Producing Ethnographies: Workplace Ethnographies in History

Abstract

Data on a large set of workplace ethnographies published from 1940 to 2002, compiled by Randy Hodson, are analyzed to show the trends over time in the production of such ethnographic work, its shifting disciplinary base, the relevance of the personal backgrounds of its authors, the contributions made by academic amateurs, the changing roles of gender and political stances, and the nature of different routes to publication. The definition of what counts as an ethnography is important to the character of the set available and has implications for its potential uses in secondary analysis. It is found that both personal and disciplinary identities and wider social factors have played roles in the production of ethnographic work that need to be understood to account for its history, though it is to be expected that the forms these take will differ for work in different subfields.

This article takes advantage of the existence of a large collection of ethnographies, bearing on one substantive area, to investigate the historical circumstances under which such ethnographic work has emerged and to consider the implications for the history of research methods in social science, and sociology in particular. The collection utilized, of workplace ethnographies, was compiled by sociologist Randy Hodson; his aim was not to contribute to the history of methods but to make secondary use of these studies to combine the strength of field observation with quantitative analysis in the study of workplaces.1 His set, thus, provides a novel attempt, of some historical interest in itself, to straddle the quantitative/qualitative divide. One cannot assume that ethnographies of workplaces constitute a representative sample of ethnographies more generally, but they are the product of general as well as subfield-specific processes; this paper makes the most of an unusual and valuable resource originally created for other purposes. It became evident that there were historical trends in the total production of workplace ethnographies, and in what kinds of ethnography had been produced by which kinds of ethnographer; the prevalence of such ethnographies was related to factors such as gender, politics, discipline, and the publication system. In addition, the decisions needed about which items to include in the set raise issues about the social processes involved in the production of different forms of empirical work. We begin with a brief overview of ethnographic methods, before introducing Hodson's set and discussing the various kinds of work that it takes to be ethnographic. We explore the characteristics of the authors and the workplaces studied, the disciplinary affiliations and political intents (if any) of the ethnographies included, and the role of publication outlets in shaping the data set. We conclude by connecting the history of workplace ethnography to that of social research methods more generally.

The Ethnographic Method

The popularity of ethnography as a method of data collection in social science has fluctuated, and its use has varied between fields. “Ethnography” as a term used to be specific to anthropology, but in recent years has become widely used to describe practices within sociology (Bryman, 2001, pp. 290–291), which had often previously gone under the heading of “participant observation.” Paul Atkinson and Martyn Hammersley, who are among its leading contemporary proponents, offer a summary description of the practices involved:

“Ethnography usually refers to forms of social research having a substantial number of the following features:

• A strong emphasis on exploring the nature of particular social phenomena, rather than setting out to test hypotheses about them
• A tendency to work primarily with “unstructured” data, that is, data that have not been coded at the point of data collection in terms of a closed set of analytic categories
• Investigation of a small number of cases, perhaps just one case, in detail
• Analysis of data that involves explicit interpretation of the meanings and functions of human actions … with quantification and statistical analysis playing a subordinate role at most.” (1994, p. 248)

Elsewhere, the same authors get closer to depicting the daily research practice of ethnographers when they say that

the ethnographer participates, overtly or covertly, in people's daily lives for an extended period of time, watching what happens, listening to what is said, asking questions, in fact collecting whatever data are available to throw light on the issue with which he or she is concerned. (Hammersley & Atkinson, 1983, p. 2)

The emphases on intensive personal involvement in naturally occurring situations, and on nonquantitative methods, have traditionally been crucial. (The operational definition used by Hodson is described below.) There is a long tradition in sociology of making quantitative vs. qualitative a key distinction among methods, although the specific styles involved have changed over time, with “case study” and then “participant observation” as predecessors of “ethnography” (Platt, 1996, pp. 11–66); in the post-war twentieth-century context, the understood “other” has been a survey method. It is widely suggested that there has been an ethnographic turn within recent sociology, though Culyba, Heimer, and Petty (2004) offer systematic data that show that, despite common assumptions, this “turn” appears to owe more to changing terminology than to a reorientation of research practice. Moreover, there appears to be a fairly widespread consensus in the sociology of work on the occurrence of change over time, from a prevalence of in-depth case studies of worker experience to their decline and marginalization, despite some national and disciplinary variation.2 These issues have been considerably discussed, in relation to various countries and disciplines; see, for instance, Frege (2005), Marginson (1998), and Whitfield and Strauss (2000). An explanation suggested for that trend is the increasing secondary use of the large survey data sets now available, which could not show the nuances and complexities of worker experience; changes in managerial practices and theories were seen as more important—and it became assumed that these were successful in suppressing worker dissent.3 Against that background, we look closely at the sources and character of the ethnographers and ethnographies in this subfield.

Hodson's Set

Hodson's set was constructed to include every book length workplace ethnography found that met certain criteria. The criteria were that the books (published from 1940 onward) be in English, report direct observation for at least 6 months, focus on a single organizational setting and one clearly identified group of workers, and provide relatively full information on the organization, the nature of the work, and employees’ attitudes and behavior.4 (In practice some trade-off between criteria was allowed, with strength on one treated as compensating for weakness on another [Hodson et al., 1993, p. 403], so some cases not meeting all the criteria were included.) Some of the books also included survey material. By 2007, a total of 204 cases of workplaces, from 158 books, had been coded. The aim was to include every book that met the criteria. Despite impressive efforts, it now appears that this was not fully achieved.5 It seems reasonable, nonetheless, to treat the set as sufficiently representative for our purposes of what it aimed to cover.

Each case was coded by Hodson's team on a range of workplace issues and a few characteristics of the project and its authors. The coded data have, with exemplary collegiality, been placed in the public domain.6 It is clear that coding categories such as those used risk not quite capturing the distinctive complexity of the studies coded, though there are corresponding advantages in the possibility of quantification. The assertion of ethnography's distinctiveness is most pronounced among researchers concerned with agency, and workplace ethnographies have special import because workplaces create contexts for specific kinds of instrumentally rational action, even if instrumentality does not always prevail. These ethnographies thus lend themselves to comparison not just between workplaces but also between those who study them.

In what follows, we build on Hodson's work to advance a contextual understanding of the production of ethnographic texts. Where for him the books provide empirical cases that can be added up to provide or support general conclusions about workplaces (see Hodson, 1999, 2001, in particular), for us the historical circumstances of their production, and the contexts in which they were published, are central. His set of cases can, independently of its compiler's intentions, be treated as one of ethnographers or of ethnographies rather than of workplaces, and that is what we do here. We have drawn on multiple additional sources to add to his data set some additional features of book and author characteristics that were potentially relevant; these include the authors’ disciplinary affiliations (if any), the department of their highest degree, their declared political positions in relation to the work,7 the specific occupations that were studied, and the ethnographic roles played. Our analysis explores quantitative and qualitative data on the connections among such factors and the published outcomes. There is scope for critical comment on some aspects of the procedure used for the purpose Hodson intended,8 but we are not concerned with that here. Our focus is on the historical processes by which it came about that a certain range of material was produced and thus became available for such secondary analysis.

First, however, we need to examine the criteria used to define this special set. Both the operational definition of “ethnography” and the focus on workplaces affect which cases are included in the data and that has consequences for the interpretation of the findings; we explore these.

What Was Counted As an Ethnography?

Some of those books would normally be described as autobiography, oral history, or journalism, but the data presented were nonetheless rich enough, and close enough to the direct experience of the workplace, to be judged fit for inclusion in Hodson's set. Some of the authors had not set out to do research using ethnographic methods, however those are defined. The work of these unintentional ethnographers challenges the boundaries of “ethnography,” including instances where:

• the author took a job because he or she needed the money, or wanted personal experience of manual work, without originally intending to write about it;
• a professional writer wrote about the subjects’ work experience, without exposing themselves to it, on the basis of what they told him or her;
• an oral historian or folklorist (sometimes an amateur one) collected the memories of participants in past work situations;
• a worker late in his career, commonly a (male) union activist, wrote a memoir of his experience; and
• the author, who may or may not have had some social scientific training, wrote from a managerial position and perspective about the workers he or she managed.

A few authors who did not set out to conduct an ethnography were aware of the genre and ended by adopting that model, but more of them remain obdurately open only to retrospective cooption as ethnographers—Messieurs Jourdain9 of social science method. Sometimes, too, the ethnographer was not the author, or even the author's research assistant. Thus, the ethnographic label was stretched in novel ways. Maybe it was the planned use of the material that led to the inclusion of some rather marginal cases in order to increase the total, and so the scope for quantitative analysis.

In what senses can all the cases be treated as of the same methodological kind? Cases as diverse as oral history material collected from local fishermen by a retired geography teacher, the memoirs of a union-activist steelworker, data acquired by a business school professor whose research staff observed factory workers, and a feminist sociology graduate student's interviews with strippers plus her own experience working as one, are each treated as one case with the same weight and methodological credentials. (We note, however, that analogous assumptions are routinely made about individual respondents in survey research.) The issue may not be problematic if one is looking for data about workplaces—though we may also note that not every author was focusing on the setting's characteristics as such; for some the theoretical interest was in economic development, community, crime, the medical care received by patients, gender roles, or stratification. But such variations make conclusions about the history of the use of ethnographic method more complex, as we discuss further below.

We do not attempt to decide which studies were, by other criteria, “really” ethnographic; any such attempt would require arbitrary decisions,10 and borderline cases are of considerable interest precisely because of their borderline status. It might, however, be more precise to describe the eventual set as quasi-ethnographic, one of cases with data of ethnographic scope or depth, however, that was achieved. The diversity of the relation of the authors both to academic ethnography and to social science urges the relevance of looking more closely at their identities.

Who Were the Authors?

The solid majority of the authors were men for books published before the 1980s, and women for books published from the 1980s onward; 83 (53 percent) of the ethnographers were sociologists; the second largest group was 26 (16 percent) anthropologists. Doctoral candidates, especially women, made a major contribution to this literature; from the 1970s onward, half or more of the total books by women were based on doctoral theses; the trend over time in the doctoral proportion was upward for women, downward for men.

The decision to use an ethnographic approach may be presumed to have been taken when fieldwork started—for those for whom it was a decision. At that stage, these ethnographies were largely a phenomenon of the 1970s and 1980s, for both the social scientists and the “amateurs”; the only amateurs whose work was not started or published in the 1970s were two who wrote from a managerial point of view. Looking instead at publication date, which also entails decisions made by publishers, the bulge for the majority writing on manual workers occurs from the mid-1970s to the early 1980s. Some of the historical features of these distinctions are developed further below, but more personal and less demographic identities were also relevant and cut across these distinctions; we start with those.

Authors approached the “ethnographic” role from different starting points, and even the professional social scientists certainly did not all fit the stereotype of the career academic anthropologist or sociologist approaching research from a theoretical starting point. Some had already held a job of the kind studied, or describe a family background in which it was typical. Eight collected data while continuing a job they already held, or retrospectively decided to use their experience as data. Of those who took a job for research purposes, a third say they had held the same type of job before, while five more reported some family background—usually of manual workers active in trade unions—which made the setting relatively familiar. It looks as though the rationale for writing at all, for choosing what kind of workplace to study, or for choosing how to collect data, was sometimes based on prior personal experience as much as on social scientific theorizing. It might, thus, be fruitful to treat these cases as examples of broader forms of autobiographical writing, not expressed as social science, and the social circumstances generating them. But autobiographical motives were not essential to such choices; practical convenience of access, the advantage of general knowledge of the setting, or a conviction that what they knew about from their own experience was inadequately represented in the literature could all point in the same direction.

In addition, some of the authors were, evidently, amateurs as social scientists. The “amateurs,” defined as those without some academic social scientific or related training, or holding academic roles, are divided fairly evenly between workers writing from personal experience and writers on the experience of others—or, in the cases of Haraszti (1978) Kamata (1982) and Palm (1977) their own experience undertaken by writers in order to see what it was like. Sharp differences between the amateurs and the professionals might be expected,11 but the personal identities of a number of the individual authors were not nearly so straightforwardly divided between those categories, as a closer look reveals. It is obvious that some of the amateur authors were far from typical manual workers, even if writing of their experience as such. Thus steelworker Spencer (1977, pp. 3–4) writes of his work:

…this book departs from the surveys, interviews, polls, sociological studies, union profiles, political projections and other eye-catching comments being presented to satisfy the current public curiosity about work and the blue collar worker. The approach here is that any and all definitions, symbols, portrayals, statements, examples, tables and theories of blue collar workers, must be tested out on the proving grounds of the workplace…. This book is not a “study” in the sense that observations, reports and documents are systematically and precisely labelled, numbered, and summarized following the method of the conventional university scholar….

We have not been able to find out more about his life, but clearly he knows something about the surveys etc. that he rejects with such verve and rhetorical fluency! Perhaps such authors should not be regarded as entirely accidental ethnographers. It is, at any rate, clearly much easier for the amateur to write his own life story than it is to produce a “study” as described by Spencer; there is no literature of working men's questionnaire reports, so those amateur contributions are likely to cluster in the quasi-ethnographic area.

The book style more generally ranges from that of the boringly recycled dissertation to one that the foreword by popular American leftist Michael Moore describes as:

a masterpiece of writing, which paints a darkly humorous account of working class life in America … it should confirm [the author's] place in publishing history as the first author to write a major book while under medical supervision—and the influence of eleven different miracle drugs…. Of course, to read what he has to say in these pages, you may wonder who the real crazy ones are…. (Hamper, 1991, p. xiv)

Hamper, an autoworker, had over a period been drawn by contact with Moore into a secondary career of political writing.

Equally, some of the nonamateur authors’ careers had been far from a straightforward progression from graduate school in their twenties to a professorship on receipt of the PhD, and some have maintained another identity before or alongside an academic identity. Thus Juravich started out as a mechanic, but got a PhD in 1984 and became a professor of labor studies—while also maintaining a career as labor folk singer; Applebaum was actively involved with professional anthropology from middle age on, taking a PhD degree in 1979, editing the Anthropology of Work Newsletter for a number of years and teaching college courses in social science part-time—while continuing throughout with the day job in construction that he started in 1956.12 Another kind of career that figures occasionally was in consultancy or the higher journalism; two books on modern technology-based financial businesses (Kidder, 1981; Stross, 2001) exemplify this, with authors who seem to move in the same worlds as their subjects as much as in academia.

Thus, the boundaries between membership categories could be blurred in practice, in ways that the traditional problems of achieving rapport and the risk of “going native” do nothing to describe. Even for those holding academic jobs, fieldwork participation could feel more like being a native already:

My “true self” includes a quite visceral sense of being a truck driver as well as a sociologist. (Ouellet, 1994, p. 17)13

Since I had had considerable experience of temporary [clerical] employment…, the problem was not so much one of coming to terms with my subjects’ world, but rather one of how to reappraise this world sociologically. (McNally, 1979, p. 192)

If one were interested only in the history of academic social science, arguably those who remained amateurs would have to be omitted—except, perhaps, to the extent, at least, that they have been retrospectively absorbed into the social science literature? Hodson's publications on this set of cases have in effect absorbed them, but is that just one idiosyncratic use? An impression of the extent to which the “amateur” works have been absorbed into social science has been gathered by looking at the first five pages found on some examples in Google Scholar.14 This somewhat crude method shows results impressionistically quite suggestive about the range of ways in which such materials may be incorporated and the factors that can influence the level of attention received. Thus, many references to Hamper (1991) were reported, no doubt helped by both the popularity of autoworkers as a topic and the sponsorship of Michael Moore, but none of those were from anthropology or sociology journals. The largest numbers came from labor history, and from legal scholars, sometimes in connection with issues of drinking in the workplace. There was also evidence of the book being used in teaching of rhetoric/creative writing/literacy, and as offering data on gender identities. Cherry's (1974) book on ironworkers similarly received a few social science references, but with an apparent tendency to treat his book as consciousness-raising educational material for undergraduates rather than as research findings or theorization to build on. Social science references to Mers, (1988) an active trade unionist among longshoremen, were mostly from sources dealing with class and trade unionism. On the other hand, two amateurs who became professional anthropologists, Applebaum and Gamst, received more research-type anthropological citations, as well as ones related to hobbyist and other kinds of interest in their areas of work, construction and railroads. Thus, we can see that these studies were used in academic social science in different ways and for a variety of purposes, some of them central and some entirely accidental in relation to the authors’ concerns. This in itself, however, does not distinguish them from conventional academic studies, also sometimes referred to in surprising contexts, and they have probably been used quite as much as many empirical studies by professors. Whether or not one treats such authors’ books as just data for sociology or social science, or as part of it, they have played a role in its history.

Occupational Coverage

If we see this as a sample drawn from the population of workplaces, the question of how representative it is arises and has consequences for the way in which it can be used.

Whether or not the studies arose from previous personal experience, all the accounts were created by authors for their own reasons, not as parts of a larger plan of occupational coverage; thus, uneven representation in them of the range of occupational possibilities is hardly surprising. Many of the authors say that they wanted to give a voice to “their” workers, otherwise unrepresented in the literature; this suggests a skew toward areas less common in other studies, though, since there is no comparable listing of nonethnographic studies of workplaces, we cannot test whether that is so.

But if the Census categories for occupations covered in the U.S. workplace15 cases are compared with Census data for the whole United States, it is clear that some categories have been over- and others underrepresented.16 For instance, sales occupations were held by 10 percent to 12 percent of the labor force in 1970 through 1990, but in the Hodson set they do not appear in the 1970s and are only 3 percent to 4 percent of its cases later.17 For the same periods, the more skilled manual workers (“precision production, craft, and repair”) are underrepresented while the semiskilled (“machine operators, assemblers, and inspectors”) are markedly overrepresented. Industry codes wholly unrepresented include, for instance, oil production, the postal service, hotels, advertising, wholesale or retail groceries, theatres, and motion pictures. Practical reasons for this must, presumably, include the suitability of the setting for ethnographic work. (One can see reasons why the occupations of the self-employed, or in religious, criminal, and military workplaces, might be underrepresented—though there have been some participant reports on such settings, but not usually ones that focus on their character as workplaces.)

On the other hand, car assembly lines and certain medical settings18 are prevalent among the cases studied—so their workers’ voices are overrepresented. One may presume that the reasons for this include the perceived social importance of such work and its theoretical or political relevance. The early post-war preoccupation with productivity in industrial reconstruction (and its accompanying funding for industrial research), and then the emphasis on the role of technology and the high rates of industrial conflict experienced, later followed by managerial fascination first with Taylorism and then with the secrets of Japanese success and attempts at imitation of Japanese practices, are among the social factors relevant—though, as Paradeise (2003, p. 7) remarks, Taylorism was never as dominant in reality as it was in industrial sociology. We may reasonably take it that the resultant set has better claims to represent anglophone workplace ethnography19 than it has to quantitative representation of workplaces, and that the choices made from the total range of available occupations reflect proclivities—also evident in nonethnographic work—for certain kinds of workplace; some of these proclivities are disciplinary.

Disciplinary Differences

Although Hodson and his colleagues using this set have published mainly in sociological journals, authors of the studies in the set also come from a range of other disciplines20—especially anthropology—and their focuses differ. The boundaries between the social science disciplines have not always been clearly marked, either organizationally or intellectually, but we find some differences here. Among the sociologists, semiskilled work on the factory floor in manufacturing industries is salient. From the 1970s through the 1980s, they consistently favored that type of workplace, and in each decade the percentage of such cases in the set overrepresents U.S. Census figures by two to one.21

Anthropologists studying semiskilled work tended to choose jobs with strong, distinctive cultures (e.g., Applebaum, 1981; Orbach, 1977), in industries where marginalized populations work (e.g., Fink, 1998), or in non-Western settings (e.g., Roberson, 1998; Kapferer, 1972). More prevalent amongst them are studies of jobs requiring highly specialized technical skills, or on the verge of disappearing and so inviting anthropological salvage (Pilcher, 1972; Gamst, 1980). The folklorists and oral historians concentrate on vanishing trades, seemingly without requiring a theoretical story. Work from business schools is naturally more management-oriented than that from departments of sociology, while area studies predictably tend to contribute broader background knowledge of foreign societies.22

For sociologists, the macro level was often a starting point; broad shifts in industry and employment trends are seen to make showing how those changes are experienced on the ground relevant; the car plant is, thus, studied to exemplify relations of production under capitalism. Even when studying one particular place, it is done with the intention of generalizing outward from a particular workplace. As Beynon says, “If [this book] … has been successful…, its implications will extend beyond its specific context—one assembly plant in Liverpool in the 1960s.” (1975, p. 14). For anthropologists, particularly in work before the 1990s, the significant context has been more that of the very specific site or group studied, often an enclosed whole, with the ethnographer aiming to unfold the local and peculiar culture of one workplace. Methods also differed somewhat. Among the sociologists, 36 percent took a job in the workplace studied, 49 percent were involved in some other form of participation, and 15 percent had little or no participation. Anthropologists participated more, half taking a job and half participating in some other way. For anthropologists, ethnographic method is taken for granted as standard practice, so the question to ask about their studies might be why they worked on workplaces and close to home, while for sociologists the question is why they chose the relatively marginal method of ethnography.23

Both disciplinary tradition and intellectual fashions have, thus, clearly influenced the historical pattern of production,24 as has the demography of academia. But those are not the only factors; we also need to look more closely at the roles of gender and politics.

Gender and Politics

Gender plays an important role in the historical pattern; the peak number of studies for men was in the 1970s, but for women in the 1980s, as shown in Table 1.25

Table 1. All, Date Fieldwork Started, by Gender

Authors	–1949	1950s	1960s	1970s	1980s	1990s*	n.d.	N
Note n.d., no rate given.
Male	10	6	14	38	15	7	5	95
Female	1	1	1	14	28	13	–	58
Male/Female	–	1	1	1	–	–	–	3
N	11	8	16	53	43	20	5	156

The changing pattern for women probably reflects their increasing participation in academia, as much as a shift among the women active; that their number of ethnographies too goes down from the 1990s suggests that both factors may be involved.26 The gender shift within the academic disciplines was sharper than this reveals, as Table 2, with amateurs (all but one of the men) excluded shows. Of fieldwork started in the 1970s (much of it published in the 1980s), the amateurs actually contributed more than a third of the total for the men, so to omit them would noticeably reduce the (quasi-)ethnographic knowledge provided for that period. The proportion of cases by American sociologists goes down for men and up for women from the 1970s to the 1990s; this suggests that the impact of feminist ideas about the greater value and appropriateness of qualitative methods (Reinharz, 1992; Oakley, 1999) was significant—and shows an ethnographic turn for women only, with the reverse for men. However, by the 1990s, fewer than half the authors were sociologists, while among the women, anthropologists and sociologists contributed equally.

Table 2. Percentage by Men, Excluding Amateurs

	–1949	1950s	1960s	1970s	1980s	1990s
Percent Male	100	[83]	93	67	35	33

But choice of method is not the only area where feminist views had some influence. Trends in topics as well as methods have been relevant, and leftist or feminist politics encouraged workplace research. The former was manifested in the political choice made by several authors to experience factory work. (For a few, their activism then led to a premature end to fieldwork.) Examples of such choices are as follows:

I had been thinking of taking a working-class job for years before I finally went to work in the factory…. Trying to keep one's feet on the ground and have a concrete grasp of ongoing working-class struggles had always seemed to me … essential if we were to understand the people on whose behalf we were supposedly theorising. (Cavendish, 1982, p. 1)

…I had become interested in trying to understand the meaning of … the work situation under an economic system … guided solely by profits…. My interest in becoming a factory worker also stemmed from a deeply felt need … to improve the working conditions in the factory by mobilizing the workers to work collectively toward such a goal. (Devinatz, 1999, pp. xii–xiii)

These authors pursued their political agendas as individuals. Linhart (1981) provides a more organized case: a Maoist group in France sent about 40 intellectuals, many drawn from Althusser's students (among whom Linhart was a leader), to establish themselves deeply in proletarian work life in order to promote revolution.27

Feminist politics, starting in the late 1960s, has implied both studying women and taking account of the way their experience has differed from that of men, as well as adopting methods of study seen as appropriate to the study of women. In the context of the study of work, this implies giving attention to the different range of occupations held by women. Of the 59 books authored by women, 19 had “women” or an equivalent reference to gender (“maid,” “ladies”) in the title, a number—overlapping with that—were about women in gender-atypical jobs (welder, car salesperson) written about as atypical, and more were on job categories dominated by women such as work in restaurants, clothing, or clerical settings, while hardly any of the books by men focused on such groups.28 The importance of qualitative methods has been a recurring theme in feminist discussion. Quantitative methods have been seen as inadequate to the study of women, inherently placing them in inappropriate categories and failing to attend to the nuances and meanings of their experience. Not all feminists have supported that line of argument, but it has been salient and may be presumed to have led to some principled political choices of method. If that is so, it suggests that women's jobs may be overrepresented/men's jobs underrepresented among ethnographies as compared with studies carried out using other methods of data collection.

A rough coding of the role of declared “political” reasons for taking the job, or undertaking the study, used these categories, based on authors’ explicit statements29:

• Took job/undertook study for leftist political reasons, or ended politicized.
• Took job/undertook study for feminist political reasons, or ended politicized in a feminist sense.
• Wanted to do good, make life better, in a less “political” way.
• Purely theoretical or descriptive aim.

(Some of the books coded into the last two categories may, of course, have had unstated political motives—as may others have done, but we consider only those of which there is explicit statement.) About a quarter of the authors whose aim was stated say they started with leftist or feminist intentions, and just over half with purely theoretical or descriptive intentions. By decade of starting fieldwork, the proportion with more political aims in the 1970s goes up to about a third while the proportion in the most academic category goes down to about half, and those proportions remain thereafter. However, the highest numbers for general leftism are in the 1960s and 1970s, while for feminism, they are in the 1970s and 1980s. Of the women authors, 36 percent gave feminist reasons, while no men did. Again there are disciplinary differences. Of the sociologists, a third had leftist or feminist intentions, slightly skewed toward feminism; anthropologists with “political” intentions were all feminists, while none of the amateurs were.

At least as significant as those facts in themselves is that these political stances were related to the types of workplace studied. Table 3 shows the differences by broad occupational categories (omitting “service” and “farm” jobs, which do not fit well into the manual/nonmanual distinction) and reveals that the leftists were concentrated on manual occupations, while those claiming a theoretical/descriptive approach studied nonmanual work situations somewhat more often than manual ones. The distinction that Isaac (2009, p. 942) makes between work produced by movement activists to advance the movement and work by outsiders drawing on movement activity for themes is applicable here, and overlaps with the amateur/professional distinction. Obviously, the causal relationship could run either way; politics affects interests, but experience of a situation may also affect politics. Whichever ways it has run in these cases, there are likely consequences—which would apply just as much to nonethnographic studies—for the themes emphasized and interpretations favored.

Table 3. Type of Job Studied × Political Approach

	Leftist	Feminist	Do Good	Theory etc.	N
Manual	24%	8%	25%	43%	87
Nonmanual	1%	10%	15%	73%	71

Despite the commonly expressed wish to give voice to the workers, some politically motivated authors were not happy with the voices they found; they wished that the workers had perceived the situation as they did and responded to it politically, and these authors were likely to conceptualize their findings as revealing oppression or false consciousness. A managerial perspective is of course equally likely to have effects of this kind, if less likely to be declared as a political motive.

Is one of the attractions of ethnography that it provides the kind of narrative that Steinmetz (1992) describes as potentially a basis for class or political mobilization? Arguably, the performance of such a function cuts across the distinction between (some) “amateur” and “professional” ethnographies; it is not only amateurs who have political motives. Isaac, McDonald, and Lukasik (2006, p. 79) have sketched a broad outline of the way in which the New Left mass movement revitalized parts of the labor movement, diffusing general militancy across sectors and influencing politics in ways favorable to union organization, while feminism and the Civil Rights movements brought groups previously less involved into various forms of activism. Industrial democracy became politically salient for some (notably political scientists contributing to this sample) from the late 1960s. As Greenberg (1986, p. 7) says:

In the late 1960s and early 1970s the political air was filled with talk of direct democracy, participatory democracy, and workers’ control. To many on the Left, these … seemed to presage the formation of a more humane, egalitarian and democratic society … such contexts were hypothesised to be rich educational environments in which people would overcome alienation in its many forms, …and become transformed into class-conscious members of the working class, willing and able to act politically in accordance with their broad class interests….

Our data show the impact of these historical movements in the climate of the wider society.30

The Publication System

But we cannot explain the genesis of such books purely in terms of wider historical trends and decisions made by their authors. The circumstances generating them must also include the decisions of publishers to publish, or even commission, such work, and market decisions may have left unpublished some projects, which would have changed the picture. By the stage of publication, 36 of the books come from the 1970s, 46 from the 1980s, and 48 from the 1990s. (Data available for most cases show that fieldwork and writing up often take a long time.) It is evident that specialist publishers—their specialisms ranging from university press series to amateur local history–have played a key role, and the work has been published by firms of varying character whose books aim at diverse audiences.

One clearly nonacademic genre, which accounts for a number of the “amateurs,” is that of the working man's autobiography published by a left-wing publisher.31 Swerdlow (1998, p. 6) chose to write in a nonacademic way, following that model, and examples she mentions include Kamata (1982) Seider (1984) and Turner (1980) from Hodson's sample. Several publishers have specialized in this area, some of whom have commissioned or actively encouraged work they eventually published. Singlejack Press, which published two of the books (Seider; Turner), was set up by Stan Weir:

…a rank and file labor activist for more than fifty years as a seaman, autoworker, teamster, housepainter, and longshoreman. He earned a doctorate … and was professor of labor and industrial relations at the University of Illinois. Weir founded Singlejack Press to publish writing by workers. (University of Minnesota Press, 2008)

Seider (1984, pp. 136–137) makes it clear that the existence of this press was crucial to him finally organizing to get his material published, with Weir playing an important role in improving his manuscript; Weir is also mentioned by Mers (1988) from the sample as having encouraged him to write when otherwise he might not have. Among the manual jobs studied, what one author calls “roughneck occupations” (Applebaum, 1981, p. vii) look overrepresented, perhaps because they often have strong occupational subcultures, and to professors and publishers seem more exotic—but other factors are suggested when we note that among the workplaces studied32 86 percent of the nonacademic “amateur” authors’ cases, but only 41 percent of the professionals’, are of manual workers; amateurs provide 20 percent of the manual cases but only 3 percent of the nonmanual ones. These last differences could reflect intellectual choices made by the academics, but they might also indicate varying opportunities for access to both workplaces and publishing houses.

Temple University Press is responsible for 11 of the books. More than half of those are in a series “Labor and Social Change,” edited by Paula Rayman and Carmen Sirianni, whose Web site states its remit as to include “books on workplace issues like worker participation, quality of work life, shorter hours, technological change, and productivity, as well as union and community organizing and ethnographies of particular occupations.” Both editors have themselves been active in research in this area, and Sirianni's support is acknowledged in three books in the sample.33 (Other major U.S. university presses contributing at least five books show no such clear editorial hand.34) Only one of the six books whose authors were manual worker amateurs (Mers, 1988) was published by a known academic house, the University of Texas Press, whose Web site declares a partly nonacademic remit: “…to serve the people of Texas, the Press also produces books of general interest for a wider audience, covering, in particular, the history, culture, arts, and natural history of the state”; this may have applied here. Another five books appeared in two anthropological series edited by George and Louise Spindler, whose foreword explains that these volumes aim to show students in beginning and intermediate-level social science courses how different life is in different places; thus, books in that series had to be about occupations interesting to beginners and written in an accessible way.

Some other presses represented are totally different; these are the museum presses, and more specialized ones that exist to publish local history and folklore. The books they produce may sometimes be of sociological interest, but it is not the social science community that they are aimed at and often not an academic one. The general trade press is also represented, though no individual publisher contributed as many as five books. It seems probable that it is in this publishing niche that the availability of a “story” of perceived general interest (“our working conditions are hell,” “women can do traditionally masculine jobs,” “some doctors behave unethically”) is most likely to affect the chances of successful publication; picturesque ethnographies have an obvious advantage here over boring tables.

Thus, the books clearly reached publication by somewhat different routes. It is evident that a difference, not wholly resting on intellectual evaluations of manuscript content, is made by the character of the available publishing houses and the relation different types of author have to them. There are also broader aspects of the publishing situation that have negative as well as positive influence on which books appear; academic authors cannot be held solely responsible for decisions publishers make about their work on good commercial grounds. Thompson (2005) has described the increasing difficulties experienced in selling enough copies of academic monographs to justify their production, and that would have made some contribution to the apparent decrease in production of ethnographic book even if ethnographic work continued to be done.

Conclusions

What, then, do these data tell us about the history of ethnography in sociology? It is clear that the total pattern of available ethnographic and quasi-ethnographic studies in the subfield considered is a product of intertwining historical trends, some within and some outside academia, and within academia some within and some outside sociology. Here, at least, the data so far suggest an ethnographic bulge in the middle of the period covered rather than an ethnographic turn that has continued, though the inclusion of the further cases mentioned in note 4 above might change this. Given what is available, other analysts might easily have made decisions different from Hodson's about which books to count as ethnographic, but the scope for decision would still have depended upon the range of existing work available to choose from. The role of publishers in determining that range is a factor that merits more historical attention than it has received.

We note the relative importance of the roles played by changes in society and changes internal to social science. Although one can detect influences from specific contributions to the literature more narrowly defined—ideas from such widely known sources as Merton on bureaucracy, Blauner on alienation, or Braverman's labor process theory have figured in the topics chosen–-the differing outlooks of critical and interpretive sociologies, and developing arguments for qualitative method, have also played their part. But one does not get the impression that those were always more salient here than less distinctively academic impulses. Within sociology and anthropology, the wider social movements of leftism and feminism have sometimes led to a commitment to ethnographic work; they have taken local disciplinary forms, which might justify treating them as academic changes, but those have not occurred independently. Even when “amateur” authors’ work is ignored, the professional literature has clearly been influenced by broader societal movements of historical change and by meanings not purely academic.35 No reader should be very surprised to learn that social science is part of the society that produces it, though this is more easily forgotten in relation to the superficially purely technical history of the choice of research methods. The trajectory of the performance and publication of these ethnographic studies has followed the trajectory of wider public movements and concerns. Thus, quite diverse streams of interest have converged on the area of close workplace investigation—and approached it diversely. To look at another area of substantive work would be unlikely to reveal the relevance of just the same streams of interest—for instance, in studies of the family, religious interests have surely been more often, and political interests less often, represented—but those differences would be well worth following up.

An author using Hodson's other criteria who had decided to limit the set used to books by individuals identified with academic sociology would have had only 83 cases to analyze, and they would have been distributed differently across the range of work situations, so substantive conclusions based on them would have been somewhat different from Hodson's. We have shown that in at least one area the boundary between amateur and professional has been quite ambiguous, and this is one of the factors that raise questions about the appropriateness of treating the history of social science as if it were only about work done by people with exclusively academic identities and holding academic jobs with the right label. (This is particularly likely to be an issue at early stages of disciplinary development.) However, attempts to avoid imposing such boundaries can also have confusing effects; the theorized practice of an institutionalized discipline does not necessarily have much in common with work generated in other milieux. What needs to be taken into account depends both on the problem addressed and on the empirical character of the particular setting studied.

Although the intention of this paper is primarily historical, our analysis draws attention to some points of methodological relevance. To the extent that ethnography does have special features that give access to realities not revealed by other methods, the variation over time in ethnographic prevalence brings about the availability of data on different aspects of the situations studied, or of data of different quality.36 To the extent that method does make a difference, a large number of methodologically impeccable individual studies of one kind can add up to a biased account of the total social realities—especially if their distribution across the relevant topic areas is clustered. The valued authenticity of intensive direct observation may be bought at the cost of disproportionate representation of different settings, and methodological choices often carry some nonmethodological baggage with them. As social scientists, we should probably favor the prevalence of a mixture of methods, since only when comparisons can be made between them can we distinguish the effects of methodological choices on the problems accessible and the findings made.

Our exploration of one valuable set of cases is too narrow to provide a good basis for an empirical history of sociological ethnography in general, but it is richly suggestive of the production processes that may be relevant to the understanding of the history of research in other subfields. It draws attention to the many lines of potential work on the history of ethnographic practice, within sociology or social science more broadly, which would contribute to a fuller account.

1. 1

   Secondary analysis of existing data sets has become increasingly prevalent in sociology, and there is much methodological discussion about the ways in which such data can be used. We note that Hodson's strategy has one important difference from the usual secondary use of survey data: the coding of the material collected was done entirely by his team, not by the original researchers, which changes some of the issues raised. (In some of his papers, however, quotations from the books are also used in a “qualitative” way, if mainly as no more than illustrations.)

2. 2

   Those identifying such a trend commonly condemn it, but clearly it can also be seen favorably. The condemnation by some of those active in the field may have helped create a welcoming reception for some of the amateur work discussed below.

3. 3

   One critical analysis, of the state of British industrial sociology (Thompson & Ackroyd, 1995), argued that patterns of worker resistance and misbehavior were being neglected because a perspective, partly affected by a Foucauldian approach, had become adopted that assumed that only management had active and effective agency.

4. 4

   Data were from the United States in 103 books (65 percent) and from Britain in 23 (15 percent). (The rest are on countries as varied as Canada, Sweden, and Zambia, though some of their authors are American or were based in America; a few further books were published in America, some in translation. Japan and transplanted Japanese practices are salient.) Studies of, and from, America are presumably most fully represented.

5. 5

   Hodson has subsequently identified 48 more books, which await coding. Of these, 24 overlap with the period of his original coverage, indicating that at least 13 percent of available books were not identified in the first round. However, the earliest was from 1986, so only the figures for the 1990s are likely to have been much affected by the missing cases. But two reviews (Morrill & Fine, 1997; Smith, 2001) of closely related areas mention more possibilities, and Edwards and Bélanger (2008) suggest others.

6. 6

   Hodson's database is available at: http://www.sociology.ohio-state.edu/rdh/Workplace-Ethnography-Project.html

7. 7

   This is not quite the same as Hodson's coding of “ethnographer's theoretical orientation,” which could be imputed rather than directly stated, and only some of whose categories could be seen as political, though it has some conceptual overlap.

8. 8

   Assessment of the validity of the data reported, once the directness and length of observation were established, was done only through the coding procedure, which checked very carefully the reported typicality of each point coded. Then, perhaps somewhat optimistically, “An underlying assumption … is that the ethnographic data constitute a realist account…” (Hodson 2004, p. 16).

9. 9

   ‘Par ma foi ! il y a plus de six mois que je fais de l'ethnographie sans que j'en susse rien…’ [Apologies to Molière].

10. 10

    The question may be raised whether unintended ethnographies should be treated as members of the class. An alternative strategy, one adopted by Lena and Peterson (2008, p. 699), was to consider a genre as existing only if “identified as such by participants and commentators”; here Hodson's role as commentator may perhaps be invoked to override other distinctions. Masson and Suteau (2008, pp. 237–238), after study of data from French journals, conclude that even where authors claimed to be doing ethnography, they were not all following the same model and that the different disciplines involved each had their own research formulae that affected the published versions whatever the authors’ declared intentions.

11. 11

    Peneff (2009, p. 95) considers the kinds of observation routinely done, without the intention of writing about it social scientifically, by groups ranging from novelists to political activists, and offers a chart distinguishing among them that helps thought on the matter.

12. 12

    On his career, see “About the Author” in Applebaum (1981, p. vi), and Greaves and Applebaum (2001).

13. 13

    He had held a trucking job before and during his graduate work, and his brother had also been a trucker; they came from a working-class trade unionist family. Moreover, he saw trucking as “more than a battle for more money or better pay rates and against monotony and fatigue. It is a place where the self is forged. Truckers attempt to manipulate the workplace to construct a positive self…” (1994, p. 11).

14. 14

    Sources by the author studied, or by Hodson, are omitted from this count, as are reviews of the book.

15. 15

    Here and below, when only U.S. cases are used, it is because only they offer enough examples for meaningful statements to be made about trends over time. We need to bear in mind that there may be many other cases from nonanglophone areas that have not been translated and could well change the picture.

16. 16

    However, this comparison is between number of workers and of workplaces; no way of overcoming this difficulty seems available.

17. 17

    For Hodson's cases, the date of publication is used and the whole decade is taken; the first means that some data were collected in the previous decade, while the second means that some were collected years after the Census date. One hopes that these cancel out. At least the examples mentioned in this rough and ready comparison should stand nonetheless.

18. 18

    Hospitals and care homes, not general practice; cancer and surgery, not dermatology or infectious diseases….

19. 19

    But Peneff (2009, p. 126) draws attention to the fact that French observation in factories has been oriented to different issues from those of American studies. It must be presumed that the small number of books translated into English were selected as ones likely to interest the anglophone audience, so cannot, as a subset, be taken to represent their home audiences.

20. 20

    There are also nurses, folklorists, political scientists, area studies specialists, members of business, and industrial relations schools; one book (Orr 1996) comes from a cross-disciplinary community only partly based in academia (Suchman, 1999). But it is not always easy to impute a disciplinary affiliation. A recurring problem is membership in departments such as business, women's studies or nursing, where members may, nonetheless, be “sociologists” or “anthropologists”; here, where they appear to have a dual affiliation, the social science one has been counted. In eight cases, however, no disciplinary affiliation could be imputed.

21. 21

    Here workplaces, not books, are counted, since some books studied both manual and nonmanual workplaces.

22. 22

    However, some of the studies under other disciplinary heads also drew on rich background knowledge because they were carried out by natives of the foreign countries studied, who could be doctoral candidates at U.S. universities.

23. 23

    It was anticipated that there might be departmental patterns, but the clusters of three or four cases from the same department among the books publishing dissertations were all ones with large cohorts of graduate students—Berkeley, Chicago, Columbia, Harvard, Northwestern—so little could be made of that as data on schools.

24. 24

    Lena and Peterson's (2008) work on the trajectories of music genres, moving over time from avant-garde to scene-based, industry-based, and traditionalist, is suggestive of analogies in this field, though we do not have space here to pursue this.

25. 25

    None started their fieldwork in the 2000s.

26. 26

    Conceivably, this pattern might also owe something to changes in occupational structure that cut the numbers of traditional male manual jobs and increased the number of nonmanual jobs commonly held by women.

27. 27

    Reid (2004) gives a general account of the movement, and Linhart's daughter has also written an interesting book about its members’ lives (Linhart, 1994).

28. 28

    This pattern may of course also owe something to the relative ease of access by different researchers to gender-typed work settings.

29. 29

    In eight cases, there was not enough information to code this.

30. 30

    One may wonder, however, why despite this there is so little focus on issues of race in these workplace studies.

31. 31

    It would be interesting to compare this with the French examples cited by Peneff (1979), who comments on the unrepresentative nature of those who had produced such work. Autobiographical accounts by ordinary workers, solicited for comparison, did not have the same characteristics.

32. 32

    Here workplaces, not books, are counted, since some books studied both manual and nonmanual workplaces.

33. 33

    Journals may play a similar role. The editors of Work, Employment and Society (Taylor et al., 2009, p. 7) have recently attempted to revive firsthand accounts of work experience by initiating a special section in the journal for them, “to enable workers’ voices to convey sociological insight into the realities of work and employment,” and it is clear that they see this as significant politically as well as academically.

34. 34

    We note also that 10 of the doctoral authors’ books were published by the university press of their doctoral institution.

35. 35

    More work such as Stott's, looking at movements of cultural feeling that cut across the distinction between the academic and nonacademic fields, would be valuable here.

36. 36

    The same inevitably applies to the findings of any other single mode of data collection. But some of the features of work done using any one method may also be represented in other work done by the same types of author; not all the characteristics of existing ethnographic studies can be imputed to the method as such.

Biographies

• JENNIFER PLATT is emeritus professor of sociology at the University of Sussex. She has been active and held office in the BSA, the ISA, and the Section on the History of Sociology of the American Sociological Association. Her main interests are in the history of sociology and aspects of research method. Major publications include A History of Sociological Research Methods in America, 1920–1960, Cambridge University Press, 1996, and The British Sociological Association: A Sociological History, Durham: Sociology Press, 2003; [e-mail]

• CHARLES CROTHERS completed his PhD at Victoria University of Wellington in Sociology after earlier study at the University of Waikato (VUW). He joined Auckland University of Technology after just over 5 years as Head of Sociology at University of Natal Durban having worked before then at the Universities of Auckland and VUW. His current areas of research include applied social theory (on the work of Robert K. Merton and on social structure) methods of social research; the sociology of the social sciences; study of the policy process in New Zealand and the role of social sciences within it; the history and current situation of sociology and social sciences; and organizational and community features of settler societies, especially New Zealand.

• MERVYN HORGAN is an assistant professor of sociology and a member of the graduate faculty in Social and Political Thought at Acadia University, where he teaches social theory and cultural sociology. His research interests include ethnographic representation and exemplarity, the cultural sociology of strangership, and case studies of neighborhood-level change in cities.