Remembering Erving Goffman
"Medical Sociology: Some Tensions Among Theory, Method, and Substance," Journal of Health and Social Behavior 1989, Vol. 30 (June):147-160
In 1961 Erving Goffman published Asylums, a series of essays based on his fieldwork at St. Elizabeths Hospital, a federal institution housing some 7,000 patients in Washington, DC (Goffman 1961). The book, issued as a paperback by Anchor Books, was enormously popular and was widely cited. Goffman's essays, in association with many the delegitimation of the mental hospital and contributed to a growing movement in the 1960s to deinstitutionalize the mentally ill. Along with Kesey's One Flew over theCuckoo's Nest and the writings of Thomas Szasz, Goffman helped portray mental hospitalization as a humiliating experience that stripped individuals of their identity and self-esteem and induced a variety of deviant adaptations that were reactions to institutional life. This critique of the mental hospital, embodied within a creative and intriguing theory of total institutions and self-identity, captured the imagination of much of the informed public well beyond the sociological community.
It is unlikely that events would have been much different had Goffman not written Asylums. Deinstitutionalization was a product of many influences that extended far beyond the critique of mental hospital care (Mechanic 1989). A democratic and preventive care ideology evolved out of experience during the Second World War, advocating early treatment in the community and a rejection of coercive interventions. The introduction of neuroleptic drugs in the middle 1950s not only allowed control of the most bizarre and disconcerting symptoms of psychosis but also gave administrators and families confidence that patients could be managed in the community. Mental health litigation and the attack on involuntary civil commitment made it more difficult to use coercion to keep patients in hospitals. The expansion of the welfare state in the middle 1960s-particularly Medicaid, disability insurance, and housing assistance-made it possible to relocate patients from mental hospitals to the community and to develop alternatives to mental hospitalization for new patients. Moreover, these new federal programs provided states with strong incentives to relocate patients to community settings so as to shift costs from the state to federal programs. Within this confluence of forces, the delegitimation of hospital care was probably important but not crucial. Thus Goffman's work had only a subsidiary role in the massive changes in mental health policy, but it was as influential as any theoretical statement or study can hope to be. It influenced the way many people perceived mental hospitals and contributed to the overall climate of opinion essential to social change.
Quite apart from Goffman's influence is the issue of the accuracy of his critique. Numerous studies have been done relevant to Goffman's depiction of the experience of the mental patient, using patient surveys. None of these studies has the theoretical brilliance of Goffman's work or the quality of his insight, but they consistently fail to replicate his view of the patient's experience (Linn 1968a, 1968b; Weinstein 1979, 1983). Most patients did not report feeling betrayed; many reported being helped by hospitalization, and viewed the hospital as a refuge from impossible problems and stresses. Moreover, some patients from disadvantaged backgrounds viewed the hospital experience as less coercive and less depriving than their usual life situation. The studies do provide evidence of stigma associated with mental illness but negate the profoundly negative conception of the experience depicted by Goffman.
The issue, however, is not simply one of deciding whether the studies based on patient interviews and questionnaires support or disconfirm Goffman. It becomes necessary to inquire more deeply whether this type of evidence invalidates the theoretical "ideal type" of mental hospitalization that Goffman developed. It seems clear that there are strong arguments on both sides; Goffman and his critics may be plumbing different levels and types of meaning with their respective methods.
As an observer, Goffman brought to the hospital his own personal biography and assumptions, which shaped how he saw events. To a middle-class, independent-minded professor, who strongly valued personal autonomy and the right to be eccentric, the regimentation of the mental hospital must have looked repressive indeed. Later in Goffman's life, after he had to live through an episode of mental illness involving another person close to him, he is said to have remarked that had he been writing Asylums at that point, it would have been a very different book.
The qualitative observer is, in a sense, a research instrument, and how he or she is
calibrated is an issue of some importance. I began my hospital fieldwork in the middle
1950s shortly after the introduction of neuroleptic drugs. Although I had visited some
mental institutions during my college years, my first intensive exposure was at Agnew
State Hospital in California in 1956. In 1961, when I initiated the medical sociology and
mental health training program at the University of Wisconsin, I required that trainees
spend a semester as visiting interns at Mendota State Hospital, a state mental hospital. It was clear to me that mental illness and the professional care of the mentally ill were a longitudinal experience, and that short cross-sectional views could be highly misleading. If students wanted to do serious work in the area, they had to become familiar with the longitudinal experiences of patients and the types of challenges they posed to staff. When students first went to the state hospital, they often came back convinced that many of the patients were not mentally ill and were inappropriately hospitalized. As they spent extended periods in the hospital, they usually
learned that "mental conditions" were more complex than they initially believed, and that
patients who appeared normal at one point in time could be highly disturbed on other
Beginning with my fieldwork in hospitals in 1956, I was impressed by the dramatic
improvement in conditions in these institutions during the next two decades. As patient
populations decreased, as staff ratios improved, and as hospitals substantially increased
their treatment and rehabilitation programs, the mental hospital became an entirely different type of institution than it was in the 1950s. Yet as each subsequent cohort of graduate students took part in their internship experience at the state hospital, they returned with anecdotes documenting
what a repressive place it was, consistent with their theoretical dispositions favorable to
labeling theory. Not having any basis for comparison with mental hospital care in prior
periods, they were calibrated as observers somewhat differently from me.
This point suggests the difficult question of how qualitative researchers, using themselves as a research instrument, properly calibrate themselves. What the researcher selects as important from an almost infinite number of possible observations, and how each observation is weighed relative to others in a larger organizing frame of reference, determine to a major extent the construction that emerges. The way such conceptualization and organization of observations take place depends substantially on the social biography of the observer.
This would seem to suggest the superiority of standardized questions or observations elicited from a representative target group, but this approach also has serious limitations. Any experienced researcher knows that how one frames a question and selects response categories substantially affects the answers received. Respondents to such questions may have difficulties in recall, may wish to withhold important information, or may wish to present themselves in a particular way contrary to the facts (Marquis and Cannell 1971; Mechanic and Newton 1965). The responses may also depend on the order and format of questioning, the timing of questioning
relative to the event being asked about, and the respondent's involvement in the interviewing process. In many surveys, respondents are only superficially involved, are poorly instructed or not particularly motivated to recall accurately, and may respond at a level that fails to reveal deeper feelings and concerns. Thus, although the survey method may seem more objective on the surface, it confronts problems comparable in importance to those characteristic of the qualitative
None of this is to suggest that good practitioners of both types of methods are not aware of these problems and do not make serious efforts to take them into account in the way they design studies or collect relevant data. We have many relevant methodological studies, and some solutions, to such problems as acquiescence, social desirability effects, format biases, recall difficulties, and many more, but the complexity of these issues means that in the average study most of these cautions are acknowledged but ignored.
In Goffman's essay on the moral career of the mental patient he depicts in a sequential form how the understandings of the prepatient are likely to be betrayed through the process of referral and admission. Goffman suggests that the prepatient passes through a "betrayal funnel" with at least a portion of the rights, liberties, and satisfactions of the civilian and ends up on a psychiatric ward stripped of almost everything (Goffman 1961, p. 140). In efforts to examine Goffman's assertions concretely, Linn (1968b) asked 185 patients, most within 48 hours of their hospital
admission, questions specifically addressed to Goffman's observations. For example, when
Linn asked patients "if they felt they have been betrayed by any of their friends or family in coming to the hospital," 75 percent of the patients denied feeling betrayed.
Linn views the discrepancy between his findings and those of Goffman as resulting primarily from sampling problems and the tendency of observers to give undue influence to more interesting, articulate, and novel responses. He acknowledges that his own results based on direct questions to patients could have been shaped by defensive adaptations to their circumstances, but rejects this idea as an adequate explanation of the highly divergent findings. Then again, it is not clear how one establishes how influential defensive adaptive needs may be in shaping responses.
In any case, the issues cut more deeply than this comparison suggests.
In an intriguing review of the effects of diet on psychological state, Wadden, Stunkard, and Smoller (1986) found that none of the existing 16 studies which measured psychological
state in a pre-post test design using standardized measures showed an adverse emotional effects of treatment, while all eight studies that used clinical measures reported adverse reactions. Unlike the standardized approach, the clinical observations, although unsystematic, were made at varying points in the treatment process and took account of a wide range of emotional distress. The standardized measures, in contrast, were more specific, were administered less frequently,
and covered a narrower range of reactions. The authors conclude that the "method of assessment shapes reports of mood changes during weight reduction and provides perhaps the single best explanation of the contradictory findings concerning dieting and depression" (p. 432). After examining these very different patterns of results, and the relation of results to methods, they suggest that these alternative modes of assessment actually address different questions. Studies depending on a pre-post test design assess the effects of weight loss on mood, while continuous
measurement speaks to the impact of dieting on mood.
* * *
Goffman's observations appear credible despite disconfirmation by surveys, because readers of his analysis find his depiction meaningful and convincing when they view themselves as the hypothetical patient in the context he describes. Thus Goffman conveys a certain kind of "truth"
that cannot be dismissed easily. This type of contextual credibility is often persuasive,
having the quality of verstehen embodied in the methodology of Max Weber (1 949).
Moreover, contextual analysis can be adapted in a stastandard and rigorous way to examine
meaning, an issue which will be examined later.