.indb 716

6/9/14 10:03 AM

22: Presenting Anthropology to Diverse Audiences

717

Sword, H. (2012). Stylish academic writing. Cambridge, MA: Harvard University Press. Tice, K. 1997. Reflections on teaching anthropology for use in the public and private sector. In The teaching of anthropology: Problems, issues, decisions, ed. C. P. Kottak, J. J. White, R. H. Furlow, and P. C. Rice, 273–84. Mountain View, CA: Mayfield. van Maanen, J. 2011. Tales of the field: On writing ethnography. Chicago: University of Chicago Press. Waterston, A., and M. D. Vesperi, eds. 2009. Anthropology off the shelf: Anthropologists on writing. Malden, MA: Blackwell. Winslow, D. 2007. What makes an NSF proposal successful? Anthropology News 48: 31. Wolf, M. 1992. A thrice-told tale: Feminism, postmodernism, and ethnographic responsibility. Stanford, CA: Stanford University Press.

14_107-Bernard.indb 717

6/9/14 10:03 AM

14_107-Bernard.indb 718

6/9/14 10:03 AM

CHAPTER 23

Public Anthropology Thomas Hylland Eriksen

The term “public anthropology” is slightly odd, for what—if anything— might a “private anthropology” look like? In a certain sense, all anthropology is public, as it entails communication of many kinds and in a variety of settings. Lectures at universities are often open to anyone who wishes to attend, and anthropological publications can be read—in theory—by anybody. There is probably no anthropologist who does not wish to have an audience for his or her ideas. So, in an important sense, all anthropology is public. However, public anthropology, as it is now being used, refers to a specific set of practices and positions within the discipline that aim to reach out beyond the confines of the academy. This can be accomplished through writing for different audiences, engaging in advocacy-oriented work in local communities, or intervening in the long, public, transnational conversation about the ills of the contemporary world and what it means to be human. The common denominator of these practices is the conviction that anthropology should matter not just as an academic pursuit of knowledge but also as a tool to engage with the world in a practical, if not political way. Robert Borofsky (2000/2007), the instigator of the web-based Public Anthropology project (www.publicanthropol ogy.org), defines public anthropology as an activity that “seeks to address broad critical concerns in ways that others beyond the discipline are able to understand what anthropologists can offer to the re-framing and easing—if not necessarily always resolving—of present-day dilemmas.” In another statement, Borofsky says that public anthropology addresses problems beyond the discipline—illuminating the larger social issues of our times as well as encouraging broad, public conversations about them with the explicit goal of fostering social change. It affirms our responsibility, as scholars and citizens, to meaningfully contribute to communities beyond the academy—both local and global—that make the study of anthropology possible. (Borofsky quoted in McGranahan 2006, 257)

In other words, public anthropology consists of an attempt to bridge the gap and overcome the alienation between the anthropological community as a closed professional group and the global society that anthropology studies and in which anthropologists take part. This chapter discusses some of the main forms of public anthropology from a methodological perspective. Some of the questions it raises are: What are the methodological

719

14_107-Bernard.indb 719

6/9/14 10:03 AM

720

Thomas Hylland Eriksen

problems peculiar to public anthropology? Do particular ethical problems arise in public anthropology? How can anthropologists retain their intellectual integrity and ambitions to contribute to knowledge if they embark on their research with explicitly political motivations to contribute to change? The structure of the chapter is as follows. I first outline a history of public anthropology before making a distinction between two main forms of public anthropology. I then discuss the main arenas of dissemination for public anthropology, identifying the main methodological problems in public anthropology, which mostly concern translation in its widest and most encompassing sense. The pressing question is “Who is it for?”; the follow-up question is inevitably “What should the consequences of the ideals of public anthropology be for the way we go about producing and dissemination anthropological knowledge?” PAST SINS AND VIRTUES

The relationship of anthropology to the wider public sphere has gone through several stages, or ebbs and flows. In the nineteenth century, anthropology scarcely existed as an independent intellectual endeavor; it was largely a gentlemanly pursuit or a side effect of exploration and colonization. Those who contributed to the emergence of anthropology as a distinctive field of scientific knowledge, from Lewis Henry Morgan in the United States to Henry Maine and E. B. Tylor in England, positioned themselves in a broader ecology of ideas and the pursuit of knowledge. The professionalization of anthropology as an academic discipline began in earnest around the turn of the twentieth century, enabling later practitioners to withdraw increasingly from social concerns and other approaches to human culture and society. Although many nineteenth-century anthropologists were not “public anthropologists” in the contemporary sense, they certainly related to a broader public—from lay readers to policymakers—in their writings more than did most academic anthropologists of the twentieth century. In addition, many early anthropologists, especially in the United States, were involved in what would today be called radical advocacy or action anthropology. Luke Lassiter notes that: [L]ong before Bronislaw Malinowski insisted that anthropologists move “off the verandah” and into the everyday lives of the natives . . . many BAE [Bureau of American Ethnology] ethnologists had moved into Native communities and were participating in people’s everyday lives, doing fieldwork in collaboration with Indian informants, and, in some cases, following in the tradition of Morgan, acting on behalf of their “subjects.” (Lassiter 2005, 86)

The increasing institutionalization of anthropology as an academic discipline in the twentieth century enabled many anthropologists to effectively withdraw from the surrounding society (Eriksen 2006; Low and Merry 2008). Concerns voiced by some, such as A. R. Radcliffe-Brown (1957), to make anthropology a “natural science of society” modeled on physics and biology, encouraged this kind of retreat into the ivory tower, and, as the internal demographics of anthropology soared after World War II, the professional community grew large enough to begin to spin a cocoon around itself. Like a growing empire, it increasingly became self-contained, self-reproducing,

14_107-Bernard.indb 720

6/9/14 10:03 AM

23: Public Anthropology

721

and self-sufficient, until the sheer demographic growth, decades later, again led to porous boundaries and defections. However, there has been no straightforward movement from openness to closure. Important anthropologists who contributed to the very institutionalization mentioned, notably Franz Boas, were engaged in broader societal issues, and Boas was a leading critic of racist pseudoscience. Among his students, Margaret Mead hardly needs an introduction, but as the author of 44 books and more than a thousand articles, keeping the pace up right up until her death in 1978, she was arguably the public anthropologist par excellence in the twentieth century. There were also many others whose work was read outside the academy and who engaged in various ways with the world at large. Bronislaw Malinowski gave lectures on primitive economics to anyone who would care to listen; Marcel Mauss was engaged in French politics as a moderate socialist; and one could go on and on. Moreover, applied anthropology had been a subfield—often unjustly disparaged by those involved in “pure research”—since well before World War II. As noted by David Mills (2006, 56–57), anthropologists had, since the early twentieth century, tried to “convince the Imperial government that anthropology served a useful purpose and deserved funding.” Although applied research was funded by the British Colonial Social Science Research Council until 1961 (Pink 2006), little basic anthropological research received such funding (Goody 1995). Anthropological methods and anthropological knowledge have, nevertheless, at various times, been deemed useful by governments and business leaders, most recently in the Human Terrain System (HTS) of the U.S. military forces, where practicioners from anthropology (and other subjects) are drawn on to enhance knowledge of local circumstances in war areas. Deeply controversial among U.S. anthropologists, the HTS was denounced in a statement issued by the American Anthropological Association in 2007. This is not the place to enter into a pro-et-contra discussion, but it may be pertinent to note that strictly speaking, giving courses to army personnel on their way to Afghanistan and Iraq can perfectly well be seen as a form of public anthropology, provided soldiers are seen as card-carrying members of the public. Such courses may lend legitimacy to the wars (which to many is problematic), but they may also, as has been argued by embedded anthropologists in U.S. and NATO forces, reduce suffering by providing a contextual and sympathetic understanding of local life-worlds. In a sense, anthropologists have always engaged with publics outside of anthropology. Sometimes, this has led to their academic marginalization—anthropologists could easily be written off as intellectually lightweight if they got involved in advocacy or applied work, say, for development agencies—and there has been, as noted by many (e.g., Borofsky 2011; Pels and Salemink 1999), a clear and arguably unproductive tendency to rank pure research above applied research. Similarly, the hierarchy ranking tough academic writing for people in the know above lucid writing for the general public is also debatable. The fact remains that most anthropologists who are familiar to students put most of their intellectual energy into basic research and theory, and we therefore need to be reminded that they have all coexisted with other anthropologists, who either went out of their way to establish a broader dialogue about the human condition or actively sought to mitigate suffering and contribute to social change.

14_107-Bernard.indb 721

6/9/14 10:03 AM

722

Thomas Hylland Eriksen

Public anthropology as such is, in other words, not something new. Nevertheless, the problematization of distinctions that were formerly taken for granted, notably between “pure” and “applied” work, and the development of a reflexive and critical discourse about the ways in which anthropology can be made relevant outside the academy, has been on the rise in recent years. This development cannot be attributed to important individual initiatives such as Borofsky’s Public Anthropology network and associated book series but must be understood as a broader structural tendency. Already in the 1980s, anthropologists working in the Global South noted that many of the people they came into contact with had highly articulate and reflexive views of their own history, culture, and identity. They certainly did not feel the need for anthropologists to identify who they were; in many parts of the world, local intellectuals had indeed read some anthropology and were familiar with its concepts. They were able to identify themselves and use some of the tools offered by anthropology to develop their own existential and political agendas. In our world of multiple transnational networks and global flows, the fiction of “us, the knowers” and “they, the objects of study,” which was always objectionable, has now become untenable, and anthropologists now venture into fields and delineate their topics of inquiry in ways that were unheard of only a generation ago (see MacClancy 2002 for a sample). As Sam Beck and Carl Maida (2012) put it, the contemporary world is borderless in every sense. The consequences of the destabilization of boundaries for the anthropological endeavor are many, and some of the most important consequences become evident in the debates around public anthropology: Who can legitimately say what, and on whose behalf can they say it? What are the benchmark criteria for good ethnography? What can anthropologists offer to the societies they study? And—in a very general sense—what is the exact relationship between anthropological research and the social and cultural worlds under study? These questions, which were always relevant, have become inevitable, and increasingly difficult to answer, in the borderless world of the twenty-first century. DOES PUBLIC ANTHROPOLOGY HAVE TO BE “ENGAGED”?

Recall Borofsky’s delineation of public anthropology. He mentions “the explicit goal of fostering social change” as a defining characteristic of public anthropology. Thinking about my own practice, mostly as a writer in many genres and as a public speaker, I cannot unequivocally subscribe to this view. In fact, I often mount the podium with more questions than answers. Sometimes, the role of the public anthropologist simply consists in raising new questions, enabling fresh perspectives on the world, without a clear normative or political message. Often, an anthropological gaze on some familiar phenomenon “at home” can be subversive in that it creates a sense of alienness or Verfremdung; it can lead to critical self-reflection or even a hint of self-irony, but the outcome in terms of fostering social change may be unclear. In my view, the “engagement” of an engaged anthropology may amount to an engagement for the sake of a broader and better understanding of the world, not necessarily an engagement for the liberation of disenfranchized groups or for greater social justice. Besides, in many situations, there will be politically engaged anthropologists representing opposing views on the issue in question. Doubtless, anthropologists involved in the HTS believed that they

14_107-Bernard.indb 722

6/9/14 10:03 AM

23: Public Anthropology

723

contributed to making the world a slightly better place by presenting army personnel with an insider view of local culture and society. On the other hand, I agree that public anthropology should be critical of concepts and critically reflexive regarding the role of anthropology outside the academy. By this token, public anthropology is not the same thing as applied anthropology, although there are important overlaps (Singer 2000). Applied anthropology is often policy anthropology and serves a particular purpose as defined by those who commisioned it; public anthropology is ideally flexible and develops through dialogue with the surrounding world. In Martijn de Koning’s view, there “are two main principles of public anthropology (that also distinguishes it from applied anthropology): 1. Public accountability, 2. Attempting to understand the structures that frame and restrict solutions to problems” (de Koning [2009]). The first criterion emphasizes the dissemination of knowledge and commitment to participate in a broad exchange of ideas that is not bounded by the academic discipline of anthropology (nor by academia at all); the other criterion commits public anthropology to grapple with questions engaging local populations, presumably framing them in new ways. Sita Venkateswar (2007) argues that anthropology should always be public in the latter sense and that concerns relating to the well-being of local communities and social justice should form part of the very backbone of the discipline, which is morally obliged to give something back to the people on whom it depends. The main challenge in this approach concerns the specification of the needs of those local communities, since such communities tend to be diverse and divided on important issues. It is not enough for a would-be action anthropologist to take “the side of the people,” for the people, in fact, are persons with differing subject-positions, needs, views, and persuasions. Supporting the poorest and most downtrodden may also not lead to positive social change; it may also be argued (rightly or wrongly), as social change is usually initiated by certain segments of the middle classes, not the poorest. THE PUBLIC ANTHROPOLOGIST AS WRITER

Public anthropology, which amounts to various alternative forms of knowledge production, is—in spite of its many potential applications—not necessarily directly applicable to social issues, although it nearly always has transformative potential. Allow me to use an example from my own working experience in public anthropology. Quite recently, I published a book in Norwegian about waste (Eriksen 2011). Intended for the general reader, it was marketed as a trade book and reviewed in the mainstream press. The book draws extensively on anthropological perspectives on waste and the purity/impurity boundary, with Mary Douglas’s seminal work (Douglas 1966) and a critical discussion of her views as the field of gravity throughout. Through both historical and contemporary examples, the book shows how the purity/impurity boundary exists everywhere but it is contested, shifts, wriggles, and varies. The aim is to encourage readers to rethink the value/rubbish boundary and think about ways in which waste can be transformed into something valuable (briefly, through recycling, storage, or change of perspective). Although there is an environmentalist agenda implicit in the book, it does not claim moral authority or otherwise try to instruct the readers. It merely offers some facts (many of them “fun facts,” as correctly pointed out

14_107-Bernard.indb 723

6/9/14 10:03 AM

724

Thomas Hylland Eriksen

by a critical reviewer) about waste at both macro and micro levels and then provides some tools with which the reader may reconfigure his or her relationship to waste. The book does not have an unequivocal political message but is implicitly political in its way of raising issues and pointing at paradoxes. The methodological point here concerns styles of writing and ways of addressing different readerships. As famously pointed out by Geertz (1988), anthropologists spend a great deal of their time writing. Writing, moreover, is a skill that can be taught and learned, although the difficulty of writing well is somewhat underestimated at most anthropology departments. When a postgraduate returns from the field and is asked by her or his supervisor to “begin to write up your ethnography,” she or he might well panic. How on earth do you begin to write up your ethnography when nobody has taught you how to do it? Writing down (field notes) is much easier than writing up. The cluster of problematic issues associated with writing in anthropology is a key challenge for public anthropology. Because its aim is to engage a nonacademic audience, not to win recognition by colleagues, public anthropology depends on a different and probably broader palette of writing styles than strictly academic anthropology. Metaphors, riddles, paradoxes, and even snippets of poetry may improve and enhance a book that is not curriculum fodder, while such elements might appear as unwelcome distractions in an academic publication. Public anthropologists need not only be aware of the social and political circumstances of the production of their work, but must also make conscious decisions as to what genre to write in (see Becker 2007; Narayan 2012; Sandelowski 1998; Van Maanen 2011, and the chapters by Kottak and by Schnegg in this volume). REWARDS AND PERILS OF GOING PUBLIC

Several commentators on public anthropology have in recent years written slightly disparagingly of “purely academic” anthropology, which allegedly, confined to its ivory tower, remains aloof from real problems in the real world and measures its significance on the number of citations from colleagues. Borofsky, for example, discusses “how becoming a ‘professional’ academic often means avoiding real problems of real people and instead focusing on problems that do not threaten the status quo” (Borofsky 2011, loc. 340). There is certainly some truth to this view, but it should be kept in mind that public anthropology is not for everyone, and in this, I disagree with Venkateswar (2006). First, although theoretically sophisticated and analytically stringent writing requires great skill (and, as a genre, often possesses its own beauty), writing for a nonacademic audience is no less difficult, but in a different way—not everyone can do it. Second, engaging with a broader public, often in a highly politicized context, is challenging in ways that not all scholarly researchers are prepared or motivated for. This may be equally true of both main kinds of public anthropology, namely civic engagement, where the anthropologist participates in a kind of action research (see Schensul et al., this volume), public interventions, and critical analyses. The proverbial pleasant-butnot-always-productive politeness of the academic seminar contrasts starkly with the heated polemics and pithy one-liners of the public debate. Those in charge of courses in public anthropology that focus on methods should probably consider media train-

14_107-Bernard.indb 724

6/9/14 10:03 AM

23: Public Anthropology

725

ing as an important element (see, e.g., the media training programs offered by the U.K.’s Economic and Social Research Council [http://www.esrc.ac.uk/funding-andguidance/impact-toolkit/tools/media/media-training.aspx]). The rewards and perils of going public with one’s knowledge and perspectives are two sides of the same coin: One is inevitably drawn into controversies. This is equally true of participants in the domestic debates on—say—cultural diversity and racism, and of activist-anthropologists working with local communities to improve their lot and to shed light on their plight. My late colleague Marianne Gullestad, who was both an eminent academic anthropologist and an engaged public anthropologist (and an excellent writer to boot), was reminded of the difficulties in public anthropology when she published a short version of an academic article in a domestic newspaper. The original article (Gullestad 2004) was an analysis of Norwegian mental schemata leading to the exclusion of black people from the category of “Norwegianness.” As an example, she used a situation when an African jogger in the forested areas outside the capital made many white Norwegians turn their heads in surprise. The man was seen as “matter out of place.” This article is now compulsory reading in several university courses in anthropology. Using the same example in the short newspaper article, Gullestad was immediately castigated both online and offline for her “hysterical political correctness” and for misunderstanding the situation: The African man was noticed not because he did not fit in (the forest as a “white” area, unlike the inner city), but because jogging Africans in these areas are so uncommon that they must inevitably be noticed. Various comments, many of them highly critical of Gullestad’s perspective, some merely pejorative but others more calm and reasoned, flourished for a few days, maybe a week, before the restless gaze of the public turned elsewhere. In other words, public debate in a mediatized public sphere is accelerated, polemic, rude, and politicized in contrast to the debate in the seminar room, which is measured, slow, considered, and polite. There is no obvious answer as to which is better, and arguing that a larger and broader audience is always better than an academic one would be both populist and anti-intellectual. The controversial public statement needs the seminar room, but the latter loses much of its purpose without the former. PUBLIC ANTHROPOLOGY AS CIVIL ENGAGEMENT

What distinguishes public anthropology from purely academic anthropology is its approach to the production and dissemination of knowledge. Much anthropology that engages with a broader public consists of civic engagement for change (see Beck and Maida 2012 for examples), where knowledge is produced, often in collaboration with the community under study, with a specific aim in mind. The debate about the ethical obligations toward local communities was intensified in the autumn of 2000, when the investigative journalist Patrick Tierney’s book Darkness in El Dorado (Tierney 2000) claimed to have documented grossly unethical behavior from a research team in South America that included anthropologists. Although the most serious allegations were eventually disproved, an AAA committee later set down some general ethical outlines for field anthropologists, stressing the collaborative nature of research more strongly than what had formerly been common:

14_107-Bernard.indb 725

6/9/14 10:03 AM

726

Thomas Hylland Eriksen

[T]he anthropology of indigenous peoples and related communities must move toward “collaborative” models, in which anthropological research is not merely combined with advocacy, but inherently advocative in that research is, from its outset, aimed at material, symbolic, and political benefits for the research population, as its members have helped to define these. . . . Collaborative research involves the side-by-side work of all parties in a mutually beneficial research program. (AAA El Dorado Task Force, quoted in Lassiter 2005, 84)

“Collaborative ethnographic practice,” Luke Lassiter argues (2005, 96), “is now converging with an engaged, public anthropology.” This is clearly true of some public anthropologists; the methodological difficulties inherent in this way of working are nonetheless obvious (see Schensul et al., this volume). Note that there are problems associated with data collection or production. In classic anthropology, the subject–object relationship during fieldwork is unquestioned. The ethnographer asks questions, makes observations, takes notes. Following the “reflexive turn” of the 1980s, the anthropologist would often position him- or herself in the text, but without problematizing the question of authorship. Collaborative ethnographies, in contrast, raise questions of methodological authority and validity. The authorship would nearly always be attributed to the anthropologist in such endeavors, but he would, to a greater or lesser extent, portray himself as the mouthpiece of the community’s interests. However, with an explicitly political agenda in mind, the production of data must inevitably be skewed. Put even more bluntly: If one already has the answer, why bother to do the research? A reasonable rejoinder to this objection is that one may have the answer to the “what” from the outset (say, clean drinking water, less drug-related violence, etc.), but not to the “how.” Moreover, there is no a priori reason to assume that normatively guided research of this kind produces less valid knowledge than “pure research.” All research is slanted one way or another, either because of the researcher’s personal interests, his or her subject position, or the theoretical and methodological training he or she has received before embarking on fieldwork. The objection to public, politically engaged anthropology as being one-sided and therefore less trustworthy than lofty academic research cannot, however, simply be brushed off. It is true that the element of intellectual surprise essential for scientific discovery may be less pronounced in politically guided research than in more openended, purely curiosity-driven projects. Yet, the impure forms of doing and writing anthropology characteristic of public anthropology, where genres mix and different interpretations of reality are confronted “as one goes along,” are no less demanding and can, in their way, be at least as illuminating of the world as the more detached variety. And, so long as the anthropologist is open and sincere about his or her aims and objectives, there is no reason to doubt the veracity of engaged anthropology. In other words, there are some methodological challenges involved here, but they are not insurmountable. It nevertheless remains true that anthropological research that explicitly aims to contribute to social change, whether commissioned (“applied”) or not, is far from the upper echelons of anthropological publishing, where detachment is seen as a guarantee for intellectual integrity (see Shore and Wright 1997; Wright 2006). This implicit

14_107-Bernard.indb 726

6/9/14 10:03 AM

23: Public Anthropology

727

hierarchy may well be challenged, not only on political or humanistic grounds but also with reference to conventional criteria for academic excellence. A problem rarely discussed by the proponents of public, collaborative, or action anthropology concerns who to represent and whose interests to advocate. Nancy ScheperHughes (1995) has argued that anthropology in the contemporary world must place itself firmly within the struggles of the oppressed and on the side of justice (see also Low and Merry 2010). Yet, it is not always obvious whose local interests should be represented and how to do it. Opinion is divided among anthropologists just as much as among policymakers, local populations, and public figures. In this regard, it may be said that anthropology has a problem with its own internal diversity and intellectual openness if it is taken for granted that all anthropologists hold roughly the same political views and are in a privileged position to make sound ethical judgements. This is, of course, debatable, and, just as Jonathan Spencer (2010) admits to feeling incompetent in a situation of political turmoil in Sri Lanka, many of us have made mistaken ethical judgments. Still, Scheper-Hughes’s point remains an important one: Rather than remaining detached and neutral in analyzing the challenges and problems facing local populations, anthropologists are ethically obliged to expose structural conditions that militate against the well-being of the people they study. In doing so, a truly public anthropology may “speak truth to power” while also producing excellent research (see, e.g., Farmer 2010; Scheper-Hughes 1992; Tsing 2005). INTERDISCIPLINARY PUBLIC ANTHROPOLOGY

A different way of carrying out public anthropology is by engaging in interdisciplinary discourse. In some ways, the methodological challenges here are similar to those of civil engagement. In both cases, translation skills are essential. Our positioned knowledge, shaped by our biography, our training, reading, and convictions, is confronted with that of others. Confrontations are inevitable, fusions of horizons possible, enrichment likely. Moreover, something important may be at stake in both cases. Although working in a local community to save adolescents from drug addiction may immediately seem more laudable than criticizing neo-Darwinian reductionism, the latter may be just as necessary, and may have similarly beneficial effects for humanity, as the former. And, to repeat, cultural translation, a key methodological skill for any anthropologist, is at the core of both endeavors (see Bergendorff 2007; Churchill 2005; Ingold 1993; Maranhão and Streck 2003; Mose Brown and de Casanova 2013). Natural scientists typically ask: “What is a human being, seen in objective terms?”; sociocultural anthropologists might ask “What does it mean to be a human being?” If one is interested in understanding the life-worlds of others, one must use the method of cultural relativism, which entails, simply, to make an effort to learn to see the world as they see it—not to become one of them, but to become capable of explaining to colleagues, students, and others in what kind of socially constructed world the others live. It is far from uninteresting to find out which ideas the Dayaks have about life after death, how the Maasai view the relationship between money and cattle, or how the Ainu in northern Japan debate the effects of climate change on their environment, but these questions do not in themselves generate answers about objective characteristics of humanity.

14_107-Bernard.indb 727

6/9/14 10:03 AM

728

Thomas Hylland Eriksen

Some time ago, I was drawn into a lengthy newspaper debate with a medical scholar about social constructivism, which he saw as an intellectually lazy and irresponsible subset of relativism. Aware that he was a long-term researcher and adviser on HIV and AIDS, I was disappointed to discover that he had no time for the cultural dimensions of the ailment. If, after all, some South Africans believe that the presence of certain large, black birds roosting in the trees near the village signal that it will soon be struck by AIDS, health workers and researchers would—in my view—be well advised to know this before launching their information campaigns. An insight from interdisciplinary public anthropology is that all truths are partial, and a piece of methodological advice is to acquire “the native’s point of view” and encourage the native (be it a biologist, a journalist or a medical scholar) to acquire yours for the interdisciplinary dialogue to be made possible. Spending most of their working time inside the cocoon, as many academics do, makes them—even many anthropologists—surprisingly inept cultural brokers and translators. IMPURE GENRES

Public anthropologists thrive in the interstices and frontier areas between genres; they challenge the conventions of academic life as much as they prod the boundaries of public discourse. Many of them cling precariously to academic positions or are both inside and outside the academy. One of the most important public anthropologists in the United Kingdom, Sean Carey, does not hold a high-prestige academic job, but contributes regularly to the New Statesman, The Guardian, and other media, including high-profile blogs. A pioneer in the anthropological blogosphere, the Dutch anthropologist of Islam, Martijn de Koning, has been active online for more than a decade, continuously updating his website (http://religionresearch.org/martijn), creating innumerable links to other sites, writing informed blog entries debunking myths about Muslims, and providing a forum for open discussion—by anthropologists and nonanthropologists, Muslims and non-Muslims—but his academic publications are mostly in Dutch and his work is so far less well known, but hardly inferior in quality, to that of many other anthropologists of Islam. Websites run by anthropologists—from the group blog Savage Minds to Barbara Miller’s anthropologyworks or Jason Antrosio’s Living Anthropologically—tend to be impure. They interact with readers, with other websites, with actuality and news items, and function in a wider ecology of ideas that is far removed from the closed circuits and slow rhythms characteristic of the academic article. Contributing to this public sphere requires a different mindset than that needed in regular academic publishing. Writing for newspapers and other mass circulation print media similarly requires mental and stylistic adjustment. The implicit taken-for-granteds, everything that “goes without saying because it comes without saying” (Bourdieu), exist there just as they do in the academic world, but the shared presuppositions and genre requirements differ. Before launching into an anthropological argument for a general audience, it is often well advised to tell a story from everyday life or mention a news item. The result is an “impure” text. Perhaps the most difficult challenge when writing anthropologically informed texts for nonacademic audiences is not to do with style or content, but with speed. Although

14_107-Bernard.indb 728

6/9/14 10:03 AM

23: Public Anthropology

729

academic publications take long to write and even longer to get into print, with a sometimes exasperating, but usually necessary review process followed by proofreading and final queries, the blog is online immediately and the newspaper article after a few days. Response is also immediate and direct, unfiltered and, as mentioned above, frequently politicized in ways anthropologists are not always prepared to deal with. Mediated communication has its pros and cons. Of the media mentioned so far, the blog probably is the closest to a conversation, Facebook and Twitter being even more immediate. Other media, in spite of their accelerated character compared with the academic publication, incorporate a delay between statement and response. Dialogue is fairly slow. An appearance on television is essentially a one-way communicative act. Viewers may react through other media and even through email, but most do not. In my experience, public meetings tend to be much more satisfactory than television appearances. In such forums, if one has the privilege of giving a talk on some anthropologically relevant topic, response is immediate, two-way, and dynamic, and a real discussion can take place. Like other ways of communicating anthropology to the public, however, this genre is impure and entails a departure, often disquieting and unpleasant (even if necessary), from the academic comfort zone, but speed and intensity can be addictive as well. POPULARIZATION OR CRITICAL INTERVENTIONS

The most public of all public anthropologists must, by most measures, be the most popular ones, the household names of the public sphere. However, popularization is not the same thing as public anthropology, and to some, the two can be opposed to each other. Popularization need not entail a wish to “foster social change” as Borofsky puts it (see also de Koning 2009; Forte 2011); it can, on the contrary, serve to confirm preexisting prejudices and worldviews. As a rule, the most popular books are rarely the most critical ones. The latter offer resistance and tend to be challenging instead of entertaining, often more painful than pleasant. Most spokespeople of public anthropology emphasize its critical dimension. Maximilian Forte, on his website Zero Anthropology, thus writes that there “is no point in being a mute bystander as public debates rage about race, the family, violence, religion, and thus act like some dog in the manger” (Forte 2011). It may seem almost self-evident that anthropologists, given their comparative method, global vision, and, especially, commitment to seeing the world from the perspective of everyday experience, should take a critical view of the status quo. A main purpose of public anthropology consists in speaking truth to power, but as every anthropologist knows, truths are partial, and it cannot be ruled out that some public anthropologists represent, or identify with, that segment of the public that is loyal to the powers that be. It should nevertheless be pointed out that the popularizer and the public anthropologist are not always the same person—in fact, they can be opposed to one another. To be popular, the popularizer has to pander to widespread sentiments; he or she has to be an organic intellectual of sorts, sometimes stating the obvious, sometimes giving an apt and accurate phrasing of notions already held semi-consciously by the imagined readership. The popularizer must convince the readers that she or he is on their side. The critical intellectual is obliged to say unpopular things in a striking way. The very

14_107-Bernard.indb 729

6/9/14 10:03 AM

730

Thomas Hylland Eriksen

point of criticism, in Foucault’s view (2001), is to speak against the “people,” against received wisdoms and representations that are dominant. In making this argument, Foucault draws on the Greek notion of parrhesia, which refers to the activity of speaking against power in a way that entails a certain personal risk. To what extent metropolitan anthropologists run a personal risk by speaking against power is debatable. As described by Jonathan Spencer (2010) in an account of the role of social scientists in the Sri Lankan conflict, domestic intellectuals in conflict areas—whether academics or not—run a far greater risk than foreign ethnographers. A more difficult question concerns the degree to which anthropologists are obliged to speak against power and to what extent they can do so efficiently. In a critical discussion of Hugh Gusterson and Caroline Besterman’s Why America’s Top Pundits Are Wrong: Anthropologists Talk Back (Gusterson and Besterman 2005), Matti Bunzl (Bunzl 2008) argues that merely showing that the world is more complex than claimed by the organic intellectuals of the powers that be, is unlikely to change the world. Indeed, he concludes, there is a risk of “anthropology collapsing under the exhaustion of perfectly rendering the world’s complexities and discarded by those not so fond of it in the first place” (Bunzl 2008, 58; see also Gusterson and Besterman 2008). Bunzl’s alternative consists in looking for alternative generalizations and simplifications that do not come across as ethnocentric or condescending. I concur (Eriksen 2006), and in a related discussion about anthropologists’ responses to Jared Diamond’s bestseller of cultural history, Guns, Germs and Steel (Diamond 1997), Borofsky makes a similar point when he concludes that: “Few anthropologists have offered alternative big picture syntheses. Rather they draw attention to themselves by criticizing Diamond’s details. It may be intellectually safe. But it lacks the courage to synthesize broadly, to move beyond a narrow niche perspective” (Borofsky 2011: loc. 3655). In spite of the apparent, and often real, contradiction between the role of the popularizer and that of the public intellectual, many seem to combine the roles rather well. Indeed, this tradition runs deep: Margaret Mead and Marvin Harris were derided as popularizers and yet did much for making anthropology visible beyond the academy. Scholars who combine these roles may do so by popularizing in provocative and challenging ways. Sometimes this can be done by making statements that are perceived as outrageous and scandalous by parts of the target group while confirming the preexisting worldview of others. The famous evolutionary biologist Richard Dawkins typically does this. His often rash, but pithy attacks on any form of belief that does not conform to scientific rationalism are hailed as pure genius (or common sense) by people who have been in agreement with him all along, and are perceived as irritating or dangerous by various others. Margaret Mead’s and Marvin Harris’s work had a similar function. The methodological problems entailed by using anthropology as a tool of social and cultural critique, away and at home (a distinction of dubious validity, by the way), thus concern translation of the academic discourse, work on genre and style, dealing with speed, and politicization (which is not always desired or desirable). The main challenge, most self-professed public anthropologists seem to agree, is not one of method but of identity: Public anthropology counts for less in academia than purely academic research.

14_107-Bernard.indb 730

6/9/14 10:03 AM

23: Public Anthropology

731

PUBLIC ANTHROPOLOGY AS A MAINSTREAM ACTIVITY

A key methodological challenge for the public anthropologist, illuminated by many of the examples in this chapter, concerns translation in the broadest sense of the word. The internal language of academia needs to be translated into a language suitable for the audience in question. The research agenda needs to be translated into an agenda perceived as relevant and, perhaps, liberating for the target group. The collaborative character of much public anthropology requires translation of concepts and perspectives in both directions. Also, when public anthropology involves interdisciplinarity, translation between traditions of learning is essential. The greatest challenge for the public anthropologist nevertheless appears to be liberating him- or herself from the hierarchies, symbolic capital, and conventions regulating academic life and academic careers. Most of the anthropologists who have sought to delineate public anthropology in recent years seem to agree that the price to pay for being of service to society consists in being academically devalued. Given this background, it may be appropriate to argue, in conclusion, that public anthropology has the makings of becoming a mainstream, indeed a core, activity in anthropological work. This is not to say that anthropology as a discipline ought to represent a particular political agenda. That would have been tantamount to intellectual suicide. Nor should anthropologists feel obliged to write for an undefined, but nonacademic readership. That would blunt our analytical instruments and weaken our methodological stringency. What I have in mind is the potential of anthropology to become a key bridging discipline in intellectual life, a discipline at the crossroads of many kinds of worlds— cultural, social, professional, intellectual—and, not least, at the crossroads between the academy that sanctions our output and the people to whom we owe the very substance of what we do. Seen in this way, anthropology could represent, as Keith Hart once put it, a continuation of the Kantian project of cosmopolitanism, a contribution to the long conversation about humanity (Hart 2003). For now, four arguments in favor of public anthropology, all of them raising particular methodological challenges, may be mentioned. • First, the comparative perspective and its accompanying Verfremdung or defamiliarization effects belong properly to anthropology. The insight that society could have been very different, that you and I could have held other values and had other expectations of life, is one of anthropology’s greatest gifts to humanity. • Second, the anthropological insistence on the primacy of the insider’s view, the lived experience of “ordinary people” and the texture of their life-worlds, is usually missing even in the better newspaper articles about far-flung places. Again, this perspective can make a difference. Public anthropologists need to improve their skills in this area, mixing the emotional with the enlightening in more compelling ways, as media impact often depends on the ability to convey gripping stories about individuals. • Third, anthropologists are rather skilled at exposing oversimplifications by insisting on a more nuanced and complex view, be it in the realm of military expansion or that of biological determinism, but again they need to become better communicators. • Fourth, anthropologists (and academics in general) should be less afraid of stating the obvious (or what, at any rate, is obvious to them). Whenever we move out of the ivory tower, the intended audience does not include people who can tell post-development

14_107-Bernard.indb 731

6/9/14 10:03 AM

732

Thomas Hylland Eriksen from post-structuralism, but that does not mean that they are stupid or that their concerns are irrelevant.

The result may be a hybrid, amphibious way of “practicing anthropology in a borderless world” (Beck and Maida 2012) while remaining true to the original ideals of the discipline, namely to make sense of human diversity and—from Boas onward—making the familiar exotic and the exotic familiar. In this, there is both a huge liberating potential and very considerable methodological challenges in the future. REFERENCES

Beck, S., and C. A. Maida, eds. 2012. Public anthropology in a borderless world. New York: Berghahn. Becker, H. S. 2007. Writing for social scientists: How to start and finish your thesis, book, or article, 2nd ed. Chicago: University of Chicago Press. Bergendorff, S. 2007. Semantic frameworks: Methodological reflections on how to understand Mekeo “sorcery” and “magic.” Field Methods 19: 407–24. Borofsky, R. 2000/2007. Defining public anthropology: A personal perspective. http://www .publicanthropology.org/public-anthropology/ (accessed September 2, 2013). Borofsky, R. 2011. Why a public anthropology? Kindle book, Center for Public Anthropology. Bunzl, M. 2008. The quest for anthropological relevance: Borgesian maps and epistemological pitfalls. American Anthropologist 110: 53–60. Churchill, C. J., Jr. 2005. Ethnography as translation. Qualitative Sociology 28: 3–24. de Koning, M. 2009. Public anthropology—10 years from researchpages to closer. http://religion research.org/martijn/2009/11/09/public-anthropology-10-years-from-researchpages-to-closer -19992000-20092010/ (accessed September 2, 2013). Diamond, J. 1997. Guns, germs and steel: A short history of everyone for the last 13,000 years. London: Jonathan Cape. Douglas, M. 1966. Purity and danger. London: Routledge & Kegan Paul. Eriksen, T. H. 2006. Engaging anthropology: The case for a public presence. Oxford: Berg. Eriksen, T. H. 2011. Søppel: Avfall i en verden av bivirkninger (Rubbish: Waste in a world of side-effects). Oslo: Aschehoug. Farmer, P. 2010. Partner to the poor: A Paul Farmer reader. Berkeley: University of California Press. Forte, M. C. 2011. Beyond public anthropology: Approaching zero. Keynote speech, 8th Public Anthropology Conference, American University, Washington DC, October 14–16. Foucault, M. 2001. Fearless speech, ed. J. Pearson. New York: Semiotext(e). Geertz, C. 1988. Works and lives: The anthropologist as author. Cambridge: Polity. Goody, J. 1995. The expansive moment: Anthropology in Britain and Africa 1918–1970. Cambridge: Cambridge University Press. Gullestad, M. 2004. Blind slaves of our prejudices: Debating “Culture” and “Race” in Norway. Ethnos 69: 177–203. Gusterson, H., and C. Besterman, eds. 2005. Why America’s top pundits are wrong: Anthropologists talk back. Berkeley: University of California Press. Gusterson, H., and C. Besterman. 2008. A response to Matti Bunzl: Public anthropology, pragmatism, and pundits. American Anthropologist 110: 61–63. Hart, K. 2003. Epilogue. In Globalisation—Studies in anthropology, ed. T. H. Eriksen, 217–27. London: Pluto.

14_107-Bernard.indb 732

6/9/14 10:03 AM

23: Public Anthropology

733

Ingold, T. 1993. The art of translation in a continuous world. In Beyond boundaries: Understanding, translation and anthropological discourse, ed. G. Palsson, 210–30. Oxford: Berg. Lassiter, L. E. 2005. Collaborative ethnography and public anthropology. Current Anthropology 46: 83–116. Low, S., and S. Engle Merry. 2010. Engaged anthropology: Diversity and dilemmas. Current Anthropology 512: S203–S226. MacClancy, J., ed. 2002. Exotic no more: Anthropology on the front lines. Chicago: University of Chicago Press. Maranhão, T., and B. Streck. 2003. Translation and ethnography: The anthropological challenge of intercultural understanding. Tucson: University of Arizona Press. McGranahan, C. 2006. Introduction: Public anthropology. India Review 5: 255–67. Mills, D. 2006. Dinner at Claridges? Anthropology and the “captains of industry,” 1947–1955. In Applications of anthropology: Professional anthropology in the twenty-first century, ed. S. Pink, 55–72. Oxford: Berghahn. Mose Brown, T., and E. M. de Casanova. 2013. Representing the language of the “Other”: African American vernacular English in ethnography. Ethnography. http://eth.sagepub.com/ content/early/recent (accessed August 31, 2013). doi doi:10.1177/1466138112471110. Narayan, K. 2012. Alive in the writing: Crafting ethnography in the company of Chekhov. Chicago: University of Chicago Press. Pels, P., and O. Salemink, eds. 1999. Colonial subjects: Essays on the practice of anthropology. Ann Arbor: University of Michigan Press. Pink, S. 2006. Introduction: Applications of anthropology. In Applications of anthropology: Professional anthropology in the twenty-first century, ed. S. Pink, 3–26. Oxford: Berghahn. Radcliffe-Brown, A. R. 1957. A natural science of society. Glencoe, IL: The Free Press. Sandelowski, M. 1998. Writing a good read: Strategies for re-presenting qualitative data. Research in Nursing & Health 21: 375–82. Scheper-Hughes, N. 1992. Death without weeping: The violence of everyday life in Brazil. Berkeley: University of California Press. Scheper-Hughes, N. 1995. The primacy of the ethical: Propositions for a militant anthropology. Current Anthropology 36: 409–20. Shore, C., and S. Wright, eds. 1997. Anthropology of policy: Critical perspectives on governance and power. London: Routledge. Singer, M. 2000. Why I am not a public anthropologist. Anthropology Newsletter 11: 12–13. Spencer, J. 2010. The perils of engagement: A space for anthropology in the age of security? Current Anthropology 51: S289–S299. Tierney, P. 2000. Darkness in El Dorado: How scientists and journalists devastated the Amazon. New York: Norton. Tsing, A. 2005. Friction: An ethnography of global connection. Princeton, NJ: Princeton University Press. van Maanen, J. 2011. Tales of the field: On writing ethnography, 2nd ed. Chicago: University of Chicago Press. Venkateswar, S. 2006. Manifesto for a public anthropologist: Insights from fieldwork. India Review 5: 268–93. Wright, S. 2006. Machetes into a jungle? A history of anthropology in theory and practice, 1981–2000. In Applications of anthropology: Professional anthropology in the twenty-first century, ed. Sarah Pink, 27–54. Oxford: Berghahn.

14_107-Bernard.indb 733

6/9/14 10:03 AM

14_107-Bernard.indb 734

6/9/14 10:03 AM

Author Index

Abbott, M., 675 Abdullah, M., 450, 456 Abedi, M., 40 Abelson, R. P., 539 Aberle, D. F., 529 Abrahamson, A. A., 376 Abu El-Haj, N., 57, 70 Abu-Lughod, L., 172, 394–95, 405, 699 Achinstein, P., 23 Adair, J. W, 444 Adams, T. L., 465 Addams, J., 7 Adkins, D. L., 8 Adler, P., 258, 262–63 Adler, P. A., 258, 262–63 Adler-Lomnitz, L., 634 Agar, M., 4–5, 97, 101, 126, 251–52, 271, 275, 466, 482, 541 Agency for Healthcare Research and Quality, 190 Ageyev, V. S., 670 Agha, A., 396, 400 Agrawal, B. C., 186 Aguilar, C., 634, 647 Ahearn, L., 395 Ahern, J., 316 Ahmed, R., 201 Aitken, Paul, 57 Alarcón, N., 165 Albert, B., 621 Albert, H., 31 al-Bindari, M., 450 al-Hadari, A., 450 Alim, H. S., 404 Alise, M. A., 9 Allen, K., 274 Alter, J. S., 257, 264

Altmann, J., 293 Alvarado, N., 367, 371, 379 Alvard, M., 3 Alvarez Roldán, A., 38 Amadiume, I., 157 American Anthropological Association, 133–34, 136, 278–79, 406 American Anthropological Association Code of Ethics, 132, 135–36, 142–43, 148 American Association of Public Opinion Research, 504 Amin, S., 450 Andersen, M., 174 Anderson, D. R., 525 Anderson, K. T., 440 Anderson-Fye, E., 316, 541 Androutsopoulos, J., 404 Andrzejewski, B. W., 535 Angeloni, E., 700 Anglin, M., 162–63 Angus, K., 663 Ansell, C. K., 638 Anthropology News, 406 APA Task Force on Statistical Inference, 244 Appadurai, A., 57, 262 Arabie, P., 369 Aral, S., 635 Archive of the Indigenous Languages of Latin America, 534 Ardener, E., 59, 63, 71, 399, 407 Arhem, K., 263 Aristotle, 81 Arnason, L., 466 Arnold, A., 421 Aronson, R., 367 Artemisia, 468 Asch, T., 441, 443

735

14_107-Bernard.indb 735

6/9/14 10:03 AM

736

Author Index

Ashley, D., 301, 307 Aswani, S., 618 Atienza, A., 301 Atkinson, P., 254 Atman, C., 242 Atran, S., 367, 541 Augsburger, D., 347 Aunger, R., 201, 272, 284, 466 Austin, D. E., 186 Austin, J. L., 57, 61–62, 395 Avard, M., 618 Avenarius, C., 106, 262–63, 278 Baba, M. L., 666 Babbie, E., 114–15 Bachelard, G., 77 Badiane, L., 672 Baer, H., 661, 663, 672 Baer, H. A., 192 Baer, R. D., 118, 362, 376, 378 Bahloul, J., 77 Bailenson, J. N., 480, 488 Bailey, F. G., 62, 74 Bailey, P. H., 542 Bainbridge, W. S., 468 Bakhtin, M. M., 395–96, 400 Bakszysz-Bymel, M., 187 Baldus, G., 301 Baldwin, J. A., 667, 669, 671, 679 Baleiro, M. C., 377, 380 Balgera, A., 481 Balieiro, M. C., 501, 508, 510, 668 Balikci, A., 444, 455 Ballerino Cohen, C., 165 Banarjee, A. V., 124 Bandura, A., 651 Bang, M., 541 Banks, M. R, 441, 446 Barbosa de Almeida, M. W., 633 Barclay, W. D., 348, 351 Bardoni, B., 153, 172 Barg, F. K., 347 Barker, S. F., 23 Barlow, J., 618 Barnes, J., 632 Barnett, G., 548, 635 Barnett, M. N., 78 Baron, D., 221 Barr, A., 122

14_107-Bernard.indb 736

Barrera, V., 199 Barrett, M., 673, 683 Barrientos, T., 371 Barriera-Viruet, H., 303 Barruti, M., 675 Barry, H. III, 574, 582 Bartlett, F. C., 539 Bartlett, P. F., 667 Bartunek, J. M., 666 Basso, E., 396 Basso, K., 394, 403, 422 Batagelj, V., 633, 644 Batchelder, W. H., 1, 42, 242, 376–78, 380, 466, 508, 523, 528, 540, 668, 674, 678, 683 Bate, J. B., 82 Bates, C., 681–82 Bates, J. E., 298 Bateson, G., 445–46, 452–53, 456 Bateson, P., 293, 297, 299 Bauman, R., 67, 395–402, 410 Bauman, Z., 313 Baym, N. K., 466 Bearison, D. J., 669 Beattie, J., 505 Beauboeuf-Lafontant, T., 541 Beavis, C., 475 Bechkoff, J., 466, 471–72, 474–75, 477 Bechtel, R., 547 Beck, C. T., 541 Beck, S. S., 186–87, 722, 725, 732 Beck, U., 313 Becker, A. B., 186, 190–91 Becker, A. L., 408 Becker, D. M., 506 Becker, H., 261, 541 Becker, H. S., 724 Beckermann, A., 25 Beck-Gernsheim, E., 313 Beebe, J., 198 Behar, R., 104, 164, 174, 257, 699 Behrens, C. A., 294, 302 Bell, D., 156, 161 Belsky, J., 298 Bender, A., 541 Bender, B., 57 Benfer, R. A., 503, 511 Bennett, J. W., 186–87, 498 Bennett, L. A., 8 Benson, P., 668, 676

6/9/14 10:03 AM

Author Index Bentley, M. E., 256 Benveniste, E., 395 Benz, B., 367 Berelson, B., 543 Berg, D. N., 6 Berg, E., 26 Berg, M., 186, 191, 202–3, 664, 669, 671 Berg, M. J., 669 Bergendorff, S., 727 Berger, P. L., 37–38, 42, 669 Berges, I. M., 377 Berkes, F., 634 Berkin, J., 601 Berkman, L. F., 635 Berlin, B. O., 346, 367 Bernard, H. R., 11, 24, 28, 40, 99, 101, 115, 225, 234, 251, 271, 275–76, 302, 316, 352, 358, 360, 367, 369, 409, 466, 479, 509, 534, 537, 541, 543–44, 547–49, 633–34, 644, 646, 650, 663, 668, 672, 675–77, 682–83, 703 Bertaux, D., 241 Besnier, N., 535 Besterman, C., 730 Béteille, A., 633 Betzig, L., 299, 303 Bhandari, M., 479 Bhaskar, R. A., 44 Bianchi, S., 295, 300 Bicker, A., 634 Biehl, J., 313 Bieleman, B., 675 Biella, P., 444 Biernacki, P., 675 Biklen, S. K., 541, 544 Bindon, J. R., 511 Bird, D. W., 296 Birdwhistell, R., 415 Birkenholtz, T., 5 Bishop, J., 447 Bishop, N., 447 Bizarro Ujpán, I., 534 Black, M., 81 Black, R. E., 5 Black, S. P., 422 Blackman, M. B., 445 Blakeley, T. D., 448 Blakely, P. A. R., 442, 444 Blakely, T. D., 442, 444

14_107-Bernard.indb 737

737

Blalock, H. M., Jr., 580, 589 Blase, J. J., 541 Bletzer, K., 538 Bliege-Bird, R., 296 Blier, S. P., 64 Blimes, J., 682 Bloch, M., 63 Blowsnake, S., 6, 13, 534 Bluff, R., 241 Blumberg, E., 613 Boas, F., 26, 100, 132, 401, 445, 533–34 Bochner, S., 281, 283, 285 Bock, J., 295 Bock, P. K., 281 Boehm, C., 68 Boehm, D., 163, 171 Boeije, H., 542 Boellstorff, T., 106, 253, 256, 259, 404, 465– 68, 470–75, 484, 487–89 Boeri, M. W., 542 Bogdan, R., 541, 544 Bohlig, A., 192 Bohman, J., 21 Bohr, N., 272 Boissevain, J., 632–33, 652 Bollen, K. A., 511 Bolles, A. L., 163 Bolstad, P. V., 601 Bond, G. C., 57 Bondarenko, D., 593 Bonvillain, N., 167 Boon, J. A., 61 Boorman, S. A., 369, 632 Booth, C., 7 Booth, R. E., 116–17 Borgatti, S. P., 115, 367, 369, 371, 375, 508, 524, 528, 545, 631, 633, 641, 650–51, 678 Borgerhoff-Mulder, M., 293, 299, 303, 592 Borges, C. D., 508, 510, 668 Borneman, J., 313 Borofsky, R., 70, 186–87, 719, 721, 724, 730 Borrero, M. G., 188, 196, 199 Boster, J., 115, 117–18, 125, 352, 359, 367, 369–70, 377–78, 528, 540, 541 Bostrom, A., 242 Bott, E., 632, 671, 678 Bourdieu, P., 43, 47, 58, 61, 64, 68, 72–73, 77, 82, 393, 649 Bourgois, P., 253, 258, 263, 276, 280, 671

6/9/14 10:03 AM

738

Author Index

Bousfield, W. A., 348, 351 Bowen, A. M., 667, 669, 671, 679 Bowen, J., 405 Bowles, S., 489 Boychuk Duchscher, J., 541 Boyd, R., 489 Boyer, D., 78, 83 Bradburn, N. M., 316, 354 Bradley, C., 575, 577, 579, 582, 584, 586 Brady, E., 405 Brady, I., 58–59, 63 Brand, S., 439, 452 Brandt, E., 408 Brass, D., 650 Braudel, F., 79 Breedlove, D., 346, 367 Breiger, R. L., 632 Breiger, W. R., 367 Brennan, D., 171 Brenneis, D., 394, 398 Brereton, D. P., 44 Brettell, C. B., 698–99 Brewer, D., 350, 376 Brewer, J., 479 Brewis, A. A., 500 Briggs, C., 316, 391, 395–402, 406, 410, 412– 13, 422, 507 Briggs, J., 264 Brim, J. A., 4, 99, 101 Briody, E. K., 662, 665–66, 671, 679 Brislin, R. W., 363 Britt, T., 543 Brodkin [Sacks], K., 158, 163–64, 168, 171, 174 Broesch, J., 634 Brondizio, E. S., 601, 603–5, 610–11, 620–23 Brondo, K., 8 Bronfenbrenner, U., 670 Brown, J. K., 158, 162 Brown, L. M., 540 Brown, M., 405 Brown, P., 662 Brown, P. J., 700 Brown, R., 396, 540 Brown, R. A., 529 Brown, R. G, 396 Browne, K., 547 Bruce, E., 613 Bruce, R., 498

14_107-Bernard.indb 738

Brumfiel, E., 166 Bruner, E. M., 57, 67 Brush, S., 367 Bryant, C., 7 Bryant, C. A., 663 Brydon-Miller, M., 185 Brymer, R. A., 258–59 Bubandt, N., 77 Bucholtz, M., 403–4, 407 Bukov, A., 358 Bullough, V. L., 566 Bunce, A., 241–42 Bunnin, N., 23 Bunzl, M., 730 Bureau of Labor Statistics, 9 Burke, H., 227, 234 Burnham, K. P., 525 Burrell, J., 466, 468–69, 480 Burris, M. A., 169 Burt, R. S., 359, 635, 649–50 Burton, M. L., 369, 371, 568, 574–75, 577, 579, 582, 584, 586, 590 Busungu, A., 230 Butin, D. W., 194 Butler, M. O., 661 Byrne, B., 111 Byrne, M., 241 Byron, E., 371, 379 Caballero, J., 347 Cabezas, A., 171 Cain, D., 347, 379 Cak, A. D., 622 Caldarola, V., 445 Calton, C., 398 Calvet, C., 1, 347, 351 Calvet, L., 634 Calvet-Mir, L., 367 Cameron, D., 404 Cammarota, J., 191, 672 Campanella Bracken, C., 544 Campbell, B. G., 27 Campbell, D. T., 111–15, 524, 526, 571, 580 Campbell, J., 162 Campos Balierio, M., 529 Canales, M., 347 Cancian, F., 29 Candea, M., 75

6/9/14 10:03 AM

Author Index Cannell, C. F., 316 Caplan, P., 161, 164 Caplan, S. E., 484 Capps, L., 538 Cardinale, J., 378 Caren, N., 545 Carey, J. W., 544 Carley, K., 541 Carlisle, B., 622 Carlson, R. G., 367 Carmines, E. G., 579–80, 587 Carmone, F. J., 375 Carnap, R., 23 Carneiro, R. L., 525 Caro, T., 293, 299, 303 Carrasco, J., 616 Carrithers, M., 83–85 Cartwright, D., 188 Casey, M. A., 680 Cashdan, E., 45 Caspari, R., 167 Cassell, J., 279 Casson, R., 539 Castañeda, Q. E., 406 Castells, M., 465 Casterline, D. C., 421 Castillo, C. O., 669 Castronova, E., 468 Caton, S., 405 Caulkins, D. D., 378 Cernea, M. M., 713–14 Chafe, W., 398, 538 Chagnon, N. A., 441, 443, 698 Chalfen, R., 447, 454 Chan, A., 545 Chan, C. S. C., 5 Chan, C. W-H., 1 Chanock, F. O., 454 Chapin, M., 622 Chapman, M., 402 Chappell, H., 538 Charmaz, K., 222, 241, 541 Chatterjee, P., 164 Chavez, L. R., 352, 376, 378, 505, 668, 678 Checker, M., 186 Cheek, J., 222 Chen, C., 480 Chen, M., 472 Chernela, J., 396

14_107-Bernard.indb 739

739

Chick, G. E., 305, 367 Chin, E., 172 Chomsky, N., 539 Christakis, M. D., 635 Chudoba, K. M., 541 Chui, K., 538 Chun, E. W., 404 Churchill, C. J., Jr., 727 Chvatal, V., 525 Ciccone, M., 545 Cipresso, P., 481 Clair, S., 189 Clark, G., 165, 169 Clark, R., 404 Clark, V. L. P., 549 Clarke, M. C., 547 Classen, C., 58, 324 Clatts, M. C., 671 Cleaveland, C., 5 Clegg Smith, K. M., 476, 487 Cliff, N., 366 Clifford, J., 12, 26, 38, 59, 61, 63, 67, 168, 260, 276, 466, 486, 698 Cliggett, L., 603 Cobb, W. J., 316 Cochran, W. G., 573 Cohen, A. P., 75 Cohen, H., 445 Cohen, J., 239, 511, 544 Cohen, J. H., 29 Cohen, P., 511 Cohen, R., 4, 99, 502 Cohen, S., 635 Cohn, C., 167 Colby, B. N., 61, 547 Coleman, E. G., 465–67, 471 Coleman, J. S., 634–35 Coley, J. D., 367 Collado-Ardon, R., 501 Collier, J., 165, 168 Collier, J., Jr., 440, 452, 455 Collier, M., 440 Collings, P., 268–69 Collins, J., 165 Collins, K., 234, 237 Collins, P. Hill, 174 Colombres, A., 450 Colorado State University Institutional Review Board, 487

6/9/14 10:03 AM

740

Author Index

Coman, E., 191, 663–64, 669, 671 Comaroff, J., 41 Comaroff, J. L., 41 Conaway, M. E., 161 Conkey, M., 166 Conklin, H., 346 Conley, J., 422 Connell, P., 163 Connelly, B. L., 631 Consorte, J. G., 111 Constable, N., 256, 258–59, 266 Convey, M., 203, 663, 667, 671 Cook, T. D., 111, 113–15, 524, 580 Cook, W., 190 Coombe, E., 405 Coombs, C. H., 370 Cooper, R., 485 Copeland, T. J., 379 Copeland-Carson, J., 661 Corbin, J., 260, 541–42, 667, 676 Corda, D., 301 Corneliussen, H. G., 468 Cornish-Ward, S. P., 358 Cornwall, A., 190 Costenbader, E., 638 Counts, D., 162 Coupland, N., 391 Cowan, J. K., 63 Coy, M., 72–73, 259 Coyle, S., 667 Coyle, S. L., 671 Coyne, I., 542 Crabtree, B. F., 544 Crain, M. M., 61 Crapanzano, V., 252, 260 Crawford, P. I., 441, 444 Cressy, D. R., 545 Creswell, J., 10, 241, 549, 681 Crick, M., 61 Croneberg, C. G., 421 Cronin, J. M., 5 Cronk, L., 124 Cronqvist, L., 545 Crosbie, P. V., 524–25 Cross, W. J., Jr., 671 Cruz, B. C., 544 Csikszentmihalyi, M., 300, 305 Csordas, T. J., 75, 316, 324 Cui, Y., 466, 471–72, 474–75, 477

14_107-Bernard.indb 740

Curling, C., 444 Currie, T. E., 592 Curtain, J., 534 Curtis, R., 679 Cushing, F. H., 254 Cushman, D., 698 Cuthill, I., 244 Dale, J. A., 664 Dallo, F., 377 Danardono, M., 544 Danching, R., 359 Dancy, J., 23, 26 D’Andrade, R. G., 8, 21, 26, 346, 360, 371, 373–74, 376–77, 539–40, 586, 668, 677 D’Andrea, A., 254 Danehy, J. J., 316 Daniel, J., 245 Daniel, L. G., 244 Daniels, D., 201 Danowski, J., 548 D’Antona, A. O., 622 Daraiseh, N., 303 Darnell, R., 402, 406 Das, V., 79 Davey, S., 545 Davids, I., 185 Davidson, C. M., 5 Davidson, L., 241 Davies, B., 399–400 Davies, C., 44 Davies, J., 252, 260, 277, 281–83, 313 Davies, R., 5 Davies, S. G., 153, 173 Davis, A., 545 Davis, S. M., 192 Davison, L., 194, 202 Dawes, R. M., 380 deBessa, Y., 358 de Brigard, E., 440, 444, 447 DeCaro, E., 301, 307 DeCaro, J., 301, 307, 540 de Casanova, E. M., 727 de Foresta, H., 621 DeFrancisco, D., 547 de Heusch, L., 445–46 de Jorio, A., 71 de Koning, M., 723, 729 Delaney, C., 159

6/9/14 10:03 AM

Author Index Delbos, G., 72 DelVecchio Good, M. J., 313 de Mello Amorozo, M. C., 347, 350 de Munck, V. C., 378 Dengah, H. J. F., 468–69, 479, 506 de Nooy, W., 644 Denton, R. K., 268 Denzin, N. K., 44, 105, 549 deOliveira, E., 230 De Silva, M. A., 669 de Silveira Cybele, C., 538 Desjarlais, R. R., 77, 252, 259–60, 324 DeVellis, R. F., 508 Devereaux, P., 479 Devereux, G., 316 DeVos, G., 662, 672 DeWalt, B. R., 265, 270, 358, 584 DeWalt, K. M., 265, 270 Dey, I., 544 Diamond, J., 730 Diawara, M., 451 Diaz, A., 675 Dibbell, J., 485 Dickinson, W. B., 680 Dickson, H. R. P., 562 Dickson, P. R., 358 Dickson-Gomez, J., 202–3, 663, 667, 671 di Leonardo, M., 169, 174 Dillman, D. A., 497 Dilthey, W., 29 Dingwall, R., 278 DiNuzzo, A., 377 Disch, W. B., 199 Ditton, J., 278 Ditton, P., 156 Divale, W. T., 568, 583–84 Dixon, L., 547 Dixon, R. B., 534 Dohrenwend, B. S., 316 Dolgin, J. D., 61 Doncel, C., 675 Donnell, E. T., 1 Doostdar, A., 469 dos Santos, J. E., 377, 380, 501, 508, 510, 529, 668 Doucet, R. C., 404 Dougherty, J. W. D., 73 Douglas, J. D., 316 Douglas, M., 57, 65, 76–78, 723

14_107-Bernard.indb 741

741

Dow, M. M., 586, 589–90 Draft Code, 138, 140 Drass, K. A., 545 Drauker, C., 542 Dresch, P., 56 Dressler, W. W., 371, 377–78, 380, 499–501, 504, 508, 510–11, 529, 668 Drews, D. R., 296, 298 Driver, H. E., 1, 568 Dryfoos, J. G., 670 Duarte, N., 706 DuBois, C., 281 Du Bois, J., 394, 398 Ducheneaut, N., 473, 480 Dudley, N., 378 Due, C., 5 Duflo, E., 124 Dugdale, P., 421 Dumont, J.-P., 257 Dunbar, R., 306, 634 Dundes, A., 64 Dungy, C. I., 348 Dunn, L. L., 501, 506 Dunphy, D. C., 540, 547 Duran, B., 191 Duranti, A., 394, 398–99, 402–3, 415, 421 Dureau, C., 395 Durkheim, E., 77, 79 Durnin, J., 298 Duval, R. D., 524 Dwyer, K., 699 Ebner-Priemer, U. W., 293, 301, 305 Eco, U., 85 Edgerton, R., 122 Edgerton, R. B., 498 Edgington, E., 116 Edginton, E. S., 524 Edholm, F., 158 Edmonson, M. S., 534 Edwards, E., 445 Edwards, G., 678 Edwards, J. A., 408 Eff, E. A., 586, 589–90 Efrom, B., 524 Eggan, F., 568 Ehlert-Lerche, S., 538 Eisenlohr, P., 534 Ekadinata, A., 621

6/9/14 10:03 AM

742

Author Index

El Guindi, F., 439, 441–44, 446–49, 452–54, 534 Ellen, R., 100, 634 Ellis, J., 392 Ellstrand, A. E., 631 El Saadawi, N., 162 Ember, C. R., 500, 545, 561, 563–68, 570, 575, 577, 579, 582–84, 588–93 Ember, M., 500, 545, 561, 565–68, 570–71, 573, 582–83, 586, 588–93 Emerson, R. M., 4, 275–76, 482 Enloe, C., 152 Ensign, J., 367 Enslin, E., 164 Ensminger, J., 45 Ensworth, P., 8 Entwisle, B., 616 Epstein, A. L., 632 Erbaugh, M. S., 538 Erickson, F., 263, 411 Erickson, P. I., 367, 672 Eriksen, T. H., 39, 720, 723, 730 Erikson, E., 322 Errington, F., 257 Ervin, A. M., 662–63 Eshleman, J., 592 Esquit-Choy, A., 371 Esser, H., 45 Ethnographic Atlas, 572–73, 576 Etienne, M., 1, 158–59, 347, 351 Eva, G., 363 Evans, G. W., 375, 378 Evans, J. M., 633 Evans-Pritchard, E. E., 74, 80 Everett, M. G., 115, 524, 631, 678 Ewick, P., 168 Ewing, K. P., 334 Eyre, S. L., 542 Fabes, R. A., 638 Fabian, J., 69 Fagan, J., 468–69, 479, 506 Fairclough, N., 391, 404 Falck, R. S., 367 Falk, D., 167 Falletta-Cowden, A., 500 Fals Borda, O., 191–92 Faris, J. C., 445 Farmer, A. P., 5

14_107-Bernard.indb 742

Farmer, P., 123, 727 Farnell, B., 58, 66, 86, 393, 396, 399, 402–3, 405, 412, 415, 418–22 Fassin, D., 324 Faubion, J. D., 313, 324 Faust, K., 358, 373, 375, 616, 631, 644, 671, 679 Fausto-Sterling, A., 153, 172 Fedirko Smith, T., 298, 303 Feld, S., 80, 398–99, 403, 413, 635 Fenigsen, J., 392, 394, 404 Ferguson, J., 76, 468–69 Fern, E. F., 352 Fernandez, J. W., 57, 59, 61, 63–64, 67–68, 71, 73–74, 76, 79–81, 85 Fernandez-Kelley, M. P., 163 Ferreira, L., 230 Ferreira-Pinto, J., 348, 367, 371 Festa, P., 56, 78, 86 Fetterman, D., 218, 662 Fink, A., 354, 497 Fink, B., 316 Finlayson, T., 230 Finn, J. T., 611 Finnström, S., 263–64 Firth, R., 252 Fischer, C. S., 635 Fischer, J. L., 580 Fischer, M., 465, 679 Fischer, M. J., 251 Fischer, M. M. J., 40, 451, 683, 698 Fischoff, B., 242 Fisher, R. A., 524 Fisher, T. D., 305 Fishman, J., 395, 397 Fiske, D., 24, 27 Fiske, M., 316 Fiske, S. J., 8 Fitzgerald, R., 363 Flannery, K., 166 Flinn, M., 296, 306 Flis, A., 25 Flocke, S., 298, 303 Flores, C., 194, 202 Florio, S., 682 Fluehr-Lobban, C., 9, 134–37, 142, 148, 186, 273, 406 Folbre, N., 168 Foley, D. E., 187

6/9/14 10:03 AM

Author Index Folke, C., 634 Fontana, A., 316 Fontanella, B. J. B., 542 Ford, C. S., 565, 585 Forman, V., 635 Forte, M. C., 729 Fossey, E., 241 Foucault, M., 37, 391, 422, 730 Fouts, H. N., 296 Fowler, F. J., 353–54, 356, 359 Fowler, J. H., 635 Fox, A., 398, 403 Fox, J., 615 Frake, C. O., 346 Francis, M. E., 546 Franklin, S., 167, 170 Franzen, L., 613 Frazer, J. G., 80 Frazier, L. J., 165 Freeman, C., 165, 171, 469 Freeman, D., 38, 101, 273 Freeman, H. E., 348, 358, 367, 376 Freeman, L., 678 Freeman, L. C., 359, 367, 524, 631–32, 642, 683 Freeman, S. C., 359, 367 Freire, P., 191, 664, 671 Freisthler, B., 613 French, B., 398 Fretz, R. I., 4, 275–76 Frey, B. S., 45 Frey, J. H., 316 Friedl, E., 156, 158 Friedman, S. R., 679 Friedrich, P., 63, 80 Friendly, M. L., 348, 351, 376 Frisby, W., 192 Frisch, O. R., 272 Fuchs, M., 26 Fuentes, C., 529, 531 Fuhrman, J., 545 Fung, H., 404 Furbee, L., 538 Furbee-Losee, L., 534 Furnham, A., 281, 283, 285 Gable, E., 57 Gable, E. H, 57 Gächter, S., 122

14_107-Bernard.indb 743

743

Gacs, U., 155 Gadamer, H.-G., 30–31 Gaggioli, A., 301 Gagnon, J. H., 669–70 Gailey, C. Ward, 158–59, 170 Gal, S., 171–72, 398, 400, 403–4 Galaskiewicz, J., 671 Galbraith, J., 367 Gallagher, K., 230 Gallaher, P., 631 Gannotti, M., 509 Gannotti, M. E., 525 Garcia, A. C., 466, 471–72, 474–75, 477 Garcia, J. G., 118 García-Álvarez, E., 540, 548 Garcia de Alba, J. E., 376, 378 Garcia de Alba Garcia, J. E., 378 García-Quijano, C. G., 378 Gardin, J.-C., 71 Garfinkel, H., 395, 397 Garrett, A. M., 316 Garrison, L., 313 Garro, L., 668 Garro, L. A., 501 Garro, L. C., 374, 378 Garro, L. Y., 120–21, 371, 376 Gartin, M., 548 Garvey, C., 404 Gatenby, B., 192 Gaudio, R.,P., 403, 405–6 Gaviria, M., 201 Gearing, F., 187 Gee, J. P., 391, 422 Geer, B., 541 Geer, J. H., 669 Geertz, C., 31–34, 58, 71, 79–80, 267, 269, 466, 484, 724 Geier, M., 23 Geilhufe, N., 186 Gelatt, R., 405 Geldsetzer, L., 29 Gelfand, C. C., 122 Genser, S. G., 671 George-Cramer, M., 592 Gerber, E. R., 506 Germano, W., 696 Gero, J., 71, 166 Gershon, I., 392, 404, 470, 476, 478 Gerson, K., 300

6/9/14 10:03 AM

744

Author Index

Geurts, K. L., 324 Gewertz, D., 257 Gibbs, R. W., 60 Giblin, J., 545 Giddens, A., 46, 68, 393 Gile, K., 231 Gillett, G., 393–95 Gilliam, A., 57 Gilligan, C., 540 Gillman, A., 396 Gindis, B., 670 Gingrich, A., 29 Ginsburg, F., 167, 170, 172, 405, 454 Ginwright, S., 672 Girke, F., 83 Giroux, H. A., 664, 671 Gittelsohn, J., 367 Givens, D., 8–9 Gladwin, C. H., 668 Glaser, B. G., 222, 260, 482, 541, 667 Gleason, G., 198 Gleason, G. R., 256, 681 Godbey, G., 297, 300 Goddard, I., 407 Goddard, P. E., 534 Godoy, R., 125–26, 501 Goffman, E., 397, 399, 415 Gold, A. G., 40 Golde, P., 155 Golder, T. V., 367 Goldschmidt, W., 99, 572 Goldstein, E., 203 Goldstein, L. J., 59 Golinelli, D., 529 Golub, A., 468, 472 Gombrich, E. H., 71 Gonçalves, B., 634 Gonzalez, M., 633 González, N. N., 186 González Ventura, J., 534 Good, B., 313 Good, B. J., 313 Good, K., 280 Goodchild, F., 601 Goodenough, W., 580 Goodenough, W. H., 525 Goodman, J., 400, 405 Goodman, L. A., 192 Goodwin, C., 336, 397, 399, 403, 405, 415, 421

14_107-Bernard.indb 744

Goodwin, M., 397, 403, 405, 421 Goody, J., 721 Goosby, E., 681–82 Gorden, R. L., 316 Gordon, D., 164, 174 Gordon, R., 663 Gorfain, P., 57 Gottlieb, A., 167 Gottschalk, L., 547 Gough, K., 155 Govone, J. S., 347, 350 Grady, H. M., 203 Graham, L. R., 392–94, 398–400, 403–6, 409–14, 422 Grand, S., 5 Granovetter, M., 634, 650 Grasseni, C., 82 Graves, T., 244, 498 Gravlee, C. C., 11, 125–26, 307, 380, 466, 479, 501, 544, 549 Gray, J. P., 574 Gray, R. D., 592 Green, G., 611 Green, G. M., 621 Green, J., 254 Green, P. E., 375 Green, S. W., 254 Greenacre, M. J., 529 Greene, J. C., 10 Greenhalgh, S., 40 Greenhill, S. J., 592 Greenson, R. R., 316 Greenwood, D., 192, 661 Grewal, I., 164 Griaule, M., 446, 453 Grier, S., 663 Griesbach, D., 5 Griffith, D. C., 352, 378 Grills, S., 251, 260–61 Grimshaw, A. D., 397 Gross, D. D., 187 Gross, D. R., 148 Gruber, J., 538 Gruenbaum, E., 162 Gruzd, A., 633 Guba, E. G., 541 Gubrium, A., 12 Gubrium, J. F., 316 Gudeman, S., 83

6/9/14 10:03 AM

Author Index Guerrón-Montero, C., 666 Guest, G., 227, 234, 241–42, 244–45, 358, 548 Guimaraes, M. J. L., 266 Guioli, S., 153, 172 Gujar, B. R., 40 Gujarati, D. N., 525 Gullestad, M., 725 Gumperz, J. J., 395, 397, 402 Guneskara, A. M., 3 Gunter, B., 479 Gupta, A., 468–69 Gurven, M., 302 Gust, G., 683 Gusterson, H., 730 Guthrie, S. E., 80 Guyatt, G., 479 Guyer, J. I., 603 Habermas, T., 538 Hackney, J. K., 161, 164, 172, 174 Haddon, A. C., 455 Hafez, S., 154, 171 Hage, G., 252, 260, 264, 271, 277 Hage, P., 633 Hager, L., 167 Hagey, R. S., 191 Hahn, R. A., 662 Haldane, H., 162, 174 Hale, C. R., 663 Halgin, D. S., 528 Hall, K., 403 Hallett, T., 633, 646 Hamberger, K., 633 Hames, R., 293, 295–96, 299, 303, 305 Hamilton-Giachritsis, C., 547 Hammel, E. A., 8 Hammersley, M., 254 Hammoudi, A., 313 Hamre, B., 298 Handcock, M., 231 Handelman, D., 57 Handelman, J., 472 Handler, R., 57 Handwerker, P., 223 Handwerker, W. P., 358, 466, 509, 511, 524– 26, 528, 540, 681 Hanks, W., 157, 396, 403, 415 Hanneman, R. A., 631 Hanssen, K., 316

14_107-Bernard.indb 745

745

Harary, F., 633 Haraway, D., 167, 465 Harding, S., 152, 155, 161, 664 Harkavy, I., 192 Harman, R. C., 367 Harper, K., 12 Harré, R., 392–95, 399–400 Harris, M., 63, 111, 697 Harris, O., 158 Harrison, B., 665 Harrison, F., 171 Hart, K., 731 Hartley, J., 540 Hartley, J. A., 115, 352 Harvey, C., 241 Harvey, S. A., 302 Hasegawa, T., 592 Hastrup, K., 43, 260, 277 Haug, M. R., 356 Haugen, E., 408 Haviland, J., 336, 392, 396, 403, 405–6, 409, 415, 421 Hawe, P., 5 Hawkes, K., 305 Hayano, D. M., 40 Hayles, N. K., 465 Hays, R. D., 367 Hays-Gilpin, K. A., 548 Heath, S., 421 Heath, S. B., 672 Hecht, T., 262 Heckathorn, D., 231, 675 Heidegger, M., 30 Heider, K. G., 440–44, 448, 453–55 Heiman, H., 367 Heiman, R., 165 Heinz, A., 367 Heise, L., 221 Hekman, S., 31 Heller, M., 398 Hendry, G., 479 Henley, N. M., 347, 371 Henrich, J., 122, 489 Henry, J., 305 Herdt, G., 316 Heritage, J., 397 Hermann, B., 122 Hermer, C., 448 Hernández, D., 405

6/9/14 10:03 AM

746

Author Index

Hernandez, T., 347 Hernández Jiménez, A., 534 Hertz, A., 634 Herzfeld, M., 55–59, 61–66, 70–71, 75, 78, 81–82, 84–85, 537 Hewlett, B. L., 274 Hewlett, B. S., 274, 295, 305–6 Hiernaux, P., 618 Higham, S., 5 Hill, C. E., 666, 672 Hill, J., 70, 392, 394–96, 398, 404 Hill, J. H., 538, 548 Hill, K., 305 Hill, W., 379 Himmelgreen, D., 200 Hindman, M. S., 465 Hine, C., 256, 259, 466, 472 Hinsley, C., 254 Hintze, W., 240 Hipsley, E. H., 298 Hirsch, J., 171 Hirsch, J. S., 29 Hirschkind, C., 399 Ho, K., 5, 106 Hobson, J., 172 Hockett, C. F., 316 Hockings, P., 440, 448–49 Hodgen, J. J., 379 Hodgen, M. T., 76 Hoff, L. A., 162–63 Hofferth, S., 300 Hoffman, S. M., 444 Hoffman, V., 542 Holden, C., 592 Hollan, D., 313, 315–16, 321, 323, 339 Hollan, D. W., 313, 315–16, 321, 334 Holland, D., 540, 668, 676 Holleristein, T., 638 Holliday, L., 167 Hollis, M., 21 Holmberg, A., 664 Holmes, S., 1, 370, 379 Holmes, T. H., 364 Holquist, M., 398 Holstein, J. A., 316 Holsti, O. R., 543 Holy, L., 256 Homiak, J. P., 449 Honigmann, J., 4

14_107-Bernard.indb 746

Hopkins, A., 631, 634 Horne, K., 601 Horowitz, R., 263 Horst, A. H., 465, 480 Hosmer, D. W., 525 Houseman, M., 633 Houts, R., 298 Howard, J., 3 Howes, D., 58, 324 Hrdy, S. Blaffer, 167 Hruschka, D. J., 122, 124, 545, 549 Hsiao, A. F., 367 Hsieh, J., 480 Huanaca, T., 501 Huang, A. C. H., 481 Hubbell, F. A., 352, 376, 378, 505, 668, 678 Hubbs-Tait, L., 298 Huberman, A. M., 115, 542, 544, 676 Hubert, L. J., 121 Hudson, A. E., 565, 585 Hufford, M., 301–2 Hughes, E. C., 541 Human Relations Area Files (HRAF), 565, 573 Humphrey, C., 405 Humphries, M., 192 Hunt, G., 533 Hunter, A., 479 Huntington, P., 479 Hurlbert, S. H., 107 Hurtado, A. M., 305 Hurtado, E., 256 Hurtig, J., 165 Husmann, R., 449 Hutchins, E., 367, 541 Hutchinson, J. W., 350, 376 Hyatt, S. B., 378 Hyde, L. L., 367 Hyde, S. T., 313 Hyland, S., 189 Hyman, H. H., 316 Hymes, D., 392–93, 395, 402, 406–8, 410, 413, 422, 535–36 Hymes, V., 536 Hyslop-Margison, E. J., 664 Hyvonen, J., 631 Iannucci, A. L., 380 Ice, G., 307

6/9/14 10:03 AM

Author Index Ice, G. H., 364, 509 Inaoka, T., 305 Ingold, T., 727 Inhorn, M. C., 662 Inoue, M., 397, 403 Institute for Community Research, 680 Iovita, A., 633, 646 Irizarry, R., 671 Ironsmith, M., 112 Irvine, J., 394, 396, 398, 400, 404, 421 Israel, B. A., 186, 190–91 Ivanič, R., 404 Ives, E. D., 405 Ivey Henry, P. K., 296 Izquierdo, C., 540 Jablonko, A., 448, 453 Jablonski, T., 8–9 Jackman, R. W., 511 Jacknis, I., 445–46, 452 Jackson, J., 273, 276 Jackson, M., 58, 73–74, 84, 252, 277 Jackson, M. O., 631 Jacobs, J., 300 Jacobs, S. E., 279 Jacobs, S.-E., 156 Jacobs-Huey, L., 664 Jacobsohn, A., 466, 479 Jacobson, D., 40 Jacobson, L., 586 Jaffe, A., 398, 404 Jaggar, A., 154 Jahan Karim, W., 161 Jakobson, R., 59, 64, 393, 395 Jameson, F., 61 Jameson, K. A., 371, 379 Janelle, D. G., 601 Janes, C. R., 499, 501, 505, 508 Janik, A., 23 Jarrett, R., 169 Jaskyte, K., 378, 500 Jaworski, A., 391 Jefferson, G., 397 Jelinek, C. E., 76 Jenkins, T., 72–73, 82 Jenks, A., 601 Jennings, J., 534 Jensen, J. R., 601–2, 611, 619 Jessri, M., 5

14_107-Bernard.indb 747

747

Jewkes, R., 190 Jhala, J., 451 Jhala, Y. S., 502 Johansen, U. C., 633 Johnsen, E. C., 633, 644, 672 Johnson, A., 293–94, 296, 298–99, 301, 304– 6, 308, 358, 540, 568 Johnson, A. W., 4 Johnson, I., 63, 78 Johnson, J. C., 98, 102, 106–7, 112, 114–18, 125, 217, 219, 223, 262–63, 278, 345, 352, 359, 367, 378, 631, 633, 671, 673–74 Johnson, J. L., 631 Johnson, L., 241–42 Johnson, M., 58, 75, 86, 539 Johnson, O., 306 Johnson, S., 404 Johnson-Laird, P. N., 541 Johnston, F. E., 358 Jones, E. W., 350 Jones, G., 481 Jones, N. A., 540–41 Jones, S., 622 Jordan, A., 137 Jordan, A. T., 665, 671 Jordan, F., 592 Jordan, G., 672 Jorgensen, J. G., 568 Jorion, P., 3, 72 Josephson, L., 582 Jowell, R., 363 Joyce, R., 166 Juan, Z., 354 Jung, Y., 78 Jupp, V., 466 Juris, J. S., 469 Kadono, T., 305 Kadushin, C., 631, 635 Kaeppler, A., 403 Kagan, J., 358 Kahn, R. L., 316 Kahneman, D., 293, 297, 301, 305 Kaldjian, L. C., 350 Kant, I., 36 Kanuha, V., 163 Kapferer, B., 57, 632 Kaplan, C. H., 675 Kaplan, H., 302, 305

6/9/14 10:03 AM

748

Author Index

Karatabsos, G., 380 Karlawish, J., 347 Karp, I., 72 Kasper, C., 631 Kasulis, J. J., 358 Katz, E., 635 Katz, J., 258, 260 Katz, L., 587 Kaufman, D. R., 154 Kaufman, J., 633 Kawulich, B. B., 265 Kay, K., 683 Kay, P. D., 346, 358 Kayembe, P., 230 Kazankov, A., 593 Keane, W., 394, 398 Kearney, M. H., 542 Keating, E., 392, 399, 415 Keiffer, M., 367, 369, 371 Kelle, U., 544 Kelley, N., 613 Kelly, R., 298, 303 Kemnitzer, D. S., 61 Kempton, W., 115, 352, 540 Kendall, M. B., 72 Kendall, P. L., 316 Kendall-Taylor, N., 316 Kendon, A., 56, 71, 336, 399, 403, 405, 415, 453 Kennedy, D. P., 125–26, 501, 509, 529 Kennedy, G., 421 Kennedy, R., 565, 585 Kenny, K., 5 Kenyatta, J., 40 Kertzer, D. I., 74 Khaltourina, D., 593 Khan, A., 155 Khan, M. E., 680 Khiabany, G., 469 Kidder, T., 124 Killworth, P. D., 302, 360, 633–34, 644, 646– 47, 650, 672, 682–83 Kim, S., 631 Kim, T.-G., 1 Kimiaie, M., 545 Kincheloe, J. L., 105 King, S. A., 487 Kirk, J., 367–68, 682 Kirk, L., 371

14_107-Bernard.indb 748

Kirk, N. E., 298 Kirkpatrick, J., 394 Kiš, A. D., 379 Kish, L., 573, 589 Kitihara, M., 592 Klapp, O. E., 75 Klein, D., 316 Klein, R. E., 356–58, 362, 367, 369, 371 Kleinbaum, D. G., 110–11, 114 Kleinman, A., 55, 313, 321, 668, 676 Kleinman, A. M., 77, 79 Kleinman, J., 77, 79, 321 Kligman, G., 171 Klima, A., 84 Klovdahl, A. S., 635, 679 Kluckhohn, C., 256, 316 Kluckhohn, F. R., 256 Knauft, B. M., 698 Knight, J., 45 Knoke, D., 679 Knowlton, A., 635 Koester, S., 116–17, 671 Kollock, P., 469 Koloss, H.-J., 449 Kondo, D. K., 72 Konstan, J., 487 Kornblum, W., 269 Korotayev, A., 593 Kortendick, O., 543 Kosminsky, E., 480 Koss, M. P., 538 Koster, J., 307, 379 Kostick, K. M., 669, 671, 676 Kottak, C. P., 697, 699, 708, 710, 712–13 Kovats-Bernat, C. J., 406 Kozaitis, K. A., 708 Kozinets, R. V., 466–67, 471–74, 476, 482, 485–87, 490 Kozulin, A., 670 Kracke, W., 316 Krackhardt, D., 359, 647, 649 Kraemer, H. C., 591 Kraidy, M. M., 262 Kramer, K., 294–95 Krathwohl, D. R., 704 Krige, E. Jenson, 157 Krige, J. D., 157 Kring, A., 538 Krippendorf, K., 543–44

6/9/14 10:03 AM

Author Index Kroeber, A. L., 1, 4–5, 26, 68, 534 Kroeber, K., 1 Kroenfeld, D., 302 Kröger, M., 545 Kronenfeld, D., 61, 360, 633, 646 Kroskrity, P., 396, 397 Krueger, A., 293, 297, 301, 305 Krueger, R. A., 680 Kruskal, J. B., 372, 375 Krygier, J. B., 609 Kuhn, T. S., 43 Kuipers, J., 400 Kulick, D., 280, 404 Kumoll, K., 39 Kupper, L. L., 110–11, 114 Kusters, K., 621 Kuzel, A., 241 Kuznar, L. A., 21 Kwong, K. K-F, 1 Labianca, L. G., 650 Labov, W., 395, 406 Labovitz, S., 591 Lacy, M. G., 468–69, 479, 502, 506 Lacy, S., 662, 700 Laderman, C., 421 Ladson-Billings, G., 672 Lagacé, R. O., 573 Lai, H.-L., 538 Lakoff, G., 58, 75, 80, 539 Lamb, M. E., 295–96, 305–6 Lamb, Z., 622 Lambek, M., 313, 324 Lambin, E. F., 603 Lampert, M. D., 408 Lampert, T., 358 Lamphere, L., 153, 155–56, 158, 164, 168, 666 Lancaster, J., 167 Landauer, T. K., 242 Landzelius, K., 465, 469 Lang, H., 21, 42, 545 Lang, J., 111 Langa, D., 298, 303 Langlois, J. H., 298 Langois-Winkle, F., 378–79 Lant, T., 548 Lantz, P. L., 191 La Paro, K., 298 Larder, N., 613

14_107-Bernard.indb 749

749

Larkin, B., 172 Larson, K., 548 Larson, R., 300, 305 Lassiter, L. E., 485, 720, 726 Laston, S. L., 367 Latkin, C., 635, 679 Latour, B., 37 Lauer, E., 582 Laumann, E., 634 Lave, J., 55, 59, 72–73, 81, 84, 472 Lawrence, D., 77 Lawrence-Zúñiga, D., 77 Lawson, A., 57 Leach, E. R., 67, 256 Leach, J. S., 64 Leacock, E. B., 155–59 Leavitt, J., 62, 75, 84 LeBaron, C., 336, 399, 405 LeBesco, K., 472 LeCompte, M. D., 4, 185, 195, 200, 202, 664, 672, 676, 682 Lederman, R., 273, 398, 403 Lee, D., 1 Lee, G. C-T, 1 Lee, R. B., 698 Leece, P., 479 Leech, N. L., 680 Leibowitz, L., 167 Leitch, A., 540–41 LeMaster, B., 415 Lemeshow, S., 525 Leonard, W. R., 125–26, 501 Leone, M. P., 71 Leontsini, E., 302 Leroi-Gourhan, A., 453 Lester, R. J., 316, 663 Le Tourneau, F.-M., 621 Levin, M., 661 LeVine, R. A., 68, 99, 101–2, 313, 316 Levinson, D., 565, 584–85 Lévi-Strauss, C., 57, 74 Levy, J., 136 Levy, R. I., 315–16, 321, 323, 325–26, 334–35, 339 Lewin, E., 163, 169 Lewin, K., 188 Lewins, A., 546 Lewis, C. E., 365 Lewis, K., 633

6/9/14 10:03 AM

750

Author Index

Lewis, M. A., 365 Lewis, M. P., 535 Lewis, N., 167 Lewis, O., 28 Li, F., 188 Li, J., 189, 202–3, 671 Liang, A. C., 403 Liao, S., 188 Lichtenstein, R., 191 Lieberman, D., 371 Lima, R. M. J., 204 Limón, J. E., 61 Lin, N., 649 Lincoln, Y. S., 105, 541 Lindeman, E. C., 256, 264 Lindenberg, S., 45 Lindesmith, A. R., 545 Lippi-Green, R., 398 List, G., 403 Litwin, A., 192 Llewelyn-Davies, M., 444 Lo, A., 404 Lobban, R., 273 Lock, M., 79 Lockhead, G. R., 376 Lofland, J., 541 Lofland, L. H., 541 Loftus, E., 359 Lohman, J. D., 256 Loizos, P., 440–41, 444–45, 454–55 Lombard, M., 544 Longacre, W., 570 Lopez, A., 367 López-Sintas, J., 540, 548 Lorde, A., 158, 160 Lorrain, F., 632 Louis, M. R., 666 Low, S., 186–87, 422, 720, 727 Low, S. M., 55, 77, 358 Lowie, R. H., 26, 100, 534 Lu, D. W., 481 Lubbers, M., 648 Luborsky, M. R., 673 Luchesi, B. M., 542 Luckman, T., 669 Luckmann, T., 37–38, 42 Lucy, J., 398, 400 Luey, B., 696 Lugosi, P., 278

14_107-Bernard.indb 750

Luhrmann, T., 260, 271, 277, 489 Lurie, N., 155, 534 Lusch, R. F., 358 Lutz, C., 168, 367, 394–95 Lutz, K., 273–74, 279 Lynam, T., 1, 347, 351, 540–41 Lynch, E. B., 1, 370, 379 Lynd, R., 4 Lyson, T. A., 500 Maas, I., 358 Mabry, E., 487 Macauda, M. M., 367 MacClancy, J., 722 MacCormack, C., 159 MacDougall, D., 440–41, 445–46, 454 Mace, R., 592 Mack, J., 65 MacKellar, D., 230 MacMahon, M. K. C., 407 MacQueen, K., 548, 683 MacVean, R. B., 358 Madge, C., 479 Madsen, R., 545 Magaña, J. R., 371, 375, 378 Maginn, P. J., 256, 265 Magnani, R., 231 Magnet, S., 467, 471, 476 Maguire, P., 192 Maher, M. M., 609 Maida, C. A., 187, 722, 725, 732 Makihara, M., 397 Malaby, T., 74, 78–79, 81, 83–84, 467–70, 484 Malinowski, B., 2, 25–27, 38, 252–53, 268, 273, 275, 466, 498, 698 Malkki, L., 78, 164 Malta, M., 204 Mandelbaum, D. G., 713 Mandell, W., 679 Manderson, L., 542 Mankekar, P., 172 Mannheim, B., 394, 400 Mantini-Briggs, C., 391 Manuel, P. L., 405 Mapatano, M., 230 Maranhão, T., 727 Marburger, W., 359 Marcelin, L., 191 Marchand, T. H. J., 63, 82

6/9/14 10:03 AM

Author Index Marcus, B., 441 Marcus, G. E., 5, 12, 38, 61, 63, 105, 168, 251, 256–57, 267, 278, 466, 486, 683, 698 Marcus, J., 166 Margolis, M., 544 Markham, A. N., 466, 472 Markovic, M., 541–42 Marks, D., 450 Marlowe, F., 299 Maroney, E. M., 538 Marriott, A., 306 Marrow, C., 671, 679 Marsden, P. V., 678 Marshall, C., 582 Marshall, J., 441 Martey, R., 487 Martin, C. L., 638 Martin, D., 263 Martin, E., 37, 167 Martin, M. M., 583, 593 Martin, P., 293, 297, 299 Martinez, M., 203 Martsolf, D. S., 542 Maruna, S., 547 Marx, K., 60 Mary-Rousseliere, G., 444 Mascia-Lees, F., 164–65, 171, 174 Massey, W. C., 568 Mathevet, R., 1, 347, 351 Mathews, H. F., 537 Mathews, M. C., 649 Mathez-Stiefel, S., 347 Mattingly, C., 324, 668 Maupin, J., 370, 374, 378–79 Mauri, M., 481 Mausel, P., 621 Mauss, M., 77 Maxwell, C. R., 466, 479 Maya-Jariego, I., 642 Mayan, M., 673, 683 Mayne, T. J., 546 Maznevski, M. L., 541 Mazzarella, W., 82 McCabe, C., 305 McCabe, J. T., 29 McCall, G. G., 251 McCarthy, M. B., 5 McCarty, C., 359, 634, 642, 647–48, 650 McCaughan, J. A., 367

14_107-Bernard.indb 751

751

McClaurin, I., 165, 171, 174 McCleary, R., 545 McClellan, E., 683 McClendon, S., 413 McCracken, D., 623 McCracken, G. D., 316 McCurdy, D. W., 4, 700 McDade, T. W., 380 McDaniel, C. K., 8 McDermott, F., 241 McDermott, L., 663 McDonald, L., 7 McElhinny, B. S., 394 McElreath, R., 122 McEwan, P. J. M., 524 McGrady, F. A., 671, 679 McGranahan, C., 719 McGregor, W. B., 538 Mcintyre, J., 155 McLaren, P. L., 105 McLaughlin, M. W., 672 McLean, D., 266 McLellan, E., 683 McMullin, J. M., 352, 376, 378, 505, 668, 678 McNabb, S., 420 McNeill, D., 399 McQuown, N. A. B, 452 Mead, M., 38, 255–56, 261–62, 273, 276, 283, 445–46, 452–53, 455 Meadows, M. S., 474, 483 Medeiros, A., 83 Medin, D., 351, 489, 541 Medina Mora, M. E., 672–73 Meek, B. A., 398, 404 Meeker, M. E., 56 Meerwarth, T. L., 665, 671, 679 Meggitt, M. J., 562 Mehra, A., 650 Meintjes, L., 403 Melo, D. G., 542 Mendoza-Denton, N., 403 Menzel, H., 635 Merbecke, J., 547 Mercer, N., 547 Merleau-Ponty, M., 75 Merlo, G., 675 Merry, S., 55, 186–87, 422, 720, 727 Merton, R. K., 316 Mertz, E., 392

6/9/14 10:03 AM

752

Author Index

Metzger, D. S., 683 Meyer, C., 83 Meztger, D., 346 Mezzich, J. E., 375 Miceli, S., 57 Michael, R., 447 Michaels, E., 405 Michaelson, A. G., 359, 367 Miles, M., 115, 542, 544, 676 Milgram, S., 631, 650 Milkie, M., 295, 300 Mill, J. S., 7, 545 Miller, D., 465–66, 468–69, 477, 480 Miller, E. M., 378 Miller, F., 405 Miller, G. A., 367 Miller, K. W., 506 Miller, M. L., 367–68, 682 Miller, S. M., 670 Miller, W. L., 544 Milligan, G. W., 375 Millroy, W., 202 Mills, D., 721 Millstein, S. G., 542 Milstein, B., 683 Milton, M., 303 Minge-Klevana, W., 295 Minh-Ha, T., 172 Minkler, M., 191 Mintz, J. R., 63–64 Mintz, S. W., 44 Miranda, D. M., 347, 350 Miranda, T. M., 347, 350 Mirus, G., 392 Mishra, S. I., 505 Mishra, V., 615 Mitchell, C. J., 632 Mitchell, J., 256 Mitchell, M., 242, 244–45 Mithun, M., 407 Mlobeli, R., 347, 379 Modell, J., 170 Moeran, B., 391 Moerman, M., 397 Mok, D., 616 Mole, J., 69 Molina, J. L., 634, 642, 647–48, 651 Monaghan, L., 398, 415, 421 Monteiro, S., 204

14_107-Bernard.indb 752

Montgomery, E., 298 Montgomery, P., 542 Montoya, R., 165 Mooney, C. Z., 524 Moore, C., 42, 371, 376, 379, 574, 590 Moore, J. H., 592 Moore, R., 367 Moore, R. J., 480 Moore, S. F., 79 Moore, Z. T., 305 Moran, E. F., 100, 603, 621, 623 Morelli, G. A., 296 Moreno, J. L., 631 Morgan, D., 541 Morgan, D. L., 352 Morgan, L., 170 Morgan, L. H., 4 Morgan, M., 242, 404, 544 Morgen, S., 159, 168, 171, 672 Morgenstern, H., 110–11, 114 Mori, J., 397 Moritz, M., 545 Morris, B., 41 Morris, M., 378–79 Morrison, E., 8 Morrison, F., 298 Morse, J., 241, 267, 673, 683 Mosack, K. E., 675 Mosavel, M., 201 Mosco, V., 465 Mose Brown, T., 727 Moss, J., 479 Mosse, D., 47 Most, D., 468–69 Most, D. E., 506 Mouttapa, M., 631 Mrvar, A., 633, 644 Muchnika, L., 635 Mühlhäusler, P., 393–94 Muller, K., 255 Mullings, L., 159–60, 163, 168–69, 266 Muna, L., 367 Munoz-Laboy, M., 171 Munroe, R. H., 569, 571–72 Munroe, R. L., 569, 571–72 Murdock, G. P., 26, 44, 565, 568, 572–76, 585–86, 590 Murphy, E., 278 Murphy, J., 612

6/9/14 10:03 AM

Author Index Murphy, R. F., 274 Murphy, S., 542 Murphy, W., 396 Murphy, Y., 274 Murray, C. R., 406 Murray, J. D., 112, 116 Murray, T. E., 406 Murthy, D., 466, 479, 485 Myers, C. S., 3 Myers, G., 671, 679 Myers-Scotton, C., 404 Myhre, K. C., 41 Nachman, S. R., 682 Nadel, S. F., 562, 632 Nader, L., 269, 273 Nagel, T., 80 Nagengast, C., 170 Nakagawa, S., 244 Nakamura, K., 415 Namey, E., 242, 244–45 Nanda, S., 172–73 Narayan, K., 161, 698, 724 Nardi, B., 466–67, 468–69, 471–73, 475, 476, 484, 487–89 Nardi, P. M., 497 Naroll, R., 4, 28, 43, 99, 566, 572–74, 580, 583–84, 589 Nash, J., 157, 159, 163, 280 Nastasi, B. K., 204, 661, 669, 671 National Academy of Sciences, 622 Nattrass, S. H., 367 Navarro, E., 363, 509 Neaigus, A., 679 Needell, B., 613 Needham, R., 55, 77 Needle, R. H., 671, 681–82 Negrón, R., 402 Negroponte, N., 465 Neidig, J. L., 683 Neofotistos, V. P., 78, 83 Nerad, M., 8 Nerlove, S. B., 371, 373–74, 376, 668, 677 Ness, R. C., 584 Netting, R., 187 Netting, R. M., 45 Netzley, S. B., 543 Ng, N., 544 Nicholas, D., 479

14_107-Bernard.indb 753

753

Nichter, M., 544 Nickell, E., 473, 480 Nielsen, J., 242 Nightingale, F., 7 Nino-Murcia, M., 404 Nisbet, R. A., 76, 79 Nishii, L., 122 Nolan, J. M., 349, 367 Noreen, E. W., 116 Norenzayan, A., 124 Norlth, M., 296, 298 Notes, 25 Nuckolls, J., 399 Nunnally, J. C., 360–61, 364, 587–88 Nyamongo, I. K., 371 Nyerges, E. A., 621 Oakley, A., 507 O’Barr, J., 422 Oberg, K., 281, 283 Oberschall, A., 7 Obeyesekere, G., 58, 316 Oboler, R. Smith, 157 O’Brien, F. J., 534 Ochs, E., 400, 404, 407, 538 O’Connor, H., 479 O’Connor, K., 313 Ogburn, W. F., 7 Ogden, L. A., 549 Ogilvie, D. M., 540, 547 Ohtsuka, R., 305 Okeyo, A. P., 156 O’Leary, T., 583, 593 Olivier, D. C., 369 Olortegui, M. P., 302 Olson, K., 5 Omori, Y., 449 Omvedt, G., 156 O’Nell, C. W., 501 Onghena, P., 116 Ongwuegbuzie, A. J., 680 Onnela, J. P., 631 Onwuegbuzie, A. J., 234, 237, 244 Opsal, T., 168 Orans, M., 100–101 Oravecz, Z., 380 Orbach, M. K., 359 Orero, P., 538 Orr, D. P., 349

6/9/14 10:03 AM

754

Author Index

Ortner, S., 46, 158–59, 313 Ortolani, A., 592 Osgood, C., 79, 547 Oths, K. S., 380, 500–501, 506 Otterbein, K. F., 571, 589–90 Outhwaite, W., 23 Ovretveit, J., 662 Owens, D. C., 191 Oxman, T. E., 547 Oxtoby, M. J., 544 Oziemkowska, M., 679 Özkök, B., 64 Pacansky-Brock, M., 707 Paccagnella, L., 487 Pachter, L. M., 378, 678 Padgett, J. F., 638 Padilla, M., 171 Padmawati, S., 544 Page, B., 191 Page, H., Jr., 583, 593 Page, J., 418, 421 Paine, R., 62, 74 Palinkas, L., 125, 359 Palmer, N. S., 501, 506 Palmquist, M., 541 Pang, B. L, 480 Panofsky, A., 545 Panourgia, N., 104 Panter-Brick, C., 500 Paolisso, M., 293, 299 Papineau, D., 23–24 Parish, S. M., 316, 324 Park, J., 277 Parker, E. A., 186, 190–91 Parker, R. G., 670 Parkes, P., 634 Parkin, D., 61 Parmentier, R., 70, 85, 392 Parr, M. G., 367 Parry, L., 618 Passmore, G., 298 Pasternak, B. E E, 577 Patterson, T., 166 Pattison, P. E., 644 Patton, M. Q., 234, 236, 543 Paulmé, D., 156 Payne, S. L., 316 Peacock, J. L., 61

14_107-Bernard.indb 754

Pearce, C., 466–68, 472–73, 475, 487–89 Peattie, L. R., 187 Pels, P., 721 Pelto, G. H., 4, 99, 114, 256, 272, 275, 497, 667, 675, 682, 703 Pelto, P. J., 4, 99, 114, 272, 275, 367, 497, 661, 667, 675, 682, 703 Peña, D. G., 37 Penka, S., 367 Pennebaker, J. W., 546 Perales, H., 367 Peregrine, P., 296, 298, 566, 570, 573 Pereira, C., 613 Peres, C., 618 Perez, P., 541 Pérez-Escamilla, R., 200 Perley, B., 398 Pero, D., 662 Perra, N., 634 Perusek, D., 264 Pesmen, D., 83 Petermann, S., 634 Peters, E., 56 Peters-Golden, H., 698 Peterson, D. J., 421 Peterson, L. C., 466 Pettit, G. S., 298 Philips, S., 403 Phillips, S. A., 5 Piaget, J., 539 Pianta, R., 298 Piault, C., 446 Pica, T., 167 Picchi, D. S., 258, 270 Picciano, J., 8 Pina-Cabral, J. d., 47 Pink, S., 55, 72, 81, 405, 421, 447, 452, 454, 721 Pinto, S., 313 Pioggia, G., 301 Pirsig, R. M., 73 Pittenger, M.-J., 305 Pittinger, R. E., 316 Plano-Clark, V. L., 10 Plattner, S., 148 Pocius, G. L., 64 Podolefsky, A., 662, 700 Polkinghorne, D. E., 673–74 Pollnac, R., 98

6/9/14 10:03 AM

Author Index Popper, K., 24, 44 Porcello, T., 403 Porter, S., 44 Porter, T. M., 98 Portes, A., 649 Postel-Coster, E., 535 Potter, J. M., 667–68 Potterat, J. J., 679 Prabandari, Y., 544 Pratt, M. L., 395 Preissle, J., 202 Presser, S., 300, 360 Pressman, J., 396 Preziosi, D., 64 Price, L., 540 Price, L. J., 667, 669, 671, 679 Price, R., 38–40 Protection of Human Subjects, Code of Federal Regulations, 142 Provost, C., 586 Pryor, F. L., 582, 585 Pumpuang, P., 650 Punch, M., 278–79 Puri, R. K., 347 Putnam, R., 300 Putnam, R. D., 649 Pyrczak, F., 498 Quételet, A., 7 Quine, W. V., 79 Quine, W. V. O., 24 Quinn, M., 542 Quinn, N., 156, 316, 334, 352, 373–74, 376, 392, 399, 402, 537, 539, 668, 676, 677 Rabinow, P., 55, 268, 283 Radcliffe-Brown, A. R., 26, 720 Radda, K., 191, 199, 663, 671 Radin, P., 6, 401, 534 Rafaeli, S., 487 Ragin, C. C., 545 Ragoné, H., 153, 164, 170 Rahe, R. H., 364 Ramos, A. R., 104 Randolph, J., 611 Rao, A., 29 Rapp, R. [Reiter], 155–56, 158, 167, 170 Rappaport, J., 70, 666 Rappaport, R. A., 57

14_107-Bernard.indb 755

755

Rapport, N., 74, 77 Rashid, S. F., 263, 265 Raven, P., 346, 367 Raver, J. L., 122 Raymond, H., 230 Reber, E., 413 Redding, T., 8 Redfield, R., 7, 28 Reed-Danahay, D., 40 Reed-Danahay, D. E., 72, 76 Regnault, F.-L., 455 Reid, C., 192 Reid, R., 192 Reinharz, S., 153 Reiter, R., 156. See also Rapp, R. [Reiter] Reitz, K., 590 Rettberg, J. W., 468 Reyes, A., 404 Reyes, C., 191, 199 Reyes-García, V., 367, 371, 379–80, 501, 634 Reyna, S., 42 Rhydwen, M., 404 Ribeiro, R. P., 380, 501 Ricas, J., 542 Rice, G. E., 539–40 Rice, R. E., 635 Richard, S. A., 525 Richards, A., 295, 498 Richards, L., 544 Richards, T., 544 Richards, W., Jr., 635 Richardson, J., 5, 68 Richardson, L., 697 Richardson, S. A., 316 Richie, B., 163 Riddle, M., 631 Ridgeway, C. L., 539 Riemer, J. W., 262 Rieser-Daner, L. A., 298 Riggs, D., 5 Rihoux, B., 545 Riley, C., 169 Rindfuss, R., 615–16 Riva, G., 481 Rivers, W. H. R., 3, 26 Robbins, J., 313, 324 Roberts, C., 363 Roberts, J. M., 367 Roberts, M., 57, 70

6/9/14 10:03 AM

756

Author Index

Robertson, J., 664 Robins, G., 644 Robinson, D., 671 Robinson, E., 221 Robinson, J., 295, 297, 300 Robinson, W. S., 545 Robison, J., 191, 199 Robson, C., 111, 116 Rocha, A. L. S., 118 Rocha, J., 379 Rodriguez-Soto, I., 500 Rogers, C., 316 Rogers, E. M., 635, 650 Roggman, L., 298 Rogoff, B., 669 Rohde, D., 133 Rohilla Shalizi, C., 635 Rohner, E. C., 588 Rohner, R. P., 534, 584, 587–88 Rohrlich Leavitt, R., 160 Rojas-Drummond, S., 547 Rollwagen, J. R., 440, 455 Romanucci-Ross, L., 662, 672 Romero-Daza, N., 200 Romney, A. K., 1, 42, 118, 242, 348–49, 354, 358–59, 367, 369–71, 373–80, 466, 508–9, 523, 528–29, 540, 574, 588, 590, 668, 674, 676–78, 683 Rorty, R., 395 Rosaldo, M., 155–56, 165, 168, 394–95, 398 Rosaldo, R., 393 Roseman, M., 412–13 Rosen, L., 55 Rosenbaum, M., 542 Rosenberg, S. D., 547 Rosenthal, G. E., 350 Rosenthal, R., 586, 588 Rosnow, R. L., 588 Ross, A., 469 Ross, H., 541 Ross, J. L., 367 Ross, K., 172 Ross, M. H., 575, 577, 579, 582–85 Ross, N., 351, 370–71, 374, 378–79, 489 Ross, R., 542 Rota, L., 634, 647 Rothenberg, R. B., 667 Roth-Gordon, J., 404 Rouch, J., 440, 442, 444, 446, 450

14_107-Bernard.indb 756

Rowe, Z., 307 Roy, R., 451 Royal Anthropological Institute of Great Britain and Ireland, 2 Roy Chowdhury, R., 601, 610–11, 622 Rubel, A. J., 501 Rubin, G., 159 Rubin, H. J., 316 Rubin, I. S., 316 Rubinstein, R. L., 673 Ruby, J., 440, 445, 454 Rudd, E., 8 Rudner, R., 499 Ruebush, T. R., II, 356–57, 362 Ruf, E., 621 Rumelhart, D. E., 376, 539 Rusch, C. D., 371, 376, 379 Rusk, T. B., 542 Russett, B., 567 Ryan, G., 541, 543–44, 546, 549 Ryan, G. W., 349, 352, 367, 369, 668 Rylko-Bauer, B., 186 Rynkiewich, M. A., 280 Sabae, H., 450 Sabin, K., 231 Sackett, R., 293–94, 299, 301, 304–5 Sacks, H., 397 Sacks, K. See Brodkin, K. Safa, H., 159 Saggurti, N., 204 Sahay, K. N., 450 Sahlins, M., 35, 58, 79, 393 Said, E. W., 450 Saidel, M. G. B., 542 Saidel, T., 231 Sailer, L., 360, 633, 646, 682–83 Sailer, L. D., 302 Salant, P., 497 Salcedo Rocha, A. L., 376, 378 Salem, S., 303 Salemink, O., 721 Salinas Pedraza, J., 40, 409, 534, 537, 548 Salmond, A., 58 Salomon, F., 404 Salzman, P. C., 56 Sammons, K., 405, 409 Samuels, D., 403 Sanday, P. Reeves, 156, 162–63

6/9/14 10:03 AM

Author Index Sandberg, J., 300 Sandelowski, M., 241, 541, 543, 673, 724 Sandstrom, K., 168 Sanjek, R., 59, 156, 252, 254–56, 264, 275–76 Sanstad, K., 203 Santelices, C. C., 367 Sapir, E., 392, 497, 534 Saramaki, J., 631 Sartre, J.-P., 60 Saussure, F. de, 392 Saville-Troike, M., 402 Sayer, L., 295, 300 Sayles, J. N., 367 Scaglion, R., 299 Scelza, B., 294 Schade, E., 671 Schafer, M. R., 399, 403 Schama, S., 77 Schank, R. C., 539 Schatz, E., 5 Schemitsch, E., 479 Schensul, J. J., 185–89, 191–92, 194–95, 199–200, 202–4, 251, 265, 661–66, 668–69, 671–74, 676, 680, 682 Schensul, S. L., 185–89, 194–96, 198–99, 201, 204, 207, 661, 665, 669, 671, 676 Scheper-Hughes, N., 41, 105, 264, 280, 727 Scherer, J. C., 445 Schieffelin, B., 396–97, 481 Schieffelin, B. B., 397, 404, 406, 411 Schkade, D., 293, 297, 301, 305 Schlegel, A., 156, 574 Schlegoff, E. A., 397 Schmaling, C., 415 Schnegg, M., 42, 547 Schneider, A., 440 Schneider, D., 99 Schneider, D. M., 35, 61 Schneidman, E. S., 547 Schnurr, P. P., 547 Schoepfle, G. M., 4, 115, 316, 676 Schor, J., 300 Schouler-Ocak, M., 367 Schrauf, R. W., 363, 509 Schultz, A., 163 Schulz, A. J., 186, 190–91, 307 Schuman, H., 360 Schunko, C., 347 Schütz, A., 37, 42, 45

14_107-Bernard.indb 757

757

Schwarz, N., 293, 297, 301, 305 Schweik, C., 611 Schweizer, T., 545, 634 Schwimmer, E., 67 Scott, G., 675 Scott, J., 358, 631, 678–79 Scott, J. C., 71, 78 Scrimshaw, N., 198 Scrimshaw, N. S., 256, 681 Scrimshaw, S., 256 Scrimshaw, S. C. M., 681 Searle, J. R., 398 Sebba, M., 404 Sechrest, L., 99 Seeger, A., 399, 405, 410, 413 Segura, D., 164, 171 Segura-Millán, S., 200 Seidel, J., 544 Sekiyama, M., 305 Seligman, B. Z., 275, 562 Seligman, C. G., 562 Seligmann, B. Z. S., 3 Seligmann, C. G., 3 Semaan, S., 231 Senft, T., 472 Senghas, R., 398, 415, 421 Sengir, G. H., 671, 679 Seremetakis, C. N., 77, 84 Sewell, W. H., 46, 59 Sexton, J., 534 Shandy, D., 4 Shannon, J., 406 Shariff, A. F., 124 Sharma, S. K., 502 Sharpe, P., 165 Shattuck, D., 227, 234 Shaw, C. R., 39–40 Shaw, L. L., 4, 275–76 Shaw, S., 189 Shelley, K., 304 Shennan, S., 592 Shensul, S., 347, 379 Sherman, S., 635 Sherzer, J., 391, 394–95, 402, 405, 407, 409, 413, 415–16, 536 Shiffman, S., 301–2 Shimmin, H. S., 572 Shin, E. H., 631 Shiu, A. T.-Y., 1

6/9/14 10:03 AM

758

Author Index

Shore, C., 662, 726 Shostak, M., 257, 698–99 Shryock, A., 56, 70 Shultz, J., 411 Shuman, A., 404 Shuy, R., 406 Siapno, J. Aquino, 154, 171 Sidnell, J., 397 Siegel, J. M., 365 Sillitoe, P., 549 Silver, C., 546 Silver, J. S., 367 Silverblatt, I., 158 Silverstein, M., 85, 393, 395–97, 400, 402, 413 Simmel, G., 635 Simmons, J. L., 251 Simmons, L. W., 565, 585 Simon, C., 201 Simon, G. M., 316 Simon, W., 669 Simonsen, J. K., 441 Singer, M., 185–89, 191–92, 194, 198, 200, 202, 204, 661, 663, 666, 671–75, 680–81, 723 Singer, M. B., 85 Singer, M. C., 367 Singh, K. S., 450–51 Singh, R., 204, 669, 671, 676 Singleton, J., 259 Sipes, R. G., 572, 574 Sirén, A. H., 620 Siupe, A., 296, 298 Slater, D., 466, 468–69, 477 Slim, H., 256 Slocum, S., 167 Smith, C., 613 Smith, C. A. S., 378 Smith, C. D. K, 5 Smith, C. S., 378–79 Smith, E., 295 Smith, K. K., 6 Smith, M. A., 469 Smith, M. E., 566 Smith, M. S., 540, 547 Smith, N. L., 704 Smith, P. K., 299 Smith, S. A., 465 Smith, T., 348, 376 Sneath, P., 375

14_107-Bernard.indb 758

Snijders, T. A. B., 644, 675 Snodgrass, J. G., 468–69, 475, 479, 502, 506 Snyder-Duch, J., 544 Soames, S., 23 Sobeilh, T., 303 Society for Linguistic Anthropology, 404–5 Sokal, R., 375 Solomon, H., 375 Somerset, S., 613 Sorenson, E. R., 449, 453–54 Sosa, E., 23, 26 Soukup, C., 471, 473 Soutar, G. N., 358 Soysal, L., 78 Spaight, T., 485 Spain, D. H., 4, 99, 101 Special Issue: Cross-Cultural and Comparative Research, 561 Spencer, D., 252, 260, 313 Spencer, J., 727, 730 Sperber, D., 67 Spindler, G., 99 Spiro, M., 668 Spittler, G., 44 Spitulnik, D., 400, 404 Spradley, J., 4, 222–23, 251–52, 262, 280, 316, 346, 466, 677, 700 Sprague, S., 479 Sreberny, A., 469 Srinivas, M. N., 633 Stack, C., 154, 156–58, 269–70 Stadhams, D., 452 Stafford, E. F., Jr., 358 Stall, R., 203 Standlee, A. I., 466, 471–72, 474–75, 477 Stange, K., 298, 303 Stanley, J. C., 524 Stanton, B. F., 367 Stark, F., 67 Stead, M., 663 Steedly, M. M., 70 Steele, S., 403 Stefflre, V. J., 369, 374 Steglich, C. E. G., 644 Stegmüller, W., 31, 44–45 Stein, J., 696 Steinkuehler, C. A., 469 Stephanson, M., 367 Stephen, L., 163, 165

6/9/14 10:03 AM

Author Index Sterk, C., 269 Stern, G., 186, 201 Steup, M., 23, 26 Stevens, R., 313 Stewart, K., 77, 479 Stillman, F. A., 506 Stinchcombe, A. L., 109 Stinson, L., 300 Stivers, T., 397 Stocking, G. W., 27 Stocking, G. W., Jr., 252 Stokoe, W., 421 Stolcke, V., 76 Stoler, A., 159 Stoller, P., 58, 84 Stoller, R. J., 316 Stone, A., 293, 297, 301–2, 305 Stone, A. A., 301 Stone, P. J., 540, 547 Stone-Jovicich, S. S., 540 Stout, D. B., 562 Strathern, M., 84, 156, 159, 170 Strauss, A., 222, 260, 482, 541–42, 667, 676 Strauss, C., 540, 547 Strauss, W. L., 256 Streck, B., 727 Strecker, I., 83 Streeck, J., 336, 399, 405 Stringer, E. T., 663 Strogatz, S. H., 650 Struever, N., 65 Studdert-Kennedy, M., 415 Stull, D., 665 Suci, G. J., 79 Sudman, S., 316, 354 Sudo, N., 305 Sudweeks, F., 487 Sugawara, K., 305–6 Sullivan, H. S., 316 Sun, C., 480 Sunakawa, C., 392 Sundararajana, A., 635 Susser, I., 163, 165, 169, 192, 257, 264 Sutton, L. A., 403 Sveningsson, M., 487 Swadesh, M., 534 Swann, B., 407 Swanson, E. C., 593

14_107-Bernard.indb 759

759

Swanton, J. R., 534 Swartz, M. C., 540 Sweet, C., 479 Sweet, E., 544, 549 Swiontkowski, M., 479 Sword, H., 697 Sydlo, S., 191, 671 Sykes, B., 160 Syme, S. L., 635 Synnott, A., 324 Szalai, A., 304 Tagliacozzo, G., 60 Takada, A., 296 Takhteyev, Y., 633 Tankersley, K. B., 379 Tannembaum, P. H., 79 Tannen, D., 167, 397, 400 Tanner, N., 167 Tanner, S., 367 Tanz, C., 403 Tartarisco, G., 301 Tashakkori, A., 10, 345, 498, 534, 549 Tatje, T. A., 586 Taureg, M., 449 Tax, S., 186–87, 663 Taylor, C., 395 Taylor, C. C., 3 Taylor, J. S., 264 Taylor, L., 448 Taylor, S. J., 541 Taylor, T. L., 466–68, 472–73, 475, 487–89 Teddlie, C., 9–10, 345, 498, 534, 549 Tedlock, B., 252, 254, 259, 265 Tedlock, D., 394, 400, 405–9, 413, 536 Teman, E., 273 Terry-Sharp, K., 8 Tesch, R., 543 Testa, R., 397 Thaiss, G., 498 Theimann, S., 591 Theis, J., 203 Thiam, A., 162 Thibault, P. J., 392 Thieberger, N., 404 Thomas, A. C., 635 Thompson, E. C., 354 Thöni, C., 122 Thorne, B., 278

6/9/14 10:03 AM

760

Author Index

Threlkeld, B., 622 Throop, C. J., 252, 260, 264, 313, 315 Tibshirani, R., 524 Tice, K., 711 Tierney, P., 135, 725 Tihanyi, L., 631 Tilly, C., 59 Timura, C., 374, 378 Timura, C. A., 370, 374, 378–79 Tom, A., 192 Tooker, E., 3 Tornetta, P., 479 Torres, L., 1 Torres, M., 668 Toulmin, S., 23 Traweek, S., 55, 82 Trechter, S., 404 Tronick, E. Z., 296 Trosset, C., 378 Trost, J. E., 543 Trotter, R. T., II, 198, 362, 378, 661–62, 665– 69, 671–73, 678–79, 681–82 Truex, G. F., 369 Trull, T. J., 293, 301, 305 Truscott, K. S., 671 Tryon, R. C., 1 Tsing, A. L., 105, 257, 727 Tsui-James, E. P., 23 Tuchman, B. W., 526 Tukey, J. W., 524 Tula, M. T., 165 Turato, E. R., 542 Turke, P., 299, 303 Turkle, S., 465 Turner, G. H., 415 Turner, L. D. 69, 408 Turner, M., 618 Turner, T. 91, 405 Turner, V., 67, 69, 75 Turton, D., 441, 444 Twine, F., 170 Twinn, S. F., 1 Tyler, S. A., 105, 541 Tylor, E. B., 4, 564, 589 Tylor, S., 40, 83

Urciuoli, B., 168, 393–94, 396, 400, 404, 422 Urry, J., 101 Ushijima, I., 449 Ushiyama, J., 449

Unger, J., 545 Urban, G., 391–94, 397, 399–400, 402–3, 406–7, 409–12

Wacquant, L., 43 Wagner-Martin, L., 164 Waiásse, C., 405

14_107-Bernard.indb 760

Vadez, V., 371, 379 Valente, T. W., 631, 635, 638, 650, 671 Valentine, L., 422 Valian, V., 158 Van Allen, J., 156 Vandebroek, I., 347 van de Bunt, G. G., 644 Vanderkerckhove, J., 380 VanderWeele, T., 635 van Dijk, T., 391 Van Holt, T., 621 van Maanen, J., 28, 34, 107, 251, 486, 698, 724 van Noordwijk, M., 621 VanWey, L., 622 Van Willigen, J., 186, 263, 267, 661–63 Varela, C., 393 Varela, C. R., 393 Varjas, C., 671 Varma, R. K., 204 Vazquez, E., 191, 663, 671 Venegas, M. D., 379 Venkateswar, S., 723–24 Verdery, K., 405 Verschueren, J., 395 Vertovec, S., 634 Vesperi, M. D., 698 Vespignani, A., 634 Vico, G., 58 Villamira, M., 481 Villa Rojas, A., 266 Vincent, J., 156 Visweswaran, K., 174 Vitzthum, V., 295, 303 Voelkl, B., 631 Vogl, C. R., 347 Vogl-Lukasser, B., 347 Vogt, E. Z., 71 Voloshinov, V. N., 395, 400 von Glasersfeld, E., 36, 38, 42 Vygotsky, L. S., 668–69

6/9/14 10:03 AM

Author Index Waitzkin, H., 543 Waldram, J. B., 540 Wali, A., 169, 266 Wallace, A. F. C., 79 Wallerstein, N., 191 Wallerstein, N. B., 191 Walley, C., 162 Walsh, S., 615 Walstrom, M. K., 472 Wanat, C. L., 263–65 Wang, C., 169 Wang, J. W., 481 Wang, P., 644 Wansink, B., 354 Ward, C., 281, 283, 285 Ward, J., 170 Warren, C., 161, 164, 172, 174 Warren, C. A. B., 273 Warschauer, M., 465 Washabaugh, W., 415 Wasserman, S., 358, 373, 375, 631, 644, 671, 679 Wasson, C., 661 Watanabe, C., 305 Waterston, A., 698 Watson, R. S., 71 Watson-Gegeo, K. A., 396 Watters, J. K., 675 Watts, D. J., 650, 679 Waugh, L. R., 64 Wax, M., 144 Wax, R. H., 254, 257, 282–83 Weatherford, E., 160 Weatherford, J. M., 106, 262–63, 278 Webb, B., 5, 7 Webb, B. P., 255 Webb, S., 7, 255 Webster, A., 397 Webster, C. M., 380 Webster’s new international dictionary, 55 Wedeen, L., 5 Weedon, C., 672 Weeks, M., 185, 203, 666, 673–74, 680 Weeks, M. R., 186, 188–89, 200, 202–3, 663, 667, 671 Wegerif, R., 547 Weinberg, R., 155 Weiner, A. B., 273 Weinstein, P., 367

14_107-Bernard.indb 761

761

Weisner, T., 546 Wellenkamp, J. C., 316, 321, 334 Weller, S., 42 Weller, S. C., 1, 118, 242, 347–49, 354, 356– 58, 362, 367, 369–71, 374–80, 466, 508–9, 523, 528–29, 540, 668, 671, 674, 676–78, 683 Wellhouse, K., 168 Wellman, B., 616, 632–33, 635 Wenger, E., 73, 472 Werbner, R., 40 Werner, O., 4, 115, 316, 676 Wertsch, J. V., 669 West, J., 613 Westmoreland, M., 450 Weston, K., 163, 169 Wetle, T., 668 Whitcher, A. L., 112 White, D., 548, 590 White, D. R., 3, 568, 573–74, 582, 590, 632–33 White, G., 396 White, G. M., 394 White, H. C., 632 Whiteford, L., 661 Whitehead, H., 159 Whitehead, N. L., 391 Whitehead, T., 161 Whiting, B., 295 Whiting, B. B., 569 Whiting, J., 295 Whiting, J. W. M., 2, 565, 567, 574, 582, 585, 590 Whitney Goodall, E., 415 Whorf, B. L., 407 Whyte, K. K., 269, 272, 276, 279 Whyte, M. K., 582 Whyte, W., 217 Whyte, W. F., 107, 192, 269, 272, 276, 279, 316 Wicks, S., 413, 416 Widayati, A., 622 Wies, J., 162, 174 Wiget, A., 403, 415 Wikan, U., 321 Wilce, J., 391, 397, 399–400 Wilder, L. B., 506 Wilkie, D., 611 Wilkie, W. L., 358

6/9/14 10:03 AM

762

Author Index

Wilkinson, L., 244 William, D. E., 633 Williams, D., 66, 393, 402, 417–18, 421, 469, 473, 484 Williams, G., 346 Williams, J., 448 Williams, J. S., 448 Williams, M., 191, 479 Williams, P., 479 Williams, S., 166, 200 Williams, T. R., 101 Williamson, K. M., 191, 202–3, 669 Willis, G. B. M, 506 Willis, P. E., 73 Wills, T. A., 635 Willson, M., 280 Wilson, A., 169, 173 Wilson, N., 191 Wilson, R. A., 41–42 Wilson, S. M., 466 Winch, P. J., 302 Winchatz, M. R., 263, 271–72 Winegar, J., 256 Winslow, D., 704 Winterhalder, B., 295 Wintle, P., 449 Wish, M., 372, 375 Witkowski, S. R., 568 Wittgenstein, L., 24 Wolcott, H. F., 4–5 Wolf, E. R., 1 Wolf, G., 449, 454 Wolf, M., 164, 699 Wolfe, A. W., 632 Wolfe, S. H., 540 Wood, D., 609 Woodbury, A., 393, 407, 413 Woodhouse, D. E., 679 Woods, S., 307 Woolard, K., 396–97, 404 Woolgar, S., 37 Worth, S., 442, 454 Worthman, C. M., 540 Wright, R., 166 Wright, S., 662, 726

14_107-Bernard.indb 762

Wu, Y., 621 Wutich, A., 300, 500–501, 548–49 Wylie, A., 152, 161, 166 Xiong, L., 473 Yamaguchi, K., 525 Yan, Y., 313 Yanagisako, S., 159, 165, 168 Yang, S., 679 Yarbrough, C., 358 Yee, N., 473, 480, 484, 488 Yin, Z., 168 Yoder, P. S., 349 Yoganathan, C. S., 671 Yogo, J., 364, 509 Young, F. W., 377, 500 Young, J. C., 120–21, 348, 371, 374, 376, 501 Young, K., 158 Young, R. C., 377 Yue Pink, X., 169 Zabusky, S. E., 82, 104 Zahle, J., 252, 260 Zaki, E. P., 316 Zavella, P., 153, 161, 164, 171, 174 Zeanah, A., 277 Zeiler, M., 612 Zeitlyn, D., 397 Zeller, A., 444 Zeller, R. A., 579–80, 587 Zenaria, E., 153, 172 Zenk, S. N., 307 Zenker, O., 39 Zentella, A. C., 404, 422 Zepeda, O., 398 Zhang, Y., 116–17, 473 Zigon, J., 313, 324 Zihlman, A., 167 Živković, M., 83 Znaniecki, F., 545 Zoran, A. G., 680 Zubin, J. A., 1 Zucker, H., 572, 574 Zyzanski, S., 298, 303

6/9/14 10:03 AM

Subject Index

Note: Letters in italics following page numbers refer to boxes (b), figures (f), and tables (t). AAA. See American Anthropological Association academic anthropology, 8–9; versus applied, 134–36 academic books, 695–97 ACASI (audio-computer-assisted-selfinterviews), 224 accountability, and method, 60 accuracy: participants and, 302–3, 682–83; questions and, 356, 359–60 action anthropology, 186–87, 720 action research, 188–89, 663; participatory, 191–92, 202, 663–65 active listening, 272 actor(s): attributes of, 642; in social network analysis, 631 actor-network-theory, 37 administrators, communicating with, 707–9 advocacy: applied anthropology and, 663; ethics and, 138; issues in, 727; public anthropology and, 720 agency: and method, 60; versus structure, 46 alter(s), 645–46 alter-chasing, 645 alter tie evaluation, 646–47, 647f American Anthropological Association (AAA): El Dorado Task Force, 136; ethics statements, 132–38, 148–49, 278; and informed consent, 142; Principles of Professional Responsibility, 134, 136–38 American Time Use Survey coding system, 304 analysis: computer-assisted, 412–13, 413f–414f; of data obtained online, 481–84; in discourse-centered methods,

410–13, 413f–414f; field notes and, 276; of person-centered interviews, 334–38; in social survey methods, 509–12; units of, 224–25 analytic induction, 545–46 analytic tradition, 22–25 anonymity, ethics and, 278–79, 487 anthropology: definition of, 137; goal of, 25; increase in degrees in, 8–9, 135; nature of, 5–8 applied anthropology, 8–9, 186, 661–93; versus academic, 134–36; methods in, 675–79; presentation of, 711–15; versus public anthropology, 723; setting for, building, 665–66; types of, 662–65 apprenticeships, participant observation and, 259–60 a priori sampling, 221–22 archaeology, feminist methods and, 166–68 architectonics, 71 archiving, discourse-centered methods and, 404–7 area data: analysis of, 617; and attributes, 615–17; collection of, 612–15; and neighborhood studies, 613b articles, 701–3; collections of, 702; popular, 702–3 attitudes: questions on, 344; scales and tests on, 362–66 audio-computer-assisted-self-interviews (ACASI), 224 autocorrelation, spatial, testing for, 590 backtranslation, 507, 509 balanced-incomplete-block designs, 371

763

14_107-Bernard.indb 763

6/9/14 10:03 AM

764

Subject Index

Bateson, Gregory, 452–53, 456 behavioral observation, 293–312; goal of, 293; versus other methods, 294–95; types of techniques in, 295–96, 296f behavior sampling, 297, 299 beliefs: questions on, 345, 376; studies of, 376–80 betweenness centrality, 643 betweenness centralization, 644 bias, 579, 587; elicitation, 646; and participant observation, 264–65, 272–73 biological anthropology, feminist methods and, 166–68 biopolitics, 37 blockmodels, 644 block sampling, 302 blogs, 704–5, 728–29 Boas, Franz, 5–6, 99–102, 132, 445, 533–34 body: and discourse, 415–21, 418f–420f; and meaning, 74–75; person-centered interviews and, 324 books, types of, 695–701 bootstrap technique, 524 boundary definition, 636 Bourdieu, Pierre, 43 breakdowns, 275–76 brokering positions, 649 Bureau of American Ethnology, 720 careers in anthropology, 8–9, 13; articles and, 701–3; books and, 695–701; grant proposals and, 703–4; public anthropology and, 724–25; public speaking and, 705–11 Carey, Sean, 728 case analysis, negative, 542 case-control study, 111t, 118–19, 119f, 501–2, 505 case studies, 698; and questionnaire development, 352 CATI (computer-assisted telephone interviews), 353 censuses, 225–26; online, 480 centrality, types of, 643 centralization, types of, 643 Chicago School, 35 cinema, 440, 442–44, 450–52 circle graphs, 642 circulation, 399–400

14_107-Bernard.indb 764

citizen action groups, working with, 196–97 clandestine research, 137, 278 classification: questions on, 345; studies on, 366–76 Clifford, James, 38 closeness centrality, 643 closeness centralization, 643 cluster analysis, 1, 522, 643; and research questions, 527t cluster sampling, 231–32, 233f, 503 coding: for behavioral observation, 298, 303–5; computer-assisted, 547; and error, 586–89; in text analysis, 543–44 cognitive systems, 647; analysis of, 676–78; content and limits of, 676–77 cohort study, 111t collaborative research, 148, 161, 165; in applied anthropology, 683; communitybased, 185–212; discourse-centered, 410–12, 421–22; and mapping, 622–23, 624f; online, 485–86; public anthropology and, 726; rapport in, 267; in text analysis, 548–49 colleagues, communicating with, 707–9 colonialism, 2, 139, 147, 722; feminist methods and, 159, 162, 164, 168, 171; visual anthropology and, 450–52 commando research, 198 Common Rule, 141–42 communication, in anthropology careers, 705–11 community-based collaborations, 185–212; training for, 203–4 community-based research institutes, and transformational research, 194–95 community development, 191 comparability, 561–63 comparisons: cross-population, large-scale, 121–22, 123f; importance of, 112; of individuals in society, 117–18, 118f; research design and, 110–14 components, 642, 644 computer-assisted analysis: in discoursecentered methods, 412–13, 413f–414f, 417f; and Galton’s Problem, 590; of text, 546–48 computer-assisted telephone interviews (CATI), 353 conceptual-incompatibility hypothesis, 120

6/9/14 10:03 AM

Subject Index confidence intervals, 523 consensus analysis, 523, 528–29; cultural, 540; and research questions, 527t consensus model/theory, 1, 42, 668–69, 678; informal, 378–80 consent. See informed consent constructivism: radical, 22, 35–37; social, 37, 42, 153, 669, 728 construct validity, 27, 357 content analysis, 542–45 content dictionaries, 547 content validity, 27–28, 357 context, 399–400 contextual research, 670–72 continuous recording, 297–98 coordinate systems, 608–9, 623 correspondence analysis, 522, 529; and research questions, 527t counting: in geospatial analysis, 617; in text analysis, 546–47 criterion validity, 27–28, 357 critical discourse analysis, 391 critical pedagogy, 191–92 critical rationalism, 24 critical realism, 44 critical theory, 671–72; public anthropology and, 723, 729–30 cross-cultural research, 561–99; applied, 683; history of, 564–66; types of, 566–70 cross-population comparison, large-scale, 121–22, 123f cross-sectional study, 111t, 501 cultural anthropology: cross-cultural research, 561–99; research strategies in, 102–7 cultural consensus analysis, 540 cultural consonance, 529 cultural domain analysis, 508 cultural intervention research, 662–63 cultural model theories, 668–69, 676 cultural poetics, 60–67 cultural relativism, 162, 727 cultural scripting theories, 669–70 culture, 25; discourse-centered methods and, 392; importance of, 727–28; lists of, 572–74; numerical analysis and, 526–29; as text, 33 culture shock, 281–83; coping with, 282–83; return, 283 cyberspeak, 705

14_107-Bernard.indb 765

765

daily life architecture, 301 data: quality control and, 584; uses of, ethics and, 139 data collection: discourse-centered methods and, 401–4; field notes and, 276; geospatial analysis and, 609–10; history of, 636–40; nature of, 526; online, 479–81; participant observation and, 252, 258–60; public anthropology and, 726; social survey methods and, 505–9; transformational research and, 199–200 Dawkins, Richard, 730 day reconstruction method, 293, 297, 300– 301 deception, online ethnography and, 475, 487–89 decision making, 525–26 deduction, 24–25 deep explanation, 46 deep participation, 44 degree centrality, 643 degree centralization, 643 deixis, 396 de Koning, Martijn, 723, 728 description, 26, 28–29, 32–33, 45; applied anthropology and, 679–82; field notes and, 276; of online life, 484–86; research design and, 103–5 descriptive statistics, 523 development agencies, 714 diachronic comparison, 566, 570 dialogic ethnography, 699 diffusion, 650 digital elevation models, 617, 628–29 discourse: embodiment of, 415–21; study of, rationale for, 393–95; term, 391 discourse analysis, 391 discourse-centered methods, 391–438; term, 391 disproportionate stratified random sample, 573–75 disruptions, in person-centered interviews, 335–36 dissemination, 695–717; online, 485; public anthropology and, 723–29; transformational research and, 204 dissertation, 695–96 dissimilarity coefficients, 521–22; and research questions, 527t

6/9/14 10:03 AM

766

Subject Index

distance calculations, 617–18 diversity: and advocacy, 727; communicating about, 708; social survey methods and, 497 documentation, 27; discourse-centered methods and, 404–7; of informed consent, 142–43; participant observation and, 274–77 domain definition, 347–52 dyad, 642 ecological momentary assessment, 293, 301 ecological theories, 670–71 edge(s), 639–40 edge list, 639, 640f; matrix from, 640f ego-centered networks, 644–49, 679; term, 632; and whole networks, 649 egocentric network, term, 644 eigenvalue, 528 El Dorado controversy, 135–36 elementary unit, 224; definition of, 216b elicitation, 65–67 elicitation bias, 646 email, 705 embeddedness, 45 emic approach, 507–9, 511–12 empiricism, 5–6, 23, 26 engaged anthropology, 187–88, 722–23, 725–27 enregisterment, 399–400 entextualization, 399–400 entrance into community, 195–96; breakthrough in, 269–70; online, 474–76; and participant observation, 265–66 environmental cognition, 622 epistemology, 21–53; feminist, 155–60; future of, 47; semantics and, 85–86 error: coding, 586–89; ethnographer, 583–86; measurement, 520; random, 580, 582–83; sampling, 589–90; and social network analysis, 645; systematic, 579–80 ethics, 131–50; applied anthropology and, 661; discourse-centered methods and, 404–7; education on, 137; feminist methods and, 154; history in anthropology, 132–34; and limits to participation, 280; online ethnography and, 475, 485–89; participant observation and, 277–80; person-centered interviews and, 339n2; principles of, 134, 149; public

14_107-Bernard.indb 766

anthropology and, 725–27; social network analysis and, 651; visual anthropology and, 450–52 ethnobiology, research design in, 117–18, 118f ethnographer error, 583–86 ethnography, 698–99; development of, 25–26; method in, 27–28; and research design, 107; term, 466, 486; visual, 442–44, 450–52 etymology, 56–58, 64–65, 67, 86n7; nonverbal, 70–71 Euclidean distance, 522 evaluation of research, criteria for, 102–3 evaluation research, 662 event-history analysis, 525 Everquest, 467–68, 484 exchange, 502; and person-centered interviews, 319 exclusion criteria, 218 experience sampling methods, 293, 300 experiential knowledge, 164–65 experiments, 110–12, 111t; field, 123–23, 124f; numerical methods and, 524–25; online ethnography and, 488–89 experts: identification of, 672–73; sampling, 674 explanation, 45; deep, 46; in research process, 107–9, 108f exploratory research, 107–9, 108f; to identify locally relevant categories, 116–17, 117f; in interview process, 343, 345–52 external validity, 113, 113t; numerical methods and, 523 factor analysis, 522, 528; and research questions, 527t falsification, 23–24, 81 feminist methods, 151–83; in action, 168–72; challenges in, 172–73; epistemologies, 155–60; ontology, 152–54; and participatory action research, 192; and social surveys, 507 field experiments, 123–23, 124f field maps, 622–23, 624f field memoir, 164–65 field notes, 274–77; discourse-centered methods and, 406–7; ethics and, 279; for online ethnography, 482–83

6/9/14 10:03 AM

Subject Index

767

fieldwork: and research design, 100–102. See also participant observation film, 440, 442–44, 450–52 focal sampling, 296 focus groups, 680–81 food transfers, coding, 305–6 formal logic, 23 Foucault, Michel, 37 framing, 58, 667 free listing, 347–48, 676–77; with contrasting questions, 348–49; interview, 343; with multiple related questions, 348; successive, 349 Freeman, Derek, 38 functional codes, 303–4 functionalism, 26; and participant observation, 256–57 funding: ethics and, 139; and transformational research, 200

graphical analysis tools, and research questions, 527t grounded theory, 541–42 group interviews, 352, 680–81 Guttman scaling, 358, 525

Gadamer, Hans Georg, 30–31 Galton’s Problem, 564, 589–90 gaming: demographics of, 484; online, 465–95 gatekeepers, 265; online, 488 Geertz, Clifford, 31–34 gender, 152–54, 158–60, 170; and online ethnography, 474–75; and participant observation, 273–74; term, 172–73; text analysis and, 537 genealogical method, 3 general information questions, 353–60 genre, public anthropology and, 728–29 geodatabases, 611–12 geodesic, term, 643 geographical information systems (GIS), 601, 680; history of, 602; resources for, 629 geographic coordinate system, 608–9 geospatial analysis, 601–29; applied, 679–80; definition of, 601; example of, 603–8, 604b, 606f; and network analysis, 616b; resources for, 627–29 GIS. See geographical information systems global cross-cultural comparisons, 566–68 global positioning systems (GPS), 601–2; and data collection, 612–15 Google Earth, 625n2 grand tour questions, 346 grant proposals, 1, 5, 703–4

image processing, steps of, 610–11 impression management, online, 474–76 independence, types of, 589 indeterminacy, 68–72 indexicality, 395–97 indices, development of, 356–58 indigenous literature: analysis of, 535; collection of, 533–34 indirect corroboration, 3–4 induction, 24–26; analytic, 545–46; and grounded theory, 541 inductive sampling, 221–22 inferential statistics, 523–24 influential case, 511 informal cultural consensus model, 378–80 informants. See participants informed consent, 135–36, 138, 142–45; and online research, 486–87; unobtrusive research and, 141–42; without forms, 146–48 inquiry: meaning and, 75–78; qualitative, 79–81; systematic, research design for, 107–16 instantaneous recording, 297–99 institutional review boards, 135, 140–42; and field notes, 279; and online research, 486–88; and sampling procedures, 222; and social network analysis, 651 instrumental positivism, 7

14_107-Bernard.indb 767

Hamming distance, 522 Harris, Marvin, 697, 700, 730 Heidegger, Martin, 30 hermeneutics, 22, 29–31; in anthropology, 31–34; positions of, 41–42 high-inference variables, 582 humanism, 6–7, 21 Human Relations Area Files (HRAF), 544– 45, 565–66, 570, 574–76, 592 Human Terrain System, 721–23 hypotheses: elements of, 498; and measurement, 577; in social survey methods, 468–99, 507

6/9/14 10:04 AM

768

Subject Index

interdisciplinarity: feminist methods and, 168; public anthropology and, 727–28; text analysis and, 548–49 internal states, coding, 305 internal validity, 113, 113t; numerical methods and, 524 international organizations, and transformational research, 194 Internet: and anthropology writing, 704–5; discourse-centered methods and, 404; and offline social spaces, 468; online ethnography, 106, 465–95; public anthropology and, 728–29; and social survey research, 506; and textbooks, 700 interpretation, participant observation and, 258–60 interpretive anthropology, 31–34, 45, 56–57 interrater reliability, 587 intersubjectivity, 421–22 interventions: critical, 729–30; cultural intervention research, 662–63; in personcentered interviews, 329–30 interviewing: applied anthropology and, 673– 74; culture and, 506–7; discourse-centered methods and, 402; geospatial analysis and, 613b; group, 352, 680–81; online, 476–79; person-centered, 313–42; for social survey research, 505–8; structured, 343–90 isolates, 642, 644 item analysis, 361 item generation, 347–52 iterative data collection, 526

language: ideologies of, 397–99; online, 473; and participant observation, 270–71; and person-centered interviews, 318 latent root, 528 layers: and attributes, 615–17, 615f; in geospatial analysis, 602, 604, 605f–606f; preparing, 611–17; thematic, 618–320 lay people, training for participatory research, 201–4 lecture style, 706–7 Levy, Janet, 136 Likert scale questioning, 508 linear programming, 525 line data, GPS, 614; and attributes, 615–17 linguistic anthropology, 392–93; feminist methods and, 166–68; and text analysis, 533–39. See also discourse-centered methods listening: active, 272; and person-centered interviewing, 326–28; in social survey research, 507 lists: of cultures, 572–74; length of, 350; linked, 349 loadings, 528 logic, 23; of situation, 45 logical positivism, 23–24 logistic regression, 525; multiple, and research questions, 527t longitudinal networks, 644 longitudinal studies, 112, 500–501 low-inference variables, 582 low-risk research, 141–42

jackknife technique, 524 jargon, in academic writing, 696–97 journals, refereed, 701–2 judgment sampling, 234

Mach, Ernst, 26 Malinowski, Bronislaw, 2, 25–29, 37–38; and ethnography, 698; and participant observation, 252–53; and research design, 99–102; and social survey methods, 498 maps, 610, 622–23, 624f, 679–80 matching, 505 matrices: and sampling, 219–21, 220f–221f; in social network analysis, 636f, 640f MDS. See multidimensional scaling Mead, Margaret, 38; and ethics, 132; and participant observation, 255–56, 261–62; and public anthropology, 721, 730; and research design, 100; and trade books, 697; and visual ethnography, 452–53, 455 mean, 520

Kant, Immanuel, 30–31, 36 kinship studies, 633 knowledge: experiential, 164–65; limits of, 21, 41–44; nature of, 7, 21–53; and power, 37; situated, 161; sources of, 44–46 knowledge tests, 344, 360–62 Kroeber, Alfred, 1, 5, 101 KWIC technique, 547 Labanotation, 418–19, 418f–420f, 421 landscape patterns: and change studies, 620, 621b; and optimal foraging studies, 618b

14_107-Bernard.indb 768

6/9/14 10:04 AM

Subject Index meaning: acquisition of, 72–74; body and, 74–75; hermeneutics and, 29–31; and inquiry, 75–78; in methods, 55–96; nonverbal, elicitation of, 65–67; in performance, 60–67; in person-centered interviews, 334–38; play of, 64–65; recovering, 67–68 measurement: in cross-cultural research, 576–90; errors in, 579–90; numerical analysis and, 519–20; in social network analysis, 642–44, 643t; in social survey methods, 505–9 measurement error, 520 measurement independence, 589 media: communication with, 709; public anthropology and, 728–29. See also publication; social media median, 520 memoirs, 164–65 mental models, 541 metaphors, 539–41 metapragmatics, 395–97 method(s), 1–9; approaches to, 1; eclectic, 8–9; grant proposals on, 703–4; history of, 3–4, 25–26; introduction to, 1–17; meaningful, 55–96; ownership of, 4–5; term, 55 midrange theory, 667; applied anthropology and, 667–72 military cooperation, 138, 721 mini-tour questions, 346 misconceptions about anthropology, combating, 708–9 misinformation, sampling and, 223–24 mixed methods, 1–2, 9; and epistemology, 44; and online ethnography, 478; social survey methods, 498, 511–12, 512f; and text analysis, 533, 549 MMORPG, 468 mode, 520 Morgan, Lewis Henry, 4 mosaicking, 611 multidimensional scaling (MDS), 372, 375– 76, 522, 522f, 548; and research questions, 527t; and social network analysis, 642 multiphase sampling, 234 multiple regression techniques, 524–25, 591 multi-stage cluster sampling, 503 multi-temporal change analysis, 620–21, 624f

14_107-Bernard.indb 769

769

multivariate analysis, in social survey research, 511 mutual learning, 197 name generators, 645, 645t name interpreters, 646 narrative(s): intra- and intercultural comparisons of, 537–39; and questionnaire development, 352. See also text analysis narrative emplotment, 75 narrative performance, patterns in, 535–37 National Association of Practicing Anthropologists, 135, 137 National Security Education Program, 133 negative case analysis, 542 netnography, term, 466 network analysis, and spatial relationships, 616b network closure, 649 network model, 644 network scale-up method, 646 node(s), 639 node list, 639–40, 640f; matrix from, 640f nominated expert sampling, 674 non-probability sampling, 225, 233, 504–5; definition of, 217b; size selection for, 241–44 nonverbal communication: etymology and, 70–71; meaning of, elicitation of, 65–67; in person-centered interviews, 336–38 note taking: in field, 274–77; in interviews, 332–33. See also field notes numerical analysis, 519–31; and culture study, 526–29; process of, 525–26; and research questions, 527t numerical methods, 519–31 numerical transformations, and research questions, 527t objectivity, 6, 36, 38, 42; feminist methods and, 152–53 observation, 271–72; behavioral, 293–312; limits to, 272–74; online, 470–73, 486; person-centered, 313–42; in social network analysis, 638; units of, 224–25. See also participant observation observational designs, 111t observer effect, 55, 302

6/9/14 10:04 AM

770

Subject Index

obviousness, 57 one-mode data, 637t, 638–39, 639f one-zero recording, 299 online ethnography, 106, 465–95; continuum of, 472; locus of, 467–70; risks of, 472–73 ontology, 21; feminist, 152–54 open coding, 543 optical remote sensing (ORS), 602–3 optimal-foraging theory, landscape patterns and, 618b ordinal, 519 ordinary least squares multiple regression, and research questions, 527t paired comparison, 370–72 panel studies, 501 papers, presentation of, 706 PAR. See participatory action research partial truths, 38–39 participant observation, 2, 4–5, 251–92; advantages of, 258–61; balance in, 261–71; basic skills of, 261–62; continuum of, 262–65; definition of, 251; development of, 252–58; online, 470–73; precautions with, 254; research design and, 106 participants: and accuracy, 302–3, 682–83; agency of, 722; communication with, 709–11; discourse-centered methods and, 406, 421–22; and error, 583–86; in interviews, 316–17; locating, 222–24; and mapping, 622–23, 624f; reactivity by, 302; selection of, 215–49; and social survey methods, 506–7; term, 224b; treatment of, 328 participation: costs of, 281–83; limits to, 280 participatory action research (PAR), 191–92, 202, 663–65 participatory methods, 185–212 paternalism, 146–48 path analysis, 525, 592 path dependency, 59 Peacock, James, 137 Pearson’s r, 521 perceptual mapping, and research questions, 527t performance, meaning in, study of, 60–67 performativity, 395–97 permutation technique, 524

14_107-Bernard.indb 770

personal networks, 644–49; life cycle of, 635 person-centered interviews, 313–42; analysis of, 334–38; beginning, timing of, 318–20; ending, 331; location of, 320; number and length of, 331–32; procedures for, 317–31; topics of, 320–26, 330–31; transcribing, 333–34 photography, and visual anthropology, 440, 444 phylogenetic methods, 592 physical anthropology, feminist methods and, 166–68 pile sorting, 366–70, 368f, 677–78 place: and sampling, 219. See also geospatial analysis play, of meaning, 64–65 poetic function, 64 poetics, cultural, 60–67 poetic wisdom, 58, 75; approach to, 84–85; as method, 60–67; postures of, 78–79 points, GPS, 612–15; and attributes, 615–17; and neighborhood studies, 613b policy research, 662 politics: and anthropology, ethics and, 131–33; feminist methods and, 152; Internet and, 469; participatory action research and, 202; poetics and, 62; public anthropology and, 722–23, 726; and sampling, 235t popularization, 729–30 positionality, 664 positivism, 7, 21–25; in anthropology, 25–29; instrumental, 7 postmodernism, 22, 37; in anthropology, 37–41, 672; positions of, 42 power: discourse-centered methods and, 410; feminist methods and, 161–66; knowledge and, 37; poetics and, 62; public anthropology and, 729–30; and questionnaire development, 538 predictive validity. See criterion validity prejudice, 31 preprocessing, in geospatial analysis, 610–11 presentation, 695–717 pretesting, 506, 579 Price, Richard, 38–39 primary data: and cross-cultural research, 566, 569; and sampling, 571–72

6/9/14 10:04 AM

Subject Index Principles of Professional Responsibility, 134–38 probabilities, and research questions, 527t probability sampling, 225–26, 503–4, 523; definition of, 217b; size selection for, 239–41 professional books, 695–97 profile data, 373–75 PROFIT analysis, and research questions, 527t projection systems, 608–9 proportionate stratified random sample, 573 proprietary research, 137–38, 140 prosody, 336–37 Proteus effect, 480 proxy measures, 581–82 public anthropology, 187, 719–33; advantages and disadvantages of, 724–25; versus applied anthropology, 723; definition of, 719; goals of, 722–23; history of, 720–22; as mainstream, 731–32 publication, 695–717; ethics and, 278–79; online, 485; public anthropology and, 723–24, 728–29; transformational research and, 204 public speaking, 705–11 purposive sampling, 234, 235t–236t Q-sort, 379 quadratic assignment, and research questions, 527t qualitative methods, 1–2, 6; meaning and, 79–81; and sampling size, 241–43; in social survey research, 511–12, 512f quantiles, 520 quantitative methods, 1–2, 6; and sampling size, 243–44; in social survey research, 511–12, 512f quasi-experimental designs, 111–12, 111t; threats to validity in, 113–14, 113t questionnaires, 343–90; construction of, 353– 80; development of, 346–52; for social surveys, 505–6, 509–10 questions: culture and, 506–7; formats for, 344, 354; open-ended, 343–44, 354; in person-centered interviews, 317, 328–30; precautions with, 354; in social network analysis, 358–59, 646; for tie definitions, 637t. See also research questions

14_107-Bernard.indb 771

771

radical constructivism, 22, 35–37 Radin, Paul, 6 radiometric resolution, 608 random error, 580, 582–83 rapid ethnographic assessment, 681–82 rapport: breakthrough in, 269–70; establishment of, 266–68; ethics and, 278 Rasch model, 358 raster data, 602 rating scales, 379 rationalism, 519; critical, 24 realism, 59; critical, 44; ethnographic, 698 reciprocity, participant observation and, 268–69 recording: discourse-centered methods and, 404–7; online ethnography and, 488; of person-centered interviews, 332–33; for questionnaire development, 350; rules, for behavioral observation, 296f, 297–301 recruitment, 244, 245t, 247t, 674 refereed journals, 701–2 reflexivity, 164; applied anthropology and, 664; community-based participatory research and, 190; discourse-centered methods and, 394, 397–99; ethnography and, 699; public anthropology and, 723 regional cross-cultural comparisons, 566, 568–69 registers, 400, 407 relationship, and knowledge, 42–43 relativism, cultural, 162, 727 reliability, 27–28, 360–62; applied anthropology and, 682–83; interrater, 587; similarity data and, 375–76; social survey research and, 509–10 remote sensing, 601; history of, 602–3 repeated measures design, 124–26, 126f reproduction, 170 research, nature of, 526 research design, 97–129; benefits of, 98–99; examples of, 111t; grant proposals on, 703–4; importance of, 97–98; influences on, 99–102; and participant selection, 215–24; in social survey methods, 499– 502; for systematic inquiry, 107–16 research film, 443–44, 453–54 research model, ethically conscious, 138 research population, defining, 215–24 research process, 107–8, 108f; ethics and, 139

6/9/14 10:04 AM

772

Subject Index

research questions: and behavioral observation, 294–95; in cross-cultural research, 563–64; numerical analysis tools and, 527t; participant observation and, 260–61 research strategies, 97–129; contemporary, 102–7; systematic, 116–26 resolution, in geospatial analysis, 607–8, 607f respondent-driven sampling, 231, 232f, 675 respondents. See participants response rate, 504 retextualization, 399–400 return culture shock, 283 rhetoric, 83 Rivers, W. H. R., 3–4 roster method, 637 Sahlins, Marshall, 34–35 salvage ethnography, 698 sampling, 114–15, 215–49, 523; and applied anthropology, 672–75; approaches to, 225–34; and behavioral observation, 295– 97; and comparison, 573–76; and content analysis, 543; and cross-cultural research, 570–76; definition of, 215; and error, 589–90; integrating approaches, 234, 237f; method selection, 237–39, 238f; planning process for, 246t–247t; size selection, 239–44, 240b, 351–52, 502, 528, 570–71, 591; snowball, 635–36, 675; and social survey methods, 502–5; terminology in, 216b–217b sampling frame: definition of, 216b; establishment of, 226–27 sampling independence, 589 satellite data, 627–28 saturation, 241–43, 242f, 542, 674; term, 248n1 scale-free networks, 650–51 scales: adaptation of, 363–64; development of, 356–58, 362–66; rating, 366; in social survey research, 508 SCCS. See Standard Cross-Cultural Sample schema analysis, 539–41 Schneider, David, 34–35 science and technology studies, 37 scientific approaches, 6–7, 21; and research design, 99–102

14_107-Bernard.indb 772

scripting theories, 669–70 secondary data: and cross-cultural research, 566, 569; and sampling, 572–73 Second Life, 465, 467, 469, 474, 484 secret research, 137, 140 segmenting, 543 Seiendes, term, 30 Sein, term, 30 self-monitoring, and person-centered interviewing, 326–28 self-reports, of behavior, 299–301 semantic network analysis, 547–48 semantics, and method, 55–96 semiotics, and method, 55–96 sentence-substitution interviews, 373–75 Sewell, William, 46 sexuality, 159–60, 169; cross-cultural comparisons and, 561–63 Shaw, Clifford, 39 shifters, 395–96 significance tests, 523 silence, in person-centered interviews, 330 similarity coefficients, 521; and research questions, 527t similarity data, 370–72, 372f–373f; profile, 373–75 simple percent agreement, 544 simple random sample, 227–28, 503, 573 simple summary statistics, and research questions, 527t simultaneity, and coding, 304–5 situated knowledge, 161 six degrees of separation, 631, 650 small world theory, 650 snowball sampling, 635–36, 675 social capital, 635, 649–50 social circles, 635 social constructivism, 37, 153, 669, 728; positions of, 42 social foci, 635 social interactions, coding, 306 social learning theory, 651 social media, 465–95, 631, 650, 704–5; and data collection, 481; and debate, 725; public anthropology and, 728–29 social network: specialized, data on, 678–79; term, 632, 651

6/9/14 10:04 AM

Subject Index social network analysis, 631–57; history of, 631–36; limitations of, 651–53; measurement and, 642–44, 643t; models in, 644; and spatial relationships, 616b; visualization in, 641–42, 641f, 648f social network questions, 358–59 social network theories, 671 social poetics, 60–67, 77–78, 80 social structures, analysis of, 678 social survey methods, 497–515 societies, lists of, 572, 574 Society for Visual Anthropology, 448 sociocognitive theories, 668–70 sociogeographic mapping, 679–80 sociology: origins of, 22–23; and text analysis, 539–49 sociometry, 632 software. See technology sojourner, term, 285n7 soundscape, 403 space: discourse-centered methods and, 404; geospatial analysis, 601–29 spatial autocorrelation, testing for, 590 spatial communication, 76–77 Spearman-Brown Prophesy Formula, 377 spectral resolution, 607 Spier, Leslie, 2 spring embedder, 641f, 642 stakeholder, 219; definition of, 217b; template for, 221f stance: discourse-centered methods and, 399–400; feminist methods and, 130–61; person-centered interviewing and, 326 Standard Cross-Cultural Sample (SCCS), 574–75, 593n1 standard deviation, 521 standpoint approaches, 152–53, 160–61 static-group comparison, 111t statistical analysis, and cross-cultural research, 590–92 statistics, 519–31; types of, 523 stratification, 218 stratified random sample, 228–29, 228f, 503–4, 573 strong ties, 650 structural codes, 303–4 structural hole, 635 structure, versus agency, 46

14_107-Bernard.indb 773

773

structured interviewing, 343–90; preparation for, 345–52; techniques for, 353–80 study population: definition of, 216b; primary, 217–18 study sample, definition of, 216b subjectivity, and research design, 101 subjects. See participants subpopulations: determining, 218–21; term, 216b sufficient redundancy, 674 summarizing, in questionnaire development, 351 summative score, 357 survey(s): in applied anthropology, 681; online, 479–81; social survey methods, 497–515; term, 353 survey method, 3, 637–38 survival analysis, and research questions, 527t symbols, 34–35; instability of, 65; nonverbal meanings in, elicitation of, 65–67; text analysis and, 537–38; use of, 56–57; visual anthropology and, 442–44 synchronic comparison, 566, 570 systematic error, 579–80 systematic sample, 229, 503; in applied anthropology, 673–74 tacit knowledge, participant observation and, 252 targeted sampling, 675 taxonomic questions, 346–47 Taylor, Carl, 3 teaching, 706–7 teamwork. See collaborative research technology: and behavior observation, 306–7; and discourse-centered methods, 404–5, 407–8; and field notes, 277; and geospatial analysis, 601–3, 608–9, 611–12, 629; and online ethnography, 474–75, 477; and person-centered interviews, 332–33; and social network analysis, 641–42, 647, 653n1–653n2; and social networks, 616b; and text analysis, 534, 546; and visual anthropology, 439–63 teleology, 76 temporal resolution, 608 test item analytical procedures, 525 test-retest reliability, 361

6/9/14 10:04 AM

774

Subject Index

text(s): culture as, 33; hermeneutics and, 29–31; importance of, 6; indigenous literature, 533–35; and questionnaire development, 352 text analysis, 533–59; history of, 533–34; intra- and intercultural comparisons, 537–39; linguistic tradition and, 533–39; sociological tradition, 539–49 textbooks, 699–701 theoretical sampling, 222, 542 theory, 1, 5–7, 55–96; analytic induction and, 545–46; building and testing, 666–67; content analysis and, 542–45; development of, 498–99, 667–72; feminist, 152–54; grounded, 541–42; hermeneutics and, 32–33; participant observation and, 256–58; and person-centered interviews, 313–14; publication and, 696; research process and, 108–9, 108f; social network analysis and, 631–36, 649–51 thick description, 32; applied anthropology and, 679–82; of online life, 484–86 tie definition, 637, 637t time: and cross-cultural research, 566, 570; and discourse-centered methods, 404; and public anthropology, 728–29; and questions, 356; and sampling, 219, 234– 37; and social network analysis, 644 time allocation, 294 time diaries, 293, 297, 299–301 time series, 525 time-space sampling, 230–31, 230f trade books, 697–98 transcriptions, 333–34; collaborative, 410–12; computer-assisted analysis and, 412–13, 413f, 417f; in discourse-centered methods, 407–10, 409f, 410–12, 414f–417f, 420f; and text analysis, 536 transformational research, 192–205; methods for, 195–205; principles of, 193 translation: in discourse-centered methods, 407–10; public anthropology and, 727, 731 translation model, 43–44 triad, definition of, 644 triad census, 644 triad designs, 370–72 triangulation, 520 trope, 56–58, 86n1 trust, and person-centered interviews, 318–20

14_107-Bernard.indb 774

truths, partial, 38–39 two-community comparative design, 120–21, 121f two-mode data, 638–39, 639f typical, 520; variation from, 520–21 uniqueness, 561–63 United States Agency for International Development, 714–15 United States Geological Survey, 610 units: of analysis, 224–25; elementary, 216b, 224; in social survey methods, 500 universals, 34–35 Universal Transverse Mercator (UTM), 609 university departments/centers, and transformational research, 193–94 use, and meaning, 56 validity, validation, 21, 26–27, 38; applied anthropology and, 682–83; cross-cultural research and, 580; numerical methods and, 519, 523; questionnaire development and, 357, 359–60; research design and, 103, 112–14, 113t; sampling and, 223; similarity data and, 375–76; social survey research and, 509–10 variables, 519; high- and low-inference, 582; range of, 521, 521f; in social network analysis, 645–46 variation, 68–72; numerical analysis and, 519–20; from typical, 520–21 vector layers, 602, 604, 605f–606f viability, 36 Vico, Giambattista, 56, 58, 64, 67, 75–76, 84 Vienna Circle, 23 Vietnam War, 134 virtual anthropology, 106 virtual ethnography. See online ethnography visual anthropology, 439–63; geneaology of, 439f, 442–44; history of, 439–42; importance of, 453–55; mainstreaming and professionalizing, 448–50; term, 442 von Glasersfeld, Ernst, 36 vulnerable populations, research with, 163 weak ties, 650 Web 2.0, 465 Webb, Beatrice, 5 Weber, Max, 32, 45

6/9/14 10:04 AM

Subject Index weighting, 226 Whiting, John, 2 whole-network analysis, 632, 679; data collection for, 636–40, 636f, 637t; definition of, 636; and ego-centered networks, 649 World Bank, 712–14 World Ethnographic Sample, 575 World Geodetic System, 609

14_107-Bernard.indb 775

775

World of Warcraft, 465, 469–71, 473–76, 479–80, 484, 486, 488–89 writing, 695–717; field notes, 274–77; grant proposals, 1, 5, 703–4; hermeneutics and, 34; positivism and, 28–29; postmodernism and, 39–41; public anthropology and, 723–24, 728–29 zero-order analysis, 524

6/9/14 10:04 AM

14_107-Bernard.indb 776

6/9/14 10:04 AM