Discussion Question 1 and 2 Due May17

Ethical Principles for Conducting Fieldwork Author(s): Joan Cassell Source: American Anthropologist, New Series, Vol. 82, No. 1 (Mar., 1980), pp. 28-41 Published by: Wiley on behalf of the American Anthropological Association Stable URL: http://www.jstor.org/stable/676126 Accessed: 17-05-2018 06:00 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://about.jstor.org/terms

American Anthropological Association, Wiley are collaborating with JSTOR to digitize, preserve and extend access to American Anthropologist

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Ethical Principles for Conducting Fieldwork

JOAN CASSELL Center for Policy Research

Federal regulations to protect human subjects assume a particular relationship between experimenter and subject based upon biomedical research. Because this relationship is not found in every type of research, the regulations are not universally applicable. Rela- tionships between those who study and those who are studied vary with the kind of re- search. Such variations depend on the relative power and control of the researcher, the direction of interaction, and the level of possible harms and benefits. Measured on these continua, fieldwork is at the opposite end of the spectrum from biomedical research. Con- sequently, the ethical system on which federal regulations are chiefly based-using utilitarian risk-benefit calculations--becomes ineffective and inappropriate when applied to proposed fieldwork. Here, investigators have relatively little power and less control of the setting and context of research, interaction flows in two directions, and calculable harms and benefits are comparatively low. Various models of fieldwork are described, each having differing relationships between investigators and subjects. It is suggested that the Kantian categorical imperative, with its principle of respect for human autonomy, might be useful in judging the ethical adequacy of these varieties offieldwork. [fieldwork, ethical principles, research relationships, federal regulations, risk-benefit calculations]

DURING THE PAST TEN YEARS, a complex network of federal regulations has been devel- oped to protect the human subjects of biomedical and behavioral research. Those regu- lations contain a model of the relationship between experimenter and subject that has been developed from biomedical research (Cassell 1978; Olesen 1979). That model is seriously flawed because the relations between those who study and those who are studied are not fixed but vary with the type of research. Consequently, the federal regulations derived from that model have limited applicability to other research methods.

JOAN CASSELL received her Ph.D. from Columbia in 1975 and subsequently studied symbolic anthropology as a Visiting Fellow at Princeton. Her book on the contemporary American women's movement was published in 1977. She is executive director of a project investigating the ethical problems of fieldwork. A book on Federal Regulations: Ethical Issues and Social Research, edited by Murray L. Wax and Joan Cassell, has just been pub- lished by Westview Press for the AAAS.

_-:-?,,:::-i:::::;::::s:::?

...-.. ::

Copyright ? 1980 by the American Anthropological Association 0002-7294/80/010028-14$1.90/1

28

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELDWORK ETHICS 29

RELATIONS BETWEEN INVESTIGATORS AND SUBJECTS

Confusion about the varying relations between those who study and those who are studied may be increased by linguistic habit. Traditionally, those who are studied by ex- perimentation have been called "subjects"; those who are surveyed have been called "respondents"; and those studied by participant observation have been called "infor- mants" (Mead 1969:361). This usage is so well established that experts who examine "human subject regulations" seem almost automatically to assume that they are dealing with experimental subjects.'1

Differences in relations between investigators and subjects in various types of research become apparent when they are compared along four dimensions: (1) the relative power of investigators, as perceived by subjects; (2) the control of the setting where research takes place; (3) the control of the research context - how the interaction is designed and defined; and (4) the direction of research interaction: whether it flows primarily in one direction or two. These dimensions are continua, and the relationships in different types of research vary systematically along them.

I am now going to review four ideal typical forms of research, measuring each along the four dimensions. The research forms reviewed are biomedical, psychological, survey, and fieldwork (participant observation). Table I summarizes the argument.

Biomedical experimentation is located at the far end of these four dimensions.2 Here, experimenters tend to be perceived by subjects as having great power, since they may de- pend upon these physicians for their medical care. Under such conditions, the patients may be afraid to refuse to take part in an experiment. For example, in a large American teaching hospital, physicians recruited their own female patients as subjects. Despite the fact that the women had signed forms consenting to the experimental procedures, 39% of those interviewed said that they did not know they were participants in a research proj-

TABLE I. RELATIONS BETWEEN INVESTIGATORS AND SUBJECTS.

Fieldwork

Biomedical Psychological Survey (participant experimentation experimentation research observation)

Investigators' power as perceived by subjects H H- M M- L Investigators' control over research H H - L O (negative) control setting by subject Investigators' control

over context of research H H M- L

Direction of

research I--S I--S I-S I-S interaction one-way one-way limited two-way two-way

H- high M- medium L- low

O - none

I - investigator S - subject

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

30 AMERICAN ANTHROPOLOGIST [82, 1980

ect, and an additional 8% felt coerced and would have preferred not to participate (Gray 1975:128). Biomedical investigators control the hospital or laboratory setting in which research is carried out; they control the context - how the experiment is designed and de- fined- and the research interaction is primarily in a single direction: experimenters act upon subjects and their diseases. This is not to say that there is no personal flow of two- way interaction between experimenters and subjects; there may be a great deal. But this personal interaction is not part of the research, and it would have to be controlled for if there were a chance that it might influence the experimental outcome. Psychological experimenters generally have less power over their subjects than do bio-

medical researchers. Rarely are their subjects hospitalized or ill. Although college sophomores- the proverbial guinea pigs for psychological experimentation - may de- pend upon researchers for good grades, this dependency is not life threatening. Admit- tedly, experimenters may invoke the prestige of a university or other large institution, as well as the more generalized prestige of "science"- all of which subjects may perceive as authoritative. For example, Milgram's controversial experiments on the human disposi- tion to be cruel in an authoritarian setting took place at Yale and were later carried out in a New Haven research laboratory (1963, 1965). It is possible that the subjects' obe- dience to the commands of the authoritarian experimenter reflected a trust that Yale University and "science" would allow nothing dangerous to happen. In the control of the setting and context of research, differences exist between laboratory and field psycholog- ical research. Laboratory experimenters have control over both, whereas psychological field researchers control only the context of experiments: it is they who design the experi- mental situation that encourages certain behaviors and discourages others. In either case, research interaction is primarily one-way: when an experimental subject reacts inappro- priately-for example, by trying to "read" or "please" experimenters (Orne 1962; Orne and Scheibe 1964)-investigators, who are interested only in a limited and predefined set of responses to their stimuli, perceive this behavior as "noise." In psychological experi- mentation, then, researchers have high-to-medium perceived power, high-to-low control of the setting, high control of the context of research, and one-way interaction (within the experimental paradigm). This situation differs in survey research. Although investigators may be perceived as

prestigious - especially when a powerful institution or "science" is invoked to increase response rate-researchers lack power and have little perceived ability to exact com- pliance or to punish for noncompliance. Investigators do not control the setting although they do exert control over the context of research: structured questionnaires allow for a limited set of responses; and even an unstructured instrument maps out in advance those areas to be explored (and those to be ignored). When a structured questionnaire is used, the research interaction is, in effect, one-way: investigators apply verbal stimuli, and those who are studied (respondents) produce responses. In an unstructured interview, in- teraction becomes increasingly two-way; although investigators try to assume primary control over the shape (and time involved) of the research interaction, they are not always successful. In survey research, then, investigators have medium-to-low perceived power, no control of the research setting, and medium-high to medium-low control of the con- text. Research interaction ranges from one-way to a somewhat limited two-way flow.

While conducting fieldwork, investigators have comparatively little power over those who are studied: informants are usually free to leave the situation or to decline to enter interaction. In point of fact, those who are studied frequently have some power over the investigators, who may depend upon the host community for shelter, food, and, on occa- sion, protection (Wax 1977:323). Those who are studied control the research setting and exert control over the context of research as well: both informants and investigators in- fluence the interaction, and each works to place it in particular social, emotional, and

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELDWORK ETHICS 31

political contexts. Contradictory definitions of the research enterprise and its relation- ships, by those who study and those who are studied, may lead to conflict or misinterpre- tations, but these contradictions can also lead to the investigators' deeper understanding of how informants organize and define reality.

Unlike other modes of investigation, fieldwork does not place human interaction out- side the research paradigm: instead, the paradigm is based upon human interaction, in all its richness, variety, and contradiction. In addition, field-workers are their own measuring instruments. These two principles, which differentiate participant observation from other varieties of research, have ethical consequences for the behavior of the field worker, as discussed in a later section. Fieldwork interaction flows freely in two direc- tions, with observers and observed participating on a comparatively equal basis. On some occasions it is the observers, and on others the observed, who initiate interaction, with the

extent of participation determined by individual situations and personal predilections. (This analysis does not characterize deceptive fieldwork, which is discussed below.) In true-or open-participant observation, field-workers do have their research agenda and must acquire skill in directing interaction and eliciting information, but those who are studied also have their own agenda and frequently exhibit remarkable skills in elicitation (see Wax 1971:181-220). In fieldwork, then, power is shared between investigators and subjects, with subjects having somewhat more power to frustrate research than research- ers have to compel them to participate. Subjects control the setting of research and in- fluence the context, with interaction flowing comparatively freely in both directions.

VARIATION IN RISKS AND BENEFITS

I have discussed how different types of research vary in the perceived power of investi- gators, the control of the setting and the context of research, and the direction of inter- action. There are similar variations when risks are weighed against benefits or, more cor- rectly, as May has pointed out, harms against benefits (1978).3 Thus, subjects of biomedical experimentation may suffer serious harm from certain procedures; on the other hand, they may benefit greatly from the same experimental procedures or may benefit others without direct benefits to themselves. Subjects of psychological experimen- tation are not exposed to the same degree of easily calculable harm as biomedical sub- jects, but then, rarely do they benefit to the same degree, although a larger class of in- dividuals might conceivably benefit, and the findings may contribute to human knowl- edge.4 (See Kelman 1968 and 1972 for a discussion of some of these issues.) Survey research would seem to have less potential for serious harm than either biomedical or behavioral experimentation, with the greatest injury being violation of privacy or confi- dentiality.5 The benefits to individuals are often less substantial than those offered by biomedical or even psychological experimentation, with the greatest benefit accruing to science or human knowledge.

In the conduct of fieldwork there is a comparatively minimal level of harm- again, primarily the violation of privacy or confidentiality. The most serious harms in fieldwork can occur when findings are disseminated or published (see Cassell 1978). In this paper, however, I am following human-subject regulations and the recommendations of the Na- tional Commission for the Protection of Human Subjects, and examining only the con- duct--not the consequences--of fieldwork. The tangible benefits of fieldwork may also be somewhat difficult to enumerate; these are primarily the correction of misperceptions about various groups and the advancement of human knowledge.

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

32 AMERICAN ANTHROPOLOGIST [82, 1980

REGULATING RESEARCH

In biomedical research, where investigators have high perceived power, high control of the setting and context of research, and where they direct research interaction toward sub- jects, it seems particularly important that the asymmetry between researchers and subjects be rectified, or at least somewhat neutralized; subjects should be given opportunity to de- cide whether they wish to take part in the research and should be informed in advance that they can freely drop out of the experiment whenever they wish. This requirement of in- formal (and documented) consent is particularly important in a research situation where subjects may be seriously harmed as well as greatly benefited. Here it seems logical that potential subjects and reviewers, in deciding whether a particular project is justified, weigh harms against benefits. Vaughan and Sjoberg point out that this risk-benefit criterion, as required by human-subject regulations,

represents a form of utilitarian reasoning, a position that argues for the calculation of pleasure over pain as a moral guideline. In its broader expression, activities are judged in terms of the "greatest good for the greatest number" standard [1978; see also Englehardt 1978].

In situations such as biomedical experimentation, where potential harms and benefits are high, this principle seems appropriate for judging projected research.

Moving along the spectrum of research, from biomedical to fieldwork, the perceived power of investigators, their control of the setting and context of research, and the uni- directionality of interaction all diminish. It follows logically that investigators will have increasing difficulty predicting just what will occur during interaction, or defining in ad- vance just what they will consider "legitimate" or studiable behavior as opposed to "noise." Because investigators are unable to predict exactly what is going to happen dur- ing the course of research interaction, it becomes self-contradictory for investigators to secure "informed" consent before the research is initiated. In addition, in the move across

the spectrum from biomedical experimentation to participant observation, calculable harms become less serious and benefits less immediate. As a result, weighing potential harms against benefits before research is carried out becomes an exercise in creativity, with little relevance to the ethical dilemmas and problems that may emerge during the research. In such research, utilitarian risk-benefit calculations become ever less appro- priate in judging the value or ethical adequacy of a research project. (Vaughan and Sjoberg [1978] also question the usefulness of utilitarian theory as a guide to action in this type of research.)

A more appropriate ethical framework for judging fieldwork might be constructed upon respect for individual autonomy based on the fundamental principle that persons always be treated as ends in themselves, never merely as means - the Kantian categorical imperative (Kant 1959; see also Englehardt 1978:19).

Before applying the categorical imperative to the conduct of fieldwork, I would like to examine participant observation more closely. The preceding discussion of various rela- tionships between investigators and subjects treated fieldwork as though it consisted of a comparatively homogeneous set of relationships.

The position of field-workers can be understood as a role that is negotiated over time, a social construction, composed in part by the ways in which those who are studied will tolerate a stranger and in part by the temperament and interests of the investigator. Thus, there is a variety of relations, and these merit separate analysis. Table II sum- marizes the argument.

RELATIONS BETWEEN FIELD-WORKERS AND INFORMANTS

Fieldwork is at the far end of the continuum when compared with methods where in-

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELDWORK ETHICS 33

TABLE II. RELATIONS BETWEEN FIELD-WORKERS AND INFORMANTS.a

Verandah Noblesse Going model oblige native Advocate

Field-workers' power as perceived by subjects H H- M L - O H- M

Field-workers' control

over research

setting H M L - O L Field-workers' control

over context of research ? L L - O M - L

Direction of research

interaction F-- I F- I F-- I variable

limited two-way two-way two-way

a The varieties of fieldwork are rated relative to each other, and the categorizations are not com- parable to those of Table I. See reference note 7 of text.

vestigators have significantly greater perceived power and control more of the setting and the context of research, and where research interaction flows primarily from investigators toward subjects. Nevertheless, variations do exist within fieldwork, and they can be measured along the same dimensions used for the other modes of research.

At one extreme - by now historically obsolete in many parts of the world - is the veran- dah model of ethnography. (This argument follows the lead of Wax and Cassell 1979.) Think of the field-worker, sitting on the verandah of the government station, sending for a "native" who will be subjected to several hours of systematic questioning about his-usually "his" and not "her"--language and customs.6 Such research is still carried out in modern institutional settings, where subordinates, be they prisoners, students, enlisted men, or mental patients, are brought to the office of the researcher for question- ing about the customs and jargon of their fellows. Such fieldwork is characterized by the relatively high perceived power and authoritativeness of ethnographers, comparatively high control of the setting (although not necessarily of the context of research), and a somewhat one-sided interaction flowing from investigators toward those studied.7

Next is the noblesse-oblige model, exemplified by the field-worker who moves into a substantial residence with a body of retainers, and comes to play in local society the role of a wealthy patron. This role was perhaps easier to assume during the imperialist era of the European presence, but it can still be played, and is, in fact, sometimes expected, by host peoples in many areas of the world. It works most easily when the field-worker can establish a peer relationship with the local elite,s and indeed in many such situations it is difficult for a relatively wealthy higher-status researcher not to be cast in such a role. The field-worker, here, has high-to-medium perceived power and some control of the research setting, though none over the context, with interaction flowing in both direc- tions.

In the next model on the fieldwork continuum, going native, the researcher tries in every conceivable fashion to adopt the way of life of the common people being studied: the researcher lives in their residences, eats their food, talks their language, and shares their lives. Such conduct is often difficult, because the researcher is attempting to learn the role of a local and responsible adult - itself a time-consuming task - while also con- ducting research and maintaining field notes and a field diary. This role is handled most easily if the researcher is or can be cast by those studied as a "native."' In this type of

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

34 AMERICAN ANTHROPOLOGIST [82, 1980

fieldwork, the investigator has little or no relative power and little or no control of the set- ting or of the context of research, and interaction flows in two directions. In the next model, the undercover agent, the investigator seeks to expose the real ac-

tivities of groups of persons who are for one reason or another concealing important parts of their lives. This infiltration is spoken of as "penetrating fronts" and "exposing the back-stage."'0 In such fieldwork, undercover agents, through deceit, obtain more power over those studied and more control of the context of the research than is realized by the hosts. It is difficult to characterize the flow of interaction in undercover research, since its course is blocked and distorted by deceit, which causes investigators to spend so much time and energy on "front work" that they tend to perceive those studied, and social life in general, in terms of deception and front work. (I believe that deceptive research is a special case that does not fit smoothly within the spectrum derived from genuine relation- ships between investigators and subjects.) On the far end of the spectrum is the advocate model of fieldwork. Many ethnogra-

phers study peoples who are economically or politically depressed, and there may be a tendency to intervene to help people improve or transform their destinies. The posture of advocacy can range from the liberal (or relatively conservative) approach known as "ap- plied anthropology" to varieties of insurgency. Equally important, if not more so, ad- vocates can range from those scarcely familiar with the hosts to those both familiar and deeply committed. Advocates wish to help those studied gain power; consequently, they frequently attempt to transfer to their hosts their own power and knowledge, or that of the group or milieu from which they come.I" Although they rarely control the research setting, advocates do in some ways shape the contexts of interaction by their definitions of the situations as those where their task is to aid or empower those who are studied. In- teraction here flows in two directions, and, in fact, certain varieties of advocacy research are set up so that those who are studied both define the context of research and control the direction of interaction (by determining the task and goals of research and helping to decide what methods will be used to achieve these [Schensul and Schensul 1978 describe this type of advocacy research]).

APPLYING ETHICAL PRINCIPLES TO FIELDWORK

It becomes clear, when considering the varieties of fieldwork, each with differing rela- tions between investigators and subjects, that different possibilities of harm and benefit are associated with each. I believe, as previously stated, that the majority of harms and benefits of fieldwork are less immediate, measurable, and serious than the harms associated with other research modes. (Thus, in thinking of harm, notorious cases that come to mind are "tearooms" [Humphreys 1970] and Camelot. The first put subjects at risk, causing much comment and disapproval among social scientists, but apparently caused no harm [Warwick 1973]. The second was canceled before it had any chance to put people at risk [Sjoberg 1967; Deitchman 1976]. In the controversial "Springdale" case [Vidich and Bensman 1960], community members were able to identify themselves from ethnographers' descriptions; the greatest harm caused by that violation of privacy, however, appears to have been anger and hurt feelings. [See Diener and Crandall 1978 for a discussion of these and other cases.] If there have been demonstrated instances where fieldwork caused more serious injury, they have not been cited in the literature, leading one to conclude that they have been successfully kept from the public eye.) I am not saying that harms should not be perceived, weighed, and avoided during the conduct of fieldwork, or that benefits not be sought, but only that such are not assessed most ef- fectively by utilitarian calculations.

I have suggested that the ethical adequacy of ethnographic research might be judged

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELD WORK ETHICS 35

more appropriately in the context of respect for individual autonomy based on the funda- mental principle stated by Kant, that persons be treated at all times as ends in themselves, never merely as means. In applying that principle, it becomes clear that certain extreme varieties of the veran-

dah model of fieldwork might be considered questionable, even if subjects suffer little or no harm and the research benefits human knowledge. It is fatally easy, when conduct- ing such research, for the researcher to treat subjects primarily as means rather than ends, and to destroy their independence by means of (implicit or explicit) coercion, thus violating the principles of autonomy. (Dworkin [1976] defines autonomy as authenticity plus independence. In this view, anything that threatens authenticity or diminishes in- dependence violates the principle of autonomy.) Because of the asymmetry between sub- jects and the relatively powerful investigators, who control the setting (if not the context) of research, and who also control, to a certain extent, the direction of interaction, a con-

scious and careful effort must be made not to treat subjects as means alone and to respect their autonomy sufficiently that they feel able to decline to enter the research situation or able to leave it at any time. If the categorical imperative were applied to such fieldwork, certain situations might be judged in advance as too coercive. In such cases, it should be possible for field-workers to alter their research designs in order to minimize their per- ceived power and their control of the setting in which research is to take place."2

In examining noblesse-oblige fieldwork, or the going-native model, incidents and rela- tionships have to be weighed on individual bases. In each case, however, the fact that an interaction or procedure would result in few harms would not be sufficient to have it judged ethically adequate. In making such judgments, it is more the quality of the inter- action than the results that must be scrutinized.

In the undercover-agent model of research, subjects are being used primarily as means rather than ends, even though there may be few harms if confidentiality is respected. In addition, in deceptive research the investigator presents an inauthentic self, making the research interaction inauthentic. Consequently, such research violates subjects' autonomy and thus is doubly questionable. This is not meant to characterize the overt, or nondeceptive, field-worker as a species of Boy Scout, who must not tell a lie, or a charac- ter from the myth of George Washington, who cannot. Fieldwork, like friendship, re- quires a number of social lies to keep interaction flowing fairly smoothly. Also, ethnog- raphers frequently become honorary or symbolic members of the community, and friends with many of their hosts, who then tend to "forget" that they are being studied. It would be a rash investigator who, at a critical or confidential moment, said: "Wait a minute - don't forget you're being studied!" Even so, field-workers who use friendships to obtain information without continually reminding informants what they are doing are, in a certain sense, deceiving them. So are researchers who agree with, or at least do not contradict, informants' remarks that they find unacceptable. Such "deceit" is probably necessary in field-work and in social life. I do not believe, however, that it violates the autonomy of those studied or treats them as means rather than ends. Participant obser- vation conducted by an undercover agent does. Covert fieldwork does cause harm: it harms investigators, whose relations with those studied, and indeed their very view of reali- ty, are distorted by their own bad faith. Here, I agree with Franz Boas, who, in a letter to The Nation in 1919, stated that a person who uses science as a cover for spying forfeits the right to be classed as a scientist. Boas was discussing political spying; that can be extend- ed to all ethnographic spying, however "just" or sympathetic one might find an in- dividual research project. Boas pointed out that the essence of a scientist's professional life is "the service of truth"; and scientists who deceive are violating a fundamental rule of science and jeopardizing the validity of their findings. I have pointed out that fieldwork is the only research method in which the interaction between investigators and subjects is

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

36 AMERICAN ANTHROPOLOGIST [82, 1980

included within the research paradigm. The interaction is the method; the ethnographer is the research instrument. Those who engage in deceptive research are using unreliable instruments, altering the course of research interaction, and obtaining distorted data. Consequently, deceptive fieldwork is unsound methodologically as well as morally. The argument has been advanced that certain information cannot be obtained except through deceit. Although that may be so, it is irrelevant. There is no principle holding that all information is equally necessary to science. For example, knowing just what goes on in brief homosexual encounters in public restrooms (Humphreys 1970), or how couples use nude beaches to arrange sexual encounters (Douglas 1976), is no more vital than much of the material found in the National Enquirer. If investigators must deceive to obtain such information, social science will survive very well without it. Should investi- gators be convinced that it is essential to act as undercover agents in order to expose a particular variety of moral, political, or social chicanery, they can do so, as spies or in- vestigative reporters, subject to the funding, ethical imperatives, and professional con- straints that govern those occupations. But they should not be allowed to protect them- selves with the mantle of science. It is unfair to other researchers, causing hostility to social science and foreclosing future investigation when the chicanery is discovered, and it usually is discovered, if only when a study is published (Diener and Crandall 1978:125-126). It is also unscientific, since the investigators are using unreliable instru- ments and altering what they study, without controlling for this fact. In addition, it may bias the findings: the investigator who presents a "false" self is affecting both the ability to judge what is going on and the course of interaction by the need for constant monitoring of one's "act" or "front." Covert fieldworkers are so preoccupied with fooling others that they frequently perceive deceit or "front work" as the essence of the activity being ob- served. (For example, Hilbert [1978], argues that covert research is the essence of partici- pant observation because it teaches the researcher about the central fact of social life, which is "passing" or "impression management.") When the categorical imperative is applied to advocacy research, it becomes clear that

there are certain varieties in which subjects might be used primarily as means rather than ends. In such cases, field-workers are using the research posture to promote predeter- mined goals for those studied, without examining indigenous values, goals, and lifeways. Thus, advocates whose fieldwork is really a species of revolutionary agitation may be violating the Kantian categorical imperative - if the host people are being used primarily as a means to gain the researchers' own revolutionary goals. Such fieldwork might be characterized as a kind of revolutionary "neocolonialism," in which researchers are con- vinced not only that they know more than their hosts about what will benefit the subjects, but also that it is worthwhile to risk the well-being of their hosts in order to chance the making of a revolution. The same critique applies to investigators who wish to keep their hosts "pure" or "untainted" by the modern world, even when those studied are anxious to obtain the techniques, ideas, and goods of that world. Because this posture regards only the values of the investigator, not those of the population, it treats people as means to alien ends and so violates their autonomy. Thus, even if one agrees with a particular field-worker, in a particular effort to ameliorate or improve a group's situation, if the change is not desired by that group, then the effort is ethically questionable. The issue of whether a researcher should intervene when in the field has always been a difficult one - especially since there seems to be no sure way of calculating possible harms and benefits in advance. By concentrating on intent rather than results, the categorical imperative makes it easier to weigh such decisions: they are acceptable when people are treated as ends, not means, and their autonomy is respected. In contrast to research that treats peo- ple as means to the field-workers' ends are advocacy programs, in which investigators work closely with the study population, helping it determine the relevant problems,

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELDWORK ETHICS 37

methods, and goals, and then working with it to achieve the goals (Jacobs 1974: Schensul 1973; Hessler and New 1972). Here, field-workers are trying to increase the autonomy of those studied, thereby treating the people most fully as ends. Thus we might consider such research a paradigm of the ideal relation between investigators and subjects.

APPLYING ETHICAL PRINCIPLES TO FIELDWORK

The principles of respect for human autonomy--regarding people primarily as ends rather than means--may sound soft and indefinite when compared with utilitarian calculations, which seem to offer the possibility of a numerical weighting of ethical ade- quacy. Unfortunately, here we come across the principle that computer experts call "gi- go"-"garbage in, garbage out." Although numbers can be satisfying, they are worth no more than what they measure, and if numbers measure something useless or irrelevant, then they are acting as symbols rather than measures, being merely a way of claiming that something significant is being accomplished. What I am saying is that irrelevant regulations, whether or not they promise to come up with hard numbers - which, in fact, the risk-benefit ratio does not - may have symbolic value but little effect upon the ethical adequacy of research, and indeed may distract attention from real moral problems.

To apply the principle of respect for human autonomy, of using people primarily as ends rather than as means, research projects might be reviewed before investigators go in- to the field and again when they return. For example, a group of experienced field- workers could review ethnographic research plans before fieldwork is initiated. When particular techniques, such as verandah ethnography or undercover-agent research ap- peared questionable, the burden of proof would be on the field-worker, to convince col- leagues that: (a) this is the only possible way to carry out such an investigation; and (b) the research is worth doing. Field-workers could also undergo a debriefing when research is completed. They would be encouraged to examine the relationships established with those studied and discuss the ways in which the data will be presented. The discussions, occurring before and after research, should in themselves encourage reflection and a more earnest and informed attempt at conducting ethical fieldwork. The review commit- tee might also employ sanctions, if necessary, including private and public censure and, in cases of flagrant abuse, refusal to allow publication of findings.

SUMMARY AND CONCLUSIONS

I have discussed human-subject regulations, showing how they are based on the rela- tionship between biomedical experimenters and their subjects. I then examined relation- ships between investigators and subjects in different types of research, indicating how they vary in investigators' perceived power over subjects, in control of the setting and of the context of research, in the direction of research interaction, and in the levels of harm

and benefit. These dimensions were described as continua, with human-subject regula- tions based on utilitarian risk-benefit calculations seen as appropriate (although not necessarily sufficient) for the biomedical end of the spectrum but becoming increasingly less appropriateas investigators' power and control, and the levels of harm and benefit, all diminish. I suggested that fieldwork might be scrutinized more realistically in terms of the principle of respect for human autonomy, based on a regard for people primarily as ends rather than means. Various types of fieldwork were then ranged along a spectrum, again according to the relative power and control of the researcher and the direction of interaction, and this ethical principle was briefly applied to different models of fieldwork.

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

38 AMERICAN ANTHROPOLOGIST [82, 1980

I realize that I am attempting to steer a difficult middle course. On the one hand are those who assert that, because fieldwork consists of talking to people, asking their opin- ions, and making notes on observations and conversations, or on just what ordinary peo- ple do in living their daily lives (Klockars 1978), it is preposterous to worry about ethical problems or try to regulate the activities of the field-worker. On the other hand are the critics who claim that investigators who discuss irrelevant regulations are merely trying to protect their own private research empires from investigation and regulation. It is simplistic to assert that scientists are good unless proved evil, or evil unless proved good. Instead, field-workers, like other scientists, must be recognized as fallible human beings, most of whom would rather benefit than harm those they study, but most of whom might also prefer not to have their activities investigated or interfered with. The ethical practice of fieldwork can and should be improved, but that will not be ac-

complished by facile denunciations or by rote application of irrelevant ethical formulas. To improve the level of fieldwork practice, investigators must examine the moral dilem- mas particular to this type of research, discover the appropriate ethical principles, and learn how best to apply them. If this is not done, regulation will become an elaborate and expensive charade, useful only in assuaging the sensibilities of legislators, who can con- vince themselves that they did their best to legislate morality without ever having both- ered to examine just what moral standards are appropriate to a particular scientific method.

NOTES

Acknowledgments. This essay was produced as part of the activities of the Project on "Ethical Problems of Fieldwork," sponsored by Washington University, Saint Louis, Missouri, and the Center for Policy Research, New York City, and supported by the program in Ethics and Values in Science and Technology of the National Science Foundation. Earlier versions were delivered at the 1978 annual meetings of the Association of Humanist Sociology and the American Anthropological Association. My colleagues on the project, William May and Murray Wax, clarified this paper in substance and style. I would also like to thank Elinor Barber, Eric Cassell, H. Tristram Engelhardt, Jr., Hans Jonas, Nancy McKenzie, and Stephen Toulmin for comments and criticism. Naturally, I accept fall responsibility for the opinions expressed.

1 For example, a report, prepared by a philosopher for the National Commission for the Protec- tion of Human Subjects of Biomedical and Behavioral Research (1978), that examines the basic ethical principles that should guide biomedical and behavioral research involving human beings limits the discussion almost entirely to experimentation, as though "research involving human sub- jects" and "experimentation" were equivalent (Engelhardt 1978).

2 Not all subjects of biomedical experimentation are hospitalized patients (they may be students, prisoners, or even the investigators themselves), and not all biomedical research consists of experi- mentation (there are observational studies as well). In the same way, psychological experiments may be carried out on hospitalized patients. The varieties of research discussed here are ideal types, and actual research projects may differ from them in many ways. The point is that the greater the per- ceived power and control of investigators, and the more unidirectional the research interaction, the closer a research project is to "the biomedical model" upon which human-subject regulations are based.

3 May (1978) notes that the terms risk and benefit are asymmetrical. Risk implies the mere possibility or probability of harm, while benefit seems to describe payoffs that are virtually certain. He suggests that it might be more accurate to weigh risk against hope or harm against benefit.

4 Even with experiments that prove in retrospect to have been extremely dangerous, it is fre- quently difficult to assess the harms until a much later time. Consider the notorious "Harvard drug controversy," in which student and experimenters took psychotropic drugs together in 1960. An early analysis (Benson and Smith 1967) discussed with evenhandedness the ethical and political

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELDWORK ETHICS 39

claims and counterclaims of experimenters, critics, and the Harvard administration. There was no mention of the long-range serious consequences that were apparent 18 years later in those who took the drugs. Because research is a venture into the unknown, it is difficult to see how regulations can guard against unexpected and delayed harms, such as the long-range consequences of drugs such as diethylstilbestrol, which causes vaginal cancer in the daughters of women who receive it during pregnancy.

s Interestingly, some of the most serious sources of harm are various legal and regulatory agen- cies, which have the power to threaten imprisonment in order to force researchers to divulge their data in violation of assurances of confidentiality given to participants (Bond 1978; Seiler and Mur- tha 1979).

6 I understand that this model was utilized by Radcliffe-Brown, who questioned Andaman Island prisoners sent to him from the British gaol.

7 When the perceived power of the field-worker is said to be relatively high in the verandah model, that is in comparison with other varieties of fieldwork. The spectrum of relations between field-workers and informants is distinct from that of the relations between investigators and sub- jects. If the two spectra were superimposed, I suspect the verandah model would be located some- where near survey research.

8 This is the way Raymond Firth studied the Tikopia (1957).

9 Colin Turnbull followed this going-native model in his work among the Pygmies of the Ituri rain forest (1962). When I studied the contemporary American Women's Movement, I was a native, working with urban, primarily white, middle-class women like myself (Cassell 1977).

o10 This is an approach that would be natural to radicals, for example, who wish to expose "the establishment." The most eloquent exposition, however, has been written by a nonradical, Douglas, who has "exposed" what goes on at sites such as massage parlors and nude beaches (1976).

SThus, in the Vicos project, sponsored by Cornell University, where Alan Holmberg, an anthro- pologist, became patron of a hacienda in Peru, with almost absolute power over the residents, a "process of power devolution" was initiated, during which the residents learned to become economically and politically self-sufficient (Holmberg 1971:47).

12 For example, Goffman, when studying a mental hospital, took precautions to avoid sociable contact with the staff and the carrying of a key (1961:ix); and Giallombardo, when studying a women's prison, took similar precautions to avoid being identified as a staff member (1966:191-223). In each case, the researchers interacted with those studied on their own territory, following as much as possible their schedules, rather than having subjects brought to a special place to be interrogated.

REFERENCES CITED

Benson, J. Kenneth, and James Otis Smith 1967 The Harvard Drug Controversy: A Case Study of Subject Manipulation and Social Struc-

ture. In Ethics, Politics and Social Research. Gideon Sjoberg, ed. pp. 115-140. Cambridge, Mass.: Schenkman Publishing Co.

Boas, Franz

1919 Scientists as Spies. The Nation 109 (#2842):797. Bond, Kathleen

1978 Confidentiality and the Protection of Human Subjects in Social Science Research: A Report on Recent Developments. The American Sociologist 13:144-152.

Cassell, Joan 1977 A Group Called Women: Sisterhood and Symbolism in the Feminist Movement. New

York: David McKay Co. 1978 Risk and Benefit to Subjects of Fieldwork. The American Sociologist 13:134-143.

Code of Federal Regulations 1977 Protection of Human Subjects. 45 CFR 46, revised as of April 1, 1977.

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

40 AMERICAN ANTHROPOLOGIST [82, 1980

Deitchman, Seymour J. 1976 The Best-Laid Schemes: A Tale of Social Research and Bureaucracy. Cambridge: MIT

Press.

Diener, Edward, and Rick Crandall

1978 Ethics in Social and Behavioral Research. Chicago: University of Chicago Press. Douglas, Jack

1976 Investigative Social Research. Beverly Hills: Sage Publishing. Dworkin, Gerald

1976 Autonomy and Behavioral Control. The Hastings Center Report 6(1):23-28. Engelhardt, H. Tristram, Jr.

1978 Basic Ethical Principles in the Conduct of Biomedical and Behavioral Research Involving Human Subjects. The Belmont Report: Ethical Principles and Guidelines for the Protection of Human Subjects of Research. Appendix Vol. 1. DHEW Publication No. (OS)78-0013. Government Printing Office: Washington DC.

Firth, Raymond 1957 We, the Tikopia. 2nd ed. London: George Allen & Unwin.

Giallombardo, Rose

1966 Society of Women: A Study of a Women's Prison. New York: John Wiley & Sons. Goffman, Erving

1961 Asylums: Essays on the Social Situation of Mental Patients and Other Inmates. New York: Doubleday Anchor Books.

Gray, Bradford H. 1975 Human Subjects in Medical Experimentation: A Sociological Study of the Conduct and

Regulation of Clinical Research. New York: John Wiley & Sons. Hessler, Richard M., and Peter Kong-Ming New

1972 Toward a Research Commune? Human Organization 31:449-451. Hilbert, Richard A.

1978 Learning From Agnes: Notes on Covert Participant Observation. Paper presented at Conference of the Association for Humanist Sociology, South Bend, Indiana, October 1978.

Holmberg, Alan 1971 The Role of Power in Changing Values and Institutions of Vicos. In Peasants, Power

and Applied Social Change. H. F. Dobyns, et al., eds. Beverly Hills: Sage Publishing. Humphreys, Laud

1970 Tearoom Trade: Impersonal Sex in Public Places. Chicago: Aldine Publishing Company. Jacobs, Sue-Ellen

1974 Doing It Our Way and Mostly for Our Own. Human Organization 33:380-382. Kant, Immanuel

1959 Foundations of the Metaphysics of Morals. Indianapolis: Bobbs-Merrill Co. (Originally published in 1785).

Kelman, Herbert C.

1968 A Time to Speak: On Human Values and Social Research. San Francisco: Jossey-Bass Inc. 1972 The Rights of the Subject in Social Research: An Analysis in Terms of Relative Power

and Legitimacy. American Psychologist 27:989-1,016. Klockars, Carl B.

1978 Comments on papers by Bond, Cassell and Stephenson. The American Sociologist 13: 164-165.

May, William F. 1978 The Right to Know and the Right to Create. Newsletter of Science, Technology and

Human Values. Spring issue. Mead, Margaret

1969 Research with Human Beings: A Model Derived from Anthropological Field Practice. Daedalus 98:361-386.

Milgram, Stanley 1963 Behavioral Study of Obedience. Journal of Abnormal and Social Psychology 67:371-378. 1965 Some Conditions of Obedience and Disobedience to Authority. Human Relations 18:

57-76.

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms

Cassell] FIELD WORK ETHICS 41

National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research 1978 Report and Recommendations: Institutional Review Boards. DHEW Publications No. (OS) 78-0008. U.S. Government Printing Office, Washington, DC.

Olesen, Virginia 1979 Federal Regulations, Qualitative Social Science and Institutional Review Boards: Com- ments on a Problematic Era. In Federal Regulations: Ethical Issues and Social Research. Mur- ray L. Wax and Joan Cassell, eds. Boulder, Colo.: Westview Press.

Orne, Martin T.

1962 On the Social Psychology of the Psychological Experiment: With Particular Reference to Demand Characteristics and Their Implications. American Psychologist 17:776-783.

Orne, Martin T., and Karl E. Scheibe

1964 Inadvertent Termination of Hypnotized and Simulating Subjects. International Journal of Clinical and Experimental Hypnosis 14:61-78.

Rynkiewich, Michael, and James P. Spradley 1976 Ethics and Anthropology: Dilemmas in Fieldwork. New York: John Wiley & Sons.

Schensul, Stephen L. 1973 Action Research: The J. Applied Anthropologist in a Community Mental Health Pro-

gram. In Anthropology Beyond the University. Alden Redfield, ed. Athens, Ga.: Southern An- thropology Society. (distributed by University of Georgia Press).

Schensul, Stephen L., and Jean J. Schensul 1978 Advocacy and Applied Anthropology. In Social Scientists as Adovcates. Beverly Hills:

Sage Publishing Co. Seiler, Lauren, and James M. Murtha

1979 Federal Regulation of Social Research: Is "Prior Review" Posing a Threat to Academic Freedom? Freedom at Issue 53:1-5.

Sjoberg, Gideon 1967 Project Camelot: Selected Reactions and Personal Reflections. In Ethics, Politics and

Social Research. Gideon Sjoberg, ed. pp. 141-161. Cambridge, Mass.: Schenkman Publishing Co.

Turnbull, Colin

1962 The Forest People: A Study of the Pygmies of the Congo: New York: Anchor Press. Vaughan, Ted R., and Gideon Sjoberg

1978 Comments on papers by Bond, Cassell and Stephenson. The American Sociologist 13: 171-172.

Vidich, Arthur J., and Joseph Bensman 1960 Small Town in Mass Society. Garden City, N.J.: Doubleday Publishing Co.

Warwick, Donald P.

1973 Tearoom Trade: Means and Ends in Social Research. The Hastings Center Studies 1(1): 27-38.

Wax, Murray L. 1977 On Fieldworkers and Those Exposed to Fieldwork: Federal Regulations and Moral Issues.

Human Organization 36:321-328. Wax, Murray L., and Joan Cassell

1979 Fieldwork, Ethics, and Politics: The Wider Context. In Federal Regulations: Ethical Issues and Social Research. Murray L. Wax andJoan Cassell, eds. pp. 85-102. Boulder, Colo.: Westview Press.

Wax, Rosalie H.

1971 Doing Fieldwork: Warnings and Advice. Chicago: University of Chicago Press.

Submitted 28 December 1978

Accepted 2 April 1979 Final revisions received 11 April 1979

This content downloaded from 199.73.44.216 on Thu, 17 May 2018 06:00:29 UTC All use subject to http://about.jstor.org/terms