The Extended Case Method*
Michael Burawoy
University of California, Berkeley
In this article I elaborate and codify the extended case method, which deploys participant
observation to locate everyday life in its extralocal and historical context. The
extended case method emulates a reflexive model of science that takes as its premise the
intersubjectivity of scientist and subject of study. Reflexive science valorizes intervention,
process, structuration, and theory reconstruction. It is the Siamese twin of positive
science that proscribes reactivity, but upholds reliability, replicability, and representativeness.
Positive science, exemplified by survey research, works on the principle of the
separation between scientists and the subjects they examine. Positive science is limited
by “context effects” (interview, respondent, field, and situational effects) while reflexive
science is limited by “power effects” (domination, silencing, objectification, and
normalization). The article concludes by considering the implications of having two
models of science rather than one, both of which are necessarily flawed. Throughout I
use a study of postcolonialism to illustrate both the virtues and the shortcomings of the
extended case method.
Methodology can only bring us reflective understanding of the means which have
demonstrated their value in practice by raising them to the level of explicit consciousness;
it is no more the precondition of fruitful intellectual work than the knowledge of
anatomy is the precondition of “correct” walking.
Max Weber—The Methodology of the Social Sciences
True, anatomical knowledge is not usually a precondition for “correct” walking. But when
the ground beneath our feet is always shaking, we need a crutch. As social scientists we are
thrown off balance by our presence in the world we study, by absorption in the society we
observe, by dwelling alongside those we make “other.” Beyond individual involvement is
the broader ethnographic predicament—producing theories, concepts, and facts that destabilize
the world we seek to comprehend. So, we desperately need methodology to keep us
erect, while we navigate a terrain that moves and shifts even as we attempt to traverse it.
*Address correspondence to the author at Department of Sociology, University of California, Berkeley, CA
94720, Burawoy@socrates.Berkeley.edu.
I have been writing this paper for twenty years. Earlier versions are barely recognizable due to dialogue,
discussion in many venues. In recent years two people in particular have sustained me in this endeavour. Erik
Wright has plied me with dozens of pages of intense argumentation to the effect that there can be only one model
of science, while Peter Evans has insisted that I persist despite all opposition. And opposition there was plenty,
from hostile receptions in talks to dismissive reviews from journal referees. The extended case method enjoyed
extended and lively public discussion in various midwest seminars at the University of Wisconsin, Madison,
Northwestern University, University of Chicago, and the University of Minneapolis, Minnesota. I have taught
participant observation since I arrived at Berkeley in 1976 and it was in those heated courses that my ideas took
shape as well as in working with graduate students on two collections of Berkeley ethnographies. The first
appeared as Ethnography Unbound (1991). Teresa Gowan, Leslie Salzinger, Maren Klawiter, and Amy Schalet
have been intent on holding me accountable for what I say, while Raka Ray, Jennifer Pierce, Charles Ragin,
Michael Goldman, Bob Connell, Nora Schaeffer, and especially Linda Blum provided more gentle stimuli over
the years. My greatest debt is to Jaap van Velsen, my first sociology teacher who, as an anthropologist, embodied
the extended case method, although he’d recoil in horror at the formalization to which I have subjected it. Finally,
I’d like to thank Craig Calhoun for steering this into print so that I can finally forget about it.
Sociological Theory 16:1 March 1998
© American Sociological Association. 1722 N Street NW, Washington, DC 20036
Like other handicaps, the ethnographic condition can be dealt with in one of two ways:
containing it or turning it to advantage. In the first strategy, we minimize our predicament
by limiting our involvement in the world we study, insulating ourselves from our subjects,
observing them from the outside, interrogating them through intermediaries. We keep our
feet on the ground by adhering to a set of data collecting procedures that assure our distance.
This is the positive approach. It is best exemplified by survey research in which
every effort is made to suspend our participation in the world we study. We try to avoid
affecting the situation we study, standardize the collection of data, bracket external conditions,
and make sure our sample is representative.
In the alternative strategy we thematize our participation in the world we study. We
keep ourselves steady by rooting ourselves in theory that guides our dialogue with participants.
Polanyi (1958) elaborates this idea in detail, rejecting a positivist objectivity based
on “sense data” in favor of a commitment to the “rationality” of theory—cognitive maps
through which we apprehend the world. This “dwelling in” theory is at the basis of what I
call the reflexive model of science—a model of science that embraces not detachment but
engagement as the road to knowledge. Premised upon our own participation in the world
we study, reflexive science deploys multiple dialogues to reach explanations of empirical
phenomena. Reflexive science starts out from dialogue, virtual or real, between observer
and participants, embeds such dialogue within a second dialogue between local processes
and extralocal forces that in turn can only be comprehended through a third, expanding
dialogue of theory with itself. Objectivity is not measured by procedures that assure an
accurate mapping of the world but by the growth of knowledge; that is, the imaginative
and parsimonious reconstruction of theory to accommodate anomalies (Kuhn 1962; Popper
1963; Lakatos 1978).
The extended case method applies reflexive science to ethnography in order to extract
the general from the unique, to move from the “micro” to the “macro,” and to connect the
present to the past in anticipation of the future, all by building on preexisting theory. In my
own use of the extended case method I used my experiences as a personnel consultant in
the Zambian copper industry to elaborate Fanon’s theory of postcolonialism. I tried to
expose the roots of consent to American capitalism by applying Gramsci’s theory of hegemony
to my experiences as a machine operator in a South Chicago factory. I have explored
the nature of work organization and class formation under socialism by combining Szelenyi’s
theory of class structure and Kornai’s theory of the shortage economy. This was based
on laboring in Hungarian factories—champagne, auto manufacturing, and steel. Most
recently I have worked my way outward from a small furniture factory in Northern Russia
in order to develop theories of the transition from socialism to capitalism using Marxist
notions of merchant and finance capital. How can I justify these extravagant leaps across
space and time, from the singular to the general, from the mundane to the grand historical
themes of the late twentieth century? That is the question that motivates this article.
Although it is more usual for ethnographic studies to confine themselves to claims
within the dimensions of the everyday worlds they examine, I am not alone in “extending
out” from the field. Indeed, this was one of the hallmarks of the Manchester School of
social anthropology, which first coined the phrase, “extended case method” (Garbett 1970;
Gluckman 1958, 1961a, 1961b, 1964; Van Velsen 1960, 1964, 1967; Mitchell 1956, 1983;
Epstein 1958). Instead of collecting data from informants about what “natives” “ought to
do,” they began to fill their diaries with accounts of what “natives” actually were doing,
with accounts of real events, struggles, and dramas that took place over space and time.
They brought out discrepancies between normative prescriptions and everyday practices—
discrepancies they traced to internal contradictions but also to the intrusion of colonialism.
They began to restore African communities to their broader, world historical context.
THE EXTENDED CASE METHOD 5
Not just in Africa but in the United States, too, there is a rich but inchoate tradition of
scholarship in the implicit style of the extended case method. Community ethnographies
have not always stopped at the tracks but incorporated the wider contexts of racism and
labor markets (Liebow 1967; Bourgois 1995) as well as urban political regimes (Whyte
1943; Susser 1982; Haney 1996). Workplace ethnographies, traditionally confined to “plant
sociology,” have also taken external factors into account, such as race and ethnicity (Lamphere
et al. 1993 ), citizenship (Thomas 1985), markets (Smith 1990), and local politics
(Blum 1991). Participant observation studies of social movements locate them in their
political and economic context (Fantasia 1988; Johnston 1994; Ray 1998). Ethnographies
of the school have always sought to explain how education is shaped by and at the same
time influences wider patterns of social inequality (Willis 1977; MacLeod 1987; and Powers
forthcoming). Family ethnographies have found it impossible to ignore influences
beyond the household (Stacey 1990; Devault 1991; Hondagneu-Sotelo 1994), upholding
Dorothy Smith’s (1987) feminist injunction to locate lived experience within its extralocal
determinations.1
The rudiments of the extended case method abound in these examples and elsewhere.
What I propose, therefore, is to bring “reflective understanding” to the extended case
method by raising it to the “level of explicit consciousness.” But, contra Weber, this is not
simply a clarificatory exercise. It has real repercussions for the way we conduct social
science. Indeed, it leads to an alternative model of social science and thus to alternative
explanatory and interpretive practices—something social scientists are reluctant to countenance.
We prefer to debate appropriate techniques or even tolerate the rejection of science
altogether rather than face the possibility of two coexisting models of science, which
would wreak havoc with our methodological prescriptions. Still I hope to demonstrate that
reflexive science has its pay-off, enabling the exploration of broad historical patterns and
macrostructures without relinquishing either ethnography or science.
By ethnography I mean writing about the world from the standpoint of participant
observation; by science I mean falsifiable and generalizable explanations of empirical
phenomena. In developing my argument it will be necessary to distinguish (a) research
method (here survey research and the extended case method), which is the deployment of
(b) techniques of empirical investigation (here interviewing and participant observation)
to best approximate (c) a scientific model (positive or reflexive) that lays out the presuppositions
and principles for producing science. In elaborating the different dimensions
of the extended case method, I hope to improve its execution, justify it as a science,
albeit a reflexive science, and draw out broader implications for the way we study the
world.
In order to illustrate and explicate the extended case method I return to a study conducted
between 1968 and 1972 in the then newly independent African country of Zambia.
Of all my studies I have chosen this one because it most effectively illustrates both the
virtues and the limits of the extended case method. First, the virtues: the extended case
method is able to dig beneath the political binaries of colonizer and colonized, white and
black, metropolis and periphery, capital and labor to discover multiple processes, interests,
and identities. At the same time, the postcolonial context provides fertile ground for recondensing
these proliferating differences around local, national, and global links. Second, the
1
Smith’s “sociology of women” begins by debunking abstract, decontextualized, and universalistic sociology
as the ideology of ruling men and turns to the concrete lived experience of women as point of departure. The
microstructures of everyday life, which women direct, become the foundation and invisible premise for macrostructures
controlled by men. This looks like the extended case method, but whereas Smith justifies it on the
grounds of the “standpoint of women,” I ground it in an alternative conception of science. In this regard I am
closer to Sandra Harding (1986, 1990), who works the terrain between androcentric science and postmodern
dismissal of science. Rather than surrender science to men, she calls for a successor science.
6 SOCIOLOGICAL THEORY
limits: the extended case method comes up against the very forces it displays. As the
renascent field of “colonial” studies makes clear, the colonies were not simply the site of
exotica but of experiments in new tactics of power, subsequently reimported back into the
metropolis (Stoler 1995; Mitchell 1988). Domination took on especially raw and exaggerated
forms, transparently implicating sociologists and especially anthropologists, coloring
their vision in unexplicated ways (Clifford and Marcus 1986; Asad 1973). Colonial and
postcolonial regimes of power throw into relief limits inherent to the extended case method.
Accordingly, the article is constructed as follows: I begin with a narrative of my study
of the Zambian copper industry (Burawoy 1972a, 1972b, 1974), highlighting the social
embeddedness of reflexive research. I then show how my study violated each of the four
principles of positive science. If there were only this model of science, then I would have
to abandon either the extended case method or science. However, the extended case method
is not alone in violating positive principles. I show that survey research, the quintessentially
positive method, transgresses its own principles because of inescapable context effects
stemming from the indissoluble connection between interviewer and respondent, and from
the embeddedness of the interview in a wider field of social relations. We can either live
with the gap between positive principles and practice, all the while trying to close it, OR
formulate an alternative model of science that takes context as its point of departure, that
thematizes our presence in the world we study. That alternative is the “reflexive” model of
science which, when applied to the technique of participant observation, gives rise to the
extended case method.
In saving both science and the extended case method, however, I do not eliminate
the gap between them. Making context and dialogue the basis of an alternative science
unavoidably brings into prominence power effects that divide the extended case method
from the principles of reflexive science. Postmodernism has done much to highlight these
power effects but, rather than make do with an inadequate science, it rejects science altogether.
I find myself working on the borders of postmodernism, without ever overstepping
the boundaries. Choosing to remain on the side of science, we have to live with its selfdetermined
limitations, whether they be the context effects of positive science or the power
effects of reflexive science. Given that the world is neither without context nor without
power, both sciences are flawed. But we do have a choice. So I finally ask when, where,
and why we deploy each of the two model-methods.
I. THE ETHNOGRAPHIC CONDITION EMBRACED
Reflexive science sets out from a dialogue between us and them, between social scientists
and the people we study. It does not spring from an Archimedean point outside space and
time; it does not create knowledge or theory tabula rasa. It starts out from a stock of
academic theory on the one side and existent folk theory or indigenous narratives on the
other. Both sides begin their interaction from real locations.
My own study of the Zambian copper mines began from publicly debated dilemmas of
the legacies of colonialism. I travelled to the copperbelt in 1968 in search of the policies
and strategies of transnational corporations toward the postcolonial regime. The two mining
companies, the Anglo American Corporation and the Roan Selection Trust, had their
roots in the colonial order of Northern Rhodesia, a British protectorate until 1964. How
were these companies responding to Zambian independence, whose stated objective was
to reappropriate control over the nation’s economy? This was not a trivial question since
the copper industry employed some fifty thousand people, 90 percent African and 10
percent expatriate. At the time of independence the mines provided 90 percent of foreign
exchange and 50 to 70 percent of government revenue. As far as Whitehall (and later the
THE EXTENDED CASE METHOD 7
Federation of Rhodesia and Nyasaland) had been concerned, Northern Rhodesia’s reason
for existence was her copper. Road and rail transportation, land and agriculture, taxation
and trade, labor and education, nationality and race were all designed to maximize the
export of copper. Zambia was the archetypal enclave economy, with copper its organizing
principle.
It was easier to study how mineworkers were faring than to disclose the mysterious
corporate practices of Anglo American Corporation and Roan Selection Trust. Mine was
not a study that could be accomplished by combing through documents since, as I was to
learn, by themselves they revealed so little. Interviews, conducted from the outside, were
no less useful since managers were protected by layers of public relations. Instead I took
advantage of my recently won mathematics degree and my contact in Anglo’s headquarters
to land myself a job in the Personnel Research Unit of the Copper Industry Service
Bureau. Located in Kitwe, at the heart of the mining region, this was the center of industrial
relations, both policy and practice.
Once there my attention turned to the more specific question of labor force localization,
or what had been called “African Advancement” but which since independence came to be
called “Zambianization.” Colonial rule left Zambia’s four million people with barely one
hundred university graduates and just over twelve hundred Africans with secondary school
certificates. So the country remained heavily dependent on white managers and experts.
Historically, the mining industry had been organized according to the color bar principle,
viz. no black person should exercise authority over any white person. It was a major aim of
the anticolonial movement to eradicate all such traces of white supremacy. How had things
changed in the postcolonial period? I began with the figures put out by the new government’s
Zambianization Committee, which painted a rosy picture of achievement. Four
years after independence there were fewer expatriates and more Zambians in “expatriate”
(white) positions. What lay behind this portrait of “deracialization”?
If comprehension of managerial strategies was largely barred to outsiders, any serious
study of Zambianization was totally off limits. Racial succession in what had been an
apartheid order was simply too explosive a question to openly investigate. Yet, of course,
it hung like a heavy cloud over all aspects of industrial relations. I could not have been
better placed to observe the different forces at work. Not only was I sitting in the mining
industry’s data gathering center but I became an active contributor to the industry’s new
job evaluation scheme that aimed to integrate black and white pay scales. As part of my job
I learned the stakes in negotiations among management, unions, and government.
So much for the perspective from the top. How did Zambianization look from inside
and from the bottom? Here I had to be more surreptitious. I organized a survey into the
working and living conditions of African miners, unrelated to Zambianization. But as
interviewers I chose young Zambian personnel officers who, I had reason to believe, were
at the storm’s eye of Zambianization. We would meet every week in the nominally desegregated
Rokana Club to discuss the progress of the survey but also Zambianization. Still
this was not enough. I worked in the Personnel Research Unit for one and a half years and
continued the research for another two years while I was an MA student at the University
of Zambia. There I recruited undergraduates to join me in studying “postcolonial” work
organization, underground and on surface. At least officially that was our goal; we were
also exploring Zambianization from below, from the standpoint of the vast majority of
unskilled and semiskilled workers. How did they feel about the Zambianization of supervisors
and lower level managers?
Our extended observations showed that white management met government Zambianization
targets as well as its own interest in maintaining the color bar by two types of
8 SOCIOLOGICAL THEORY
organizational maneuvers. The first strategy was blanket Zambianization. In the days of
colonialism the personnel manager was king, reigning over his African supplicants and to
a lesser extent over whites too. The personnel department lorded over the company town,
over life in the mine and the “compound” (Epstein 1958). An obvious target, the department
was entirely and rapidly Zambianized but at the same time shunted aside and stripped
of its powers, especially over expatriate employees, who were placed under the guardianship
of the newly created “Staff Development Advisor”—one of the former white personnel
managers.
The second stratagem was shadow Zambianization. During the three and a half years of
our research the position of mine-captain, that is the highest level of underground supervision,
was Zambianized. A number of old white captains were promoted into newly created
posts of assistant underground manager, and took with them many of their old powers
and responsibilities. Any Zambian successor had to operate in the shadow of his predecessor.
He became a buffer between his subordinates and the “real” mine-captain, now removed
to a comfortable office on surface.
These maneuvers to maintain the color bar had several, not hard to anticipate, consequences.
First, the organization became increasingly top heavy as the layers of management
thickened. Second, there was increased conflict between workers and their new
Zambian supervisors who were less effective, even if less abusive, than their predecessors.
Maintaining the color bar through Zambianization was a recipe for organizational rupture,
conflict, and inefficiency.
If blanket and shadow Zambianization undermined the organization, why did it continue?
What were the forces behind the retention of the color bar? How could a nationalist
black government ignore the continuity of the racial order, as it effectively did in its
Zambianization report? I sought the answer in the broader constellation of interests. First,
while the government embraced the rhetoric of Zambianization, African trade unions,
representing the unskilled and semiskilled miners, were more interested in higher wages
and better working conditions than in the upward mobility of supervisors. Second, Zambian
successors, caught between black subordinates and white bosses, were a lightening
rod for racial and class tensions. They were organizationally weaker than white management,
which retained a virtual monopoly of knowledge and experience.
Third, corporate executives in the industry had long fought for raising the level of the
color bar and replacing white with black since this reduced labor costs. If they faced
organized resistance from white staff before, now they were threatened by exodus. Fourth,
the Zambian government regarded the mining industry as a sacred cow, the source of
revenue for its “nation-building” projects. It did not dare jeopardize profits from copper.
Moreover, it was content to let expatriates run the industry because, although they had
economic power, they did not pose a political threat. They were on limited three-year
contracts that could be terminated at will. Zambian managers, as a powerful fraction of the
dominant class, could pose many more problems for a Zambian government. This balance
of forces meant that, despite independence, the overall class and racial patterns in the
mines did not change substantially.
From the microworlds of Zambianization I “extended out” to the sources of underdevelopment.
Obstacles to development arose not only from dependence on copper in a world
economy controlled by advanced capitalist nations but also from the reproduction of class
relations inherited from colonialism. An emergent African “national bourgeoisie” had class
interests in a racial order that inhibited economic transformation. My study had, thus,
reconstructed and reconfigured indigenous narratives into a class analysis of postcolonialism
which, as we shall see, fed back into society in unanticipated ways.
THE EXTENDED CASE METHOD 9
II. POSITIVE SCIENCE REVISITED
What is positive science? For August Comte, sociology was to replace metaphysics and
uncover empirical laws of society. It was the last discipline to enter the kingdom of science
but once admitted it would rule over the unruly, producing order and progress out of chaos.
Thus, positivism is at once science and ideology. Today sociology has, for the most part,
dropped its pretensions to a ruling ideology and we can call this stripped down version of
positivism simply positive science. The premise that distinguishes positive from reflexive
science is that there is an “external” world that can be construed as separate from and
incommensurable with those who study it. Alvin Gouldner (1970) once called this premise
“methodological dualism”—social scientists are exempt from the theories they develop
about others. Positive science calls for the distancing of observer from the object of study,
a disposition of detachment. The purpose of positive science is to produce the most accurate
mapping of the workings of this external world, to mirror the world (Rorty 1979).
Constituting the observer as outsider requires an effort of estrangement, facilitated by
procedural objectivity. In his exemplary discussion of “analytic fieldwork,” Jack Katz
(1983) lays out the “4Rs,” what I refer to as the four prescriptive tenets of positive science.
First, sociologists must avoid affecting and thus “distorting” the worlds they study. This is
the injunction against reactivity. Second, the external world is an infinite manifold, so we
need criteria for selecting data. This is the principle of reliability. Third, the code of selection
should be formulated unambiguously so that any other social scientist studying the
same phenomena could produce the same results. This is the principle of replicability.
Fourth, we must guarantee that the slice of the world we examine is typical of the whole.
This is the principle of representativeness.
Katz accepts these principles as definitive of social science. He tries to show how
participant observation can live up to the 4Rs if it follows “analytic induction,” or what he
prefers to call “analytic research.” However, in the process he radically destabilizes his
methodological principles, embracing rather than proscribing “reactivity,” dissolving the
boundary between fact and fiction, and summoning readers to replicate findings from their
own experiences. Still he holds on to the 4Rs. I take the opposite tack, forsaking positive
science for “reflexive” science, more appropriate to the extended case method. I justify
invoking and elaborating this alternative by first showing how the extended case method
violates the 4Rs, and then how even survey research fails to live up to those same positive
principles. My intention here is not to reject positive science but to show how positive
science rejects the extended case method and in particular my study of Zambianization.
Positive Science Violated
The extended case method makes no pretense to positive science. First, my Zambianization
research broke the injunction against reactivity. I was anything but a nonintervening
observer. I entered the Personnel Research Unit just as it was undertaking a mammoth job
evaluation exercise to categorize the complex industry-wide occupational structure with a
view to bringing white and black pay structures into a single hierarchy. It was critical that
the job hierarchy already established within each racial group should be maintained. In
order to give the impression of “fairness,” integrating the two pay scales was based on a
joint team of “experts” from union and management “evaluating” each job according to a
pregiven set of characteristics, experience, education, dexterity, effort, and so forth. An
English consulting company, brought in to attempt to match the evaluation of jobs with the
preexisting hierarchy, failed abysmally. With my mathematical training I was able to turn
10 SOCIOLOGICAL THEORY
the task into a simple problem of linear programming and thereby helped to reproduce the
very racial order that became the focus of my research in The Colour of Class.
Reliability was also violated. Having a fixed coda or prism through which to observe
and extract information makes one unresponsive to the flux of everyday life. Living in the
time and space of those one studies makes it difficult to fit the world into a predefined
template. One begins with one set of questions and ends with very different ones. Thus, I
entered the mining industry in search of some company policy guiding relations with the
Zambian government. It was only by working for the company executives that I realized
that there was no such policy. Nor was it rational, as I subsequently realized, to follow a
predetermined strategy in situations of great uncertainty—political uncertainty (frequent
government crises, changes in ministerial personnel, or surprise moves such as the nationalization
of the mines); economic uncertainty (especially the volatile world price of copper);
and technical uncertainty (unexpected problems of excavation, mine falls). In such a
turbulent environment managers need to be flexible and not be hamstrung by detailed
plans. As I discovered, those policies that did exist were constructed in post-hoc fashion,
by “experts” like myself, to justify decisions already made. Had I not been a participant in
these processes I would still be looking for that elusive company policy, or more likely
would have concocted a policy from company rationalizations. In short, with the extended
case method, dialogue between participant and observer provides an ever-changing sieve
for collecting data. This is not to deny that we come to the field with presuppositions,
questions, and frameworks but that they are more like prisms than templates and they are
emergent rather than fixed.
By the same token replicability was also problematic. The data I gathered was very
much contingent on who I was—a white male recently graduated from a British university
with a degree in mathematics, a newcomer to colonialism, and an idealist to boot. Every
one of these characteristics shaped my entry and performance in social situations and how
people spoke to me of racial issues. More than that, anyone who replicated my study of
Zambianization at a subsequent point in time would come up with very different observations.
History is not a laboratory experiment that can be replicated again and again under
the same conditions. There is something ineffably unique about the ethnographic encounter.
It certainly would have been interesting for someone else to repeat the study, either
simultaneously or subsequently, not as a replication but as an extension of my own study.2
And so, of course, we come to the inevitable question of representativeness. How representative
were my observations of the process of Zambianization within my two cases?
How representative were my case studies of all the possible case studies at the one mine I
studied, let alone of the six other mines or indeed of industries beyond? How could I draw
any conclusions beyond my two unique cases? And if I could not generalize, why did I
bother to devote three and a half years to the study!
These are valid criticisms from the standpoint of positive science, and if this were the
only model of science I would indeed have wasted my time. However, there is a second
approach to science, a reflexive approach that also seeks generalizable and falsifiable
explanations. This alternative is not drummed up out of thin air but, true to its own principles,
arises from a critical engagement with positive science. But first I must show that
no method, not even the best survey research, can live up to positive principles, for it is
from this irrevocable gap between positive theory and its practice that the principles of
reflexive science spring.
2
In another study (Burawoy 1979), this time of a factory in South Chicago, I found myself in the same plant that
had been studied by another sociologist thirty years earlier. I could have tried to show why his theory of “output
restriction” was wrong but instead I used it as a baseline from which to extend my own study back into history.
THE EXTENDED CASE METHOD 11
Positive Science Delimited
Survey research is avowedly positive in its method. It tries to live up to the 4Rs by delivering
the 4Ss. First, in order to overcome the problem of reactivity, the interview is constructed
as a uniform, neutral stimulus that elicits variegated responses. The respondent is
supposed to react to the question and the question alone, stripped of the medium in which
it is posed. Second, to confront the problem of reliability and achieve a consistent set of
criteria for the selection of data, the interview is standardized; identical questions are
asked in identical ways of each respondent. Third, for replicability not only has the question
to be a stimulus “isolated” from the interview but the external conditions must be
controlled, that is stabilized or deemed irrelevant. Finally, for representativeness the respondents
must be a carefully selected sample of the broader targeted population.
Despite their best efforts, survey researchers have always and inevitably fallen short of
their positive goals. The interview is a social context, embedded in other contexts, all of
which lend meaning to and are independent of the question itself. There are four types of
context effects. First, there are well-documented interview effects, which create the problem
of reactivity, in which interviewer characteristics (for example, race or gender) or the
interview schedule itself (for example, order or form of questions) significantly affect
responses (Hyman et al. 1954; Converse and Schuman 1974; Schuman and Presser 1981).
Second, there are respondent effects, in which the meaning of questions have an irreducible
ambiguity, dependent on the different worlds from which the respondents come. Standardization
of questions cannot eliminate respondent effects (Cicourel 1967; Forsyth and
Lessler 1991). Third, there are field effects, which simply recognize that interviews cannot
be isolated from the political, social, and economic contexts within which they take place.
Responses to interviews conducted at different points in time or in different places will be
shaped by such extraneous conditions. Replication is thwarted by external factors we do
not control. We cannot even disentangle their unmediated impact on the respondent from
their mediated impact via the interview itself.3
Finally, there are situation effects, which
threaten the principle of representativeness. Insofar as meaning, attitudes, and even knowledge
do not reside with individuals but are constituted in social situations,4
then we should
be sampling from a population of social situations and not a population of individuals
(Stinchcombe 1980). But we have no idea how to determine the population of relevant
social situations, let alone how to draw a sample.
There is nothing new here—serious survey researchers spend their lives trying to minimize
and0or control for context effects, assuming them to be noise that can be investigated
if not expurgated. If early research into survey research simply revealed interview
effects, more recent work has begun to theorize those effects (Suchman and Jordan 1990;
Schaeffer 1991; Tanur 1992). The interview is viewed as a distorted conversation in which
one of the interlocutors is absent (the researcher), in which the conversation follows a
3
Just how difficult it is to control context effects can be seen in ethnographically sensitive survey research. In
order to reduce “interview effects” survey research matches the race of interviewer and interviewee, but this can
exaggerate “respondent effects” and “field effects.” Sanders (1995) shows that the wider racial field invades the
interview so much that some black respondents imputed “whiteness” to their “black” telephone interviewers.
Moreover, those blacks who identified their interviewer as white adopted more conciliatory attitudes. In their pen
experiment, Bischoping and Schuman (1992) show that the divergent polling results prior to the Nicaraguan
election of 1991 was due to the respondents’ perceived partisanship of the polling organization. They conclude
that this was an artifact of the polarized political situation in Nicaragua but exactly how that field affected the
responses remained unclear.
4
Here we are appealing to a methodological situationism (Knorr-Cetina 1981; Cicourel 1964) to replace a
methodological individualism. Survey researchers might try to build in social situation as a “variable,” examining,
for example, how a person’s race is affected by situation, but that is very different from methodological
situationism in which the situation rather than the individual is the unit of analysis. Thus, Cicourel (1982) raises
the problem of “referentiality”—what can we know about a given situation from a conversation that takes place
in another situation.
12 SOCIOLOGICAL THEORY
predetermined trajectory with prescribed responses, and in which dialogue is precluded
(Clark and Schober 1992).5
Unable to establish common ground with the respondent, the
interviewer cannot avoid misunderstandings and mistakes. One response, therefore, is to
move toward a more “narrative” interview. Instead of foisting the standardized interview
on respondents, the interviewer allows respondents to tell their own story, to offer their own
“narrative” (Mishler 1986). The interviewer proceeds through dialogue, reducing distortion
but at the expense of reactivity, reliability, replicability, and often representativeness.
In other words, no one denies the importance of context effects. Survey researchers
look upon them as a challenge—they must be measured, reduced, and controlled. However,
if one takes the view that context is not noise disguising reality but reality itself, then
improving survey research is tackling the wrong problem with the wrong tools. Thus,
many regard the ineluctability of context effects as a demonstration of the irremediable
flaws of positive science, justifying abandoning science altogether in favor of an interpretive
approach to the social world. We can find influential representatives of this “hermeneutic”
school across the disciplines: philosophers such as Hans Gadamer and Richard
Rorty (1979) reduce social science to dialogue and conversation; anthropologists such as
Clifford Geertz (1973, 1983) regard the art of ethnography as thick description or the
excavation of local knowledge; sociologists such as Zygmunt Bauman (1987) argue that
intellectuals should abandon their legislative pretensions for an interpretive role, mediating
between communities; feminists such as Donna Haraway (1991, chapter 9) call for
networks of “situated knowledges.”
This is not the approach I propose to follow here. Faced with the ineluctable gap between
positive principles and research practice, I neither abandon science altogether nor resign
myself to refining practice to approach unachievable positive principles. Instead, I propose
an alternative model of science, a reflexive science, that takes context as point of departure
but not point of conclusion.
III. REFLEXIVE SCIENCE DEFINED
Reflexivity in the social sciences is frequently regarded as the enemy of science. Long ago
Peter Winch (1958) argued that individual reflexivity, that is the self-monitoring of behavior,
leads to an irrevocable uncertainty in human action, making scientific prediction impossible.
All social science can do is reveal the discursive and nondiscursive worlds of the
people it studies. Similar views have become common in anthropology wherever the “linguistic”
or “interpretive” turn has taken hold. In its extreme form we are so bound by our
own preconceptions that we can do little more than gaze into our biographies. Within
sociology reflexivity has been put to more positive use. Alvin Gouldner (1970) turned
sociology onto itself to uncover the “domain assumptions” of reigning paradigms in “Western”
sociology, arguing that they were increasingly out of sync with the world they claimed
to mirror. More recently Pierre Bourdieu (1977, 1990), with the help of his interlocutor
Loïc Wacquant (Bourdieu and Wacquant 1992), invites us to a reflexive sociology that
5
Sniderman and Piazza (1993) try to build dialogue into their surveys by presenting respondents with predetermined
counterarguments. For example, respondents are first asked whether they approve of government support
for blacks. If respondents approve of spending increases then they are asked whether they would feel the same
way if blacks were singled out for special treatment. If, on the other hand, respondents do not approve of more
spending they are asked if they would feel the same if this meant that blacks would continue to be poorer than
whites. Their data show that 44 percent of whites were “talked out” of their original position. In the case of
affirmative action only 20 percent changed their minds in the face of counterarguments. It is not clear why there
should be such changes, whether Sniderman and Piazza are tapping context specific attitudes, whether attitudes
of whites toward race are pliable and superficial, or whether this is simply an artifact of the interview situation
itself in which the respondent flows with an expected answer. Whatever else, these changes in responses suggest
the importance of studying the interview itself as a social situation.
THE EXTENDED CASE METHOD 13
explicitly seeks to deepen the scientific foundations of sociology. Recognizing our own
place within the disciplinary field enables us to objectify our relation to those we study,
which will make us better scientists.
I take a slightly different approach. Rather than arguing that there is one model of
science that is best carried out with reflexive awareness, I propose a methodological duality,
the coexistence and interdependence of two models of science—positive and reflex-
ive.6
Where positive science proposes to insulate subject from object, reflexive science
elevates dialogue as its defining principle and intersubjectivity between participant and
observer as its premise. It enjoins what positive science separates: participant and observer,
knowledge and social situation, situation and its field of location, folk theory and academic
theory. The principles of this reflexive science can be derived from the context
effects that pose as impediments to positive science.
1. Intervention
The first context discussed above was the interview itself, which is not simply a stimulus
to reveal the true state of the interviewee but an intervention into her life. The interview
extracts her from her own space and time and subjects her to the space and time of the
interviewer. In the view of reflexive science, intervention is not only an unavoidable part
of social research but a virtue to be exploited. It is by mutual reaction that we discover the
properties of the social order. Interventions create perturbations that are not noise to be
expurgated but music to be appreciated, transmitting the hidden secrets of the participant’s
world. Institutions reveal much about themselves when under stress or in crisis, when they
face the unexpected as well as the routine. Instead of the prohibition against reactivity,
which can never be realized, reflexive science prescribes and takes advantage of intervention.
2. Process
The second context effect is the multiple meanings attached to the interviewer’s “stimulus,”
which undermines the reliability of research. One can standardize the question but
not the respondent’s interpretation of the question. Respondents come to the interview
with multiple experiences derived from different situations that they are then asked to
collapse into a single data point. Even asking someone’s race or gender can turn out to be
complicated, requiring that the respondent reduce a diverse array of experiences to a single
item of a check list. There is a double reduction: first aggregation and then the condensation
of experience.
Reflexive science commands the observer to unpack those situational experiences by
moving with the participants through their space and time. The move may be virtual, as for
example in historical interpretation; real, as in participant observation; or some combination
of the two, as in the clinical interview. But there is another complication. Not only
does each situational experience produce its own “situational knowledge,” but that knowl-
6
This distinction can be extended to the natural sciences. There are philosophers of the natural sciences, such
as Michael Polanyi (1958), who refuse the separation of subject and object. His theory of personal knowledge
gives centrality to the natural scientist who makes contact with and dwells in “nature.” Similarly, Evelyn Fox
Keller (1983, 1985) makes the case that natural scientists, like social scientists, may also be part of the world they
study, that they have a human relation to the objects under investigation. In her feminist view, what is distinctive
is not the objects of science but the gendered way we approach them. Finally, from a realist standpoint, Roy
Bhaskar (1979) insists on intervention and experiment as central to both the natural and social sciences. The
distinction between reflexive and positive science does not have ontological foundations; it does not depend on
the nature of the world being studied. The distinction between the two models lies not in its object (human as
opposed to nonhuman) but in the relation of scientist to object.
14 SOCIOLOGICAL THEORY
edge may be discursive or nondiscursive. If the discursive dimension of social interaction,
what we may call narrative, can be reached through interview, the nondiscursive, that is
the unexplicated, unacknowledged, or tacit knowledge, sometimes referred to as practical
consciousness, which underlies all social interaction, calls for more. It may be discovered
through “analysis,” for example, or through participation, “doing” things with and to those
who are being studied (Garfinkel 1967).
The task of reflexive science does not stop with situational comprehension, with the
recovery of situational knowledge. First, there are always multiple knowledges, reflecting
the position of different actors within a social situation. Reflexive science would be impossibly
cumbersome if its goal were the display of multiple narratives, multiple voices. But
worse still, situational knowledge is knowledge located in a specific space and time. Neither
space nor time can be frozen and so situational knowledges are in continual flux.
Therefore, like any other science, reflexive science has to perform some reduction. In this
instance the reduction is an aggregation—the aggregation of situational knowledge into
social process. Just as survey research aggregates data points from a large number of cases
into statistical distributions from which causal inferences can be made, reflexive science
collects multiple readings of a single case and aggregates them into social processes. The
move from situation to process is accomplished differently in different reflexive methods
but it is always reliant on prior theory. Below we will see how it works with the extended
case method.
3. Structuration
The third context is the external field within which the interview occurs. The field cannot
be held constant, so the purpose of replication is thwarted. It is not simply that social
scientists shape the world they study in idiosyncratic and therefore nonreplicable ways but
that the external field has its own autonomous dynamic. This wider field of relations
cannot be bracketed or suspended, yet it is also beyond the purview of participant observation.
We therefore look upon the external field as the conditions of existence of the
locale within which research occurs. We therefore move beyond social processes to delineate
the social forces that impress themselves on the ethnographic locale. These social
forces are the effects of other social processes that for the most part lie outside the realm
of investigation. Viewed as external to the observer these social forces can be studied with
positive methods that become the handmaidens of reflexive science.7
Reflexive science insists, therefore, on studying the everyday world from the standpoint
of its structuration, that is by regarding it as simultaneously shaped by and shaping
an external field of forces.8
This force field may have systemic features of its own, operating
with its own principles of coordination and contradiction, and its own dynamics, as
it imposes itself on multiple locales.
7
In other words I follow Abbott (1992a, 1992b) and Somers and Gibson (1994) in distinguishing the “narrative”
of social process from the causality of social forces, but where they want to replace the second with the first,
I insist on retaining a place for social forces as methodological expedient and experiential reality framing and
confining social processes.
8
Anthony Giddens (1984) has made structuration the leitmotif of his work. He seeks to transcend the dualism
of subject and object, agency and structure, micro and macro by substituting the notion of duality in which
practices simultaneously reproduce the conditions that enable them. He stresses how structure facilitates rather
than constrains action, much as language allows speech. In the end, intuitive notions of structure evaporate and
we are left with a voluntarist vision that emphasizes the control we exercise over our worlds. I return to a more
conventional notion of structuration in which “structure,” or “social forces,” really do confine what is possible,
although they are themselves continually reconfigured. What he understands as “structuration” is closer to what
I call “process,” but even here I will give more centrality to structures of micropower that are beyond the control
of individuals.
THE EXTENDED CASE METHOD 15
4. Reconstruction
The fourth context effect relates back to the second, the priority of the social situation over
the individual, which problematizes sampling on the basis of individuals. If representation
is not feasible, is there any other way of producing generality? Instead of inferring generality
directly from data, we can move from one generality to another, to more inclusive
generality. We begin with our favorite theory but seek not confirmations but refutations
that inspire us to deepen that theory. Instead of discovering grounded theory we elaborate
existing theory.9
We do not worry about the uniqueness of our case since we are not as
interested in its “representativeness” as its contribution to “reconstructing” theory.10
Our
theoretical point of departure can span the range from the folk theory of participants to an
abstract law. We require only that the scientist consider it worth developing.
But what distinguishes a “progressive” rather than a “degenerate” reconstruction? Following
Karl Popper (1963, chapter 10) and Imre Lakatos (1978) we seek reconstructions
that leave core postulates intact, that do as well as the preexisting theory upon which they
are built, and that absorb anomalies with parsimony, offering novel angles of vision. Finally,
reconstructions should lead to surprising predictions, some of which are corroborated.
These are heavy demands that are rarely realized but ones that should guide progressive
reconstruction of theory.
Dialogue is the unifying principle of reflexive science. It is dialogical in each of its four
dimensions. It calls for intervention of the observer in the life of the participant; it demands
an analysis of interaction within social situations; it uncovers local processes in a relation
of mutual determination with external social forces; and it regards theory as emerging not
only in dialogue between participant and observer, but also among observers now viewed
as participants in a scientific community. Theories do not spring tabula rasa from the data
but are carried forward through intellectual debate and division. They then reenter the
wider world of participants, there to be adopted, refuted, and extended in intended and
unintended ways, circulating back into science.11
Science offers no final truth, no certainties,
but exists in a state of continual revision.
IV. THE EXTENDED CASE METHOD
Reflexive science is to the extended case method what positive science is to survey research—
the relation of a model to method, legitimating principle to situated practice. Just as we
codified survey research so we must now do the same for the extended case method. I
return to my Zambianization study to illustrate the extended case method, pointing to ways
in which it might have benefited from greater methodological self-consciousness. In Section
V, I will use my case study in the opposite way, to cast light on inherent limitations of
reflexive science.
1. Extending the Observer to the Participant
In the positive view participant observation brings insight through proximity but at the
cost of distortion. The reflexive perspective embraces participation as intervention pre-
9
There is a substantial body of philosophy of science, informed by historical exploration of the growth of
knowledge, arguing that science moves forward through the absorption of anomalies within paradigms (Kuhn 1962)
or research programs (Lakatos 1978), as well as through competition among paradigms or research programs.
10
Rebecca Emigh (1997) has made the critical distinction between “deviant case” analysis, in which the outliers
increase the generalizability of our theory, and “negative case” analysis, which increases the “empirical
content” of theory, what I have called theory reconstruction.
11
Again, Anthony Giddens (1992) has made much of this interchange between academic and lay theory, arguing
that sociology appears not to advance because its discoveries become conventional wisdom. The reflexivity
of social theory, he argues, is one of the distinctive features of modernity.
16 SOCIOLOGICAL THEORY
cisely because it distorts and disturbs. A social order reveals itself in the way it responds to
pressure. Even the most passive observer produces ripples worthy of examination, while
the activist who seeks to transform the world can learn much from its obduracy.12
The most seismic interventions are often entry into and departure from the field. Any
group will often put up a great deal of formal and informal resistance to being studied at
close quarters—resistance that discloses much about the core values and interests of its
members as well as its capacity to ward off danger. Leaving the field is also an “intervention”
since it is then that participants often declare well-kept secrets, or pose revealing questions
that they had never dared ask before. But the biggest bombshell often comes when
outsiders return their findings to the participants. There are many reasons for this. There are
few who like to be partialized, reduced to reified forces or in any other way made an object
of sociological research. Second, most communities are riven by conflicts so that it is
impossible to navigate them to everyone’s satisfaction no matter how careful the observer.
However painful, ethnographers always learn a great deal from their final intervention.
When I had completed my study of Zambianization I decided to seek permission for
publication from the top executives of Anglo American who had first employed me and
then sponsored the research I had conducted on the mines. They had no idea that I had been
studying Zambianization for three years. When I showed them my report they were shocked
and dismayed that I had dared to broach such a “sensitive” issue. After reading the manuscript
they bluntly refused to allow publication on the grounds that it was politically explosive.
I countered that the report was based on their own data. They finally threw me a token
concession. Since the mines had just been nationalized, the publication decision was no
longer their’s but a government responsibility. I took my manuscript to the person responsible
for Zambianization. He was an expatriate, new to the job but not to the mines, who
saw the report as a way of making his mark by challenging the practices of the mining
companies. Based as it was on careful, detailed, inside research he considered it a powerful
weapon to advance Zambianization. “Because it criticizes the government, the trade unions,
the Zambian successor, the expatriates and the corporations; because it criticizes everyone
it must be objective,” he said!
The monograph was duly published by the Institute for African Studies at the University
of Zambia. It received a lot of publicity in the media. Its class analysis was hostile to
the mining companies as well as to the government and expatriates. Yet corporate managers
in Lusaka used it to discipline mine management on the copperbelt. The stamp of
academic certification made it an effective weapon in the hands of the mining companies—a
happy marriage of science and power.
No claims to “impartiality” can release us from either the dilemmas of being part of the
world we study or from the unintended consequences of what we write. What we write
circulates into the world we seek to comprehend and from there sprays dirt in our face. As
I shall suggest below, this response represented both a confirmation and a challenge to the
theory expounded in The Colour of Class.
2. Extending Observations over Space and Time
Such dramatic culminations of research happen in miniature every day. Ethnographers join
participants for extended periods of time as well as in different places. Each day one enters
the field, prepared to test the hypotheses generated from the previous day’s “intervention.”
Fieldwork is a sequence of experiments that continue until one’s theory is in sync with the
12
My position here is not unlike John van Maanen’s (1988) three “tales of the field”—realist tales that privilege
the participant, confessional tales that privilege the observer, and impressionist tales that accent the interaction of
the two. The latter, which is the one he favors, is parallel to the interventionist approach I am advocating here.
THE EXTENDED CASE METHOD 17
world one studies. It is a process of successive approximation that can, of course, go awry.
Wild perturbations between observations and expectations signify poor understanding,
while occasional shocks force one into a healthy rethinking of emergent theorizing. At this
level theorizing is compiling situational knowledge into an account of social process. How
does this work?
Situations involve relations of copresence, providing the conditions for practices that
reproduce relations. The archetype of this conceptualization of social situations is the
Marxian treatment of production. As workers transform nature into useful things, so they
simultaneously produce their own means of existence (necessary labor) and the basis of
profit (surplus labor), that is they reproduce the worker on one side and the capitalist on the
other. But this process continues; that is, laborers return the next day, because they have no
alternative source of survival. They are therefore subject to the power of capital, or what I
have called the political regime of production, which regulates the division of labor, the
mobility between positions in the division of labor, rewards, and so on. The point is simple:
production becomes reproduction only under a particular structure of power.13
We can
compile situational knowledge into an account of social process because regimes of power
structure situations into processes.
This can be applied to my case study. Zambianization takes place under the erosion of
“colonial despotism” toward a less punitive production regime but one still based on the
color bar. Working with the vocabulary of Giddens (1984) and Sewell (1992), one can say
that, within this political regime, resources (money, skill, education, prestige, etc.) are
distributed along racial lines supported by schemas (norms, beliefs, theories, etc.) of racial
supremacy. The Zambianization process is set in motion when a Zambian is promoted to
replace an expatriate. The expatriate seeks to preserve his job (a resource) and looks upon
the new incumbent as inferior to himself (schema). Management intervenes to open a new
job for the expatriate, who takes with him some of his old authority and responsibility,
leaving his successor with fewer resources. The successor’s subordinates, seeing him as a
diminished version of his predecessor, withdraw their support and confidence. Unable or
unwilling to seek support from his white boss, the new Zambian supervisor resorts to more
authoritarian rule, which confirms his subordinates’ worst suspicions. In their view the
new Zambian successor is worse than his white predecessor—he is trying to recreate the
despotism of the past. Subordinates further withdraw cooperation and the cycle continues
until a new equilibrium of force and consent are reached. The regime of power, that is the
color bar, is reproduced.
Three issues are noteworthy. First, a social situation becomes a social process because
social action presupposes and reproduces its regime of power. It is by participating in
terms of the color bar that the color bar is reproduced. Second, in the struggles around the
regime of power, history and macrostructures are invoked as resources and schema within
the social situation. The Zambian successor complains that whites continue to rule the
roost, that independence has brought no change. Zambian workers see their new black
boss as recreating the despotic past or imposing a new tribal supremacy. Third, interventions
from outside the social situation have consequences structured by the regime of
power. Management may create positions for displaced expatriates as “aids” to the Zambian
successor but the effect is to weaken him. Management may recruit high school
graduates to improve the quality of personnel managers but its effect is to exacerbate
conflict between Old Timers and Young Turks.
13
Thus I am closer to Bourdieu and Foucault than Giddens and Sewell who have little to say about how power
enters into the constitution of the conditions of our existence.
18 SOCIOLOGICAL THEORY
The reproduction of the color bar causes hierarchical social relations to change: relations
between black and white become more distant and indirect while relations between
black and black become more tense and conflictual. Reproduction of the regime of power
is assured from the inside through the deployment of resources and schemas. It is also
reproduced from the outside, beyond the realm of participant observation, but this requires
the analysis of social forces.
3. Extending Out from Process to Force
I could have closed my study of Zambianization with a demonstration of the general law of
the color bar: however the organization changes, authority always flows from white to
black. I could have given the law even more power by drawing on evidence from the very
different context of the United States, where gender and racial lines also have an uncanny
way of reproducing themselves.14
This would be the strategy of inductive generalization,
namely to seek out common patterns among diverse cases, so that context can be discounted.
This might be called the segregative or horizontal approach, in which cases are
aggregated as though they were independent atoms. The extended case method, on the
other hand, deploys a different comparative strategy, tracing the source of small difference
to external forces. This might be called the integrative or vertical approach. Here the
purpose of the comparison is to causally connect the cases. Instead of reducing cases to
instances of a general law, we make each case work in its connection to other cases.
The Colour of Class offered two such connected comparisons. The dominant one was a
comparison of Zambianization after independence with African Advancement under colonial
rule. The second, much less developed, compared the bottom up Zambianization of
the mines with the top down Zambianization of government. In order to understand why
the color bar remained on the copperbelt despite democratization and the formal dissolution
of racism, I dug back into history. Under colonial rule the mining companies had
persistently tried to “advance Africans” into positions hitherto monopolized by whites.
What little was accomplished took place through job fragmentation and deskilling of white
jobs. African trade unions were always ambivalent about this view of African Advancement
since the majority of their members were more interested in wage increases and
improved working conditions. The colonial regime was pressured by the mining companies
and the colonial office in London to support gradual African Advancement, as much
as a safety valve for frustrated aspirations as for profit. The white settler community was
an influential counterweight that opposed any upward mobility for Africans. For the most
part, the colonial state tried to keep out of the fray, entering only as adjudicator when the
machinery of industrial relations broke down.
The successor Zambian government, no longer tied to London, became even more
beholden to the mining companies as a major source of revenue. While white managers
lost their formal political power, their leverage remained since the mines depended upon
their expertise. For its part, the Zambian political elite retained expatriates in the commanding
heights of the copper industry because it did not want to depend on an indigenous,
potentially rival, economic elite. Still, the postcolonial government had to respond
to nationalist clamor that Zambians run their own country. It did so, not by a more vigorous
pursuit of Zambianization but by nationalizing the mines, which left internal organization
14
There is a large literature here starting from Rosabeth Kanter’s (1977) analysis of organization processes to
Ruth Milkman’s (1987) analysis of the forces shaping the position of the gender line to Linda Blum’s (1991)
class analysis of the contending forces of affirmative action and comparable worth (parallel to the two meanings
of African Advancement).
THE EXTENDED CASE METHOD 19
untouched. Zambianization from above in the capital propelled Zambianization from below
on the copperbelt.
Far from being independent, the two cases inversely determine each other. The roots of
color bar persistence on the copperbelt lie in its erosion within government. This is the
principle of structuration—locating social processes at the site of research in a relation of
mutual determination with an external field of social forces. But can we go further and ask
whether these extralocal forces exhibit a processual character of their own? Do they have
a certain “systematicity” that tends to reproduce itself? Once more we can only proceed to
such questions with the aid of theory, in this case Marxist theory. The Colour of Class
partook in a debate about the capitalist state, arguing that the postcolonial state preserved
the overall class structure not because it was an instrument of capital but because it was
institutionally autonomous from but dependent upon capital. Here was an emergent understanding
of the structuring of class forces—a tendency for them to be reproduced domestically
on the basis of a national regime of power.
I could have extended the principle of structuration by regarding the arrangement of
state and classes within Zambia as a structured process nested in an external constellation
of international forces. Instead I stopped at the national level and looked upon “international
forces” not as constraints but as resources mobilized by the ruling elite to legitimate its
domination. The new African elite focused on forces beyond national control—terms of
trade, price of copper, Western experts, transnational corporations—in order to obscure the
class character of postcolonialism. The African governing class deployed “neocolonialism”
in their own version of the extended case method, denying their own class power by
claiming impotence before external forces. This perspective of the new elites found its
representative within academic discourse as underdevelopment theory, popularized by Paul
Baran and then Gundar Frank. Later it would be challenged by comparative studies that
focused on the capacity of the state to engineer “dependent development” within a changing
world economy. The debate continues today with the emphatic rejection of the entire
“developmentalist” project as destructive of third world countries (Escobar 1995).
However, my interest at the time lay in confronting “neocolonialism” and underdevelopment
theory with class analysis, which confined both the local and the extralocal to
national boundaries. Looking back now I underestimated the importance of international
forces. Zambia’s dependence on a single commodity, copper, whose price has continued to
fall on world markets, brought it under the spell of the IMF and its structural adjustment
programs. Twenty-five years after nationalizing the copper mines, the Zambian government
is now trying to sell them off, to reprivatize them. They have brought back expatriate
managers to make the mines more attractive to foreign investors. The Zambian economy is
being recolonized at the behest of its own African government.
4. Extending Theory
Our first three “extensions”—intervention, process, and structuration—all call for prior
theory. But our stance toward theory is kamikaze. In our fieldwork we do not look for
confirmations but for theory’s refutations. We need first the courage of our convictions,
then the courage to challenge our convictions, and finally the imagination to sustain our
courage with theoretical reconstruction. If these reconstructions come at too great a cost
we may have to abandon our theory altogether and start afresh with a new, interesting
theory for which our case is once more an anomaly.
I was not methodologically self-conscious about theory extension in The Colour of
Class, but the strategy pervaded the monograph. The very concept of succession was
20 SOCIOLOGICAL THEORY
drawn from Alvin Gouldner’s (1954) case study of the organizational reverberations of a
managerial succession.15
But where his was a “natural succession,” Zambianization was a
case of “forced succession,” imposed from above and resisted from below. The Zambian
successor had to contend with suspicion from his subordinates, resistance or indifference
from his supervisor, as well as his own doubts about his abilities.
Theorization of social process was extended to theorization of the broader social forces.
First, I deconstructed the government’s Zambianization report. Hidden behind its data lay
the real processes of forced succession under the color bar principle. Contrary to the implications
of the report, expatriates were as firmly in control of the industry as ever. On the
other hand, I drew back from the “neocolonial” thesis that blamed Zambia’s continued backwardness
on a conspiracy of international forces.Again the point was not that the claims were
wrong—obviously Zambia was held in the vice of multinationals and international trade—
but rather that their partiality obscured the class interests of the new ruling elite.
I was more forthright in rejecting theories that attributed underdevelopment to the cultural
backwardness of Zambian workers or, as was more common, to their anomic and
undisciplined industrial behavior. Robert Bates (1971), for example, claimed that the postindependence
Zambian government had failed to control the mineworkers. However, careful
examination of his and other data on productivity, absenteeism, turnover, disciplinary cases,
and strikes provided no basis for his claims. He had simply adopted management’s and
government’s class ideology of the “lazy Zambian worker,” blaming workers for the inefficiencies
and conflicts whose sources lay elsewhere, such as in the continuing color bar
(Burawoy 1972b).
Frantz Fanon’s ([1952] 1968a, [1961] 1968b) theory of the “postcolonial revolution”
guided my analysis. Although I was not as explicit in my reconstruction as I would be now,
I sought to extend his theory to Zambia, a colony without a peasant-based national liberation
struggle. My analysis of the multinationals, mineworkers, Zambian managers, and
expatriates paralleled his dissection of the class interests of the national bourgeoisie, intellectuals,
and the peasantry. I turned the government’s claims of worker indiscipline, indolence,
and anomie against the new ruling elite itself, whose extravagance and selfindulgence
emanated from rapid upward mobility. As to the mineworkers themselves, they
were the prototype of Fanon’s labor aristocracy. They pursued their narrow economic
interests, showed little concern for the color bar, and saw nationalization of the mines as a
government ruse to impose harsher discipline. The Colour of Class did more than recast
Fanon’s class categories, it set the class map in motion by connecting the macroforces,
propelling the movement from African Advancement to Zambianization, to the microprocesses
of succession.
Theory is essential to each dimension of the extended case method. It guides interventions,
it constitutes situated knowledges into social processes, and it locates those social
processes in their wider context of determination. Moreover, theory is not something stored
up in the academy but itself becomes an intervention into the world it seeks to comprehend.
Indeed, The Colour of Class became its own self-refuting prophecy. My man in the
ministry, then the media, and finally the mining companies all set out to change the world
I had described. They sought to overturn the new governing elite’s interest in reproducing
the color bar on the copperbelt.
15
Starting from tensions within Weber’s analysis of bureaucracy and refusing Weber’s monolithic characterization,
Gouldner develops three types of bureaucracy—mock, representative, and punishment centered. In so
doing Gouldner brackets the context of his gypsum plant and misses the historical specificity of his ideal types.
The extended case method would have tried to locate the plant in its political, economic, and geographical
context (Burawoy 1982).
THE EXTENDED CASE METHOD 21
This refutation, like any other, is not cause for theoretical dejection but an opportunity
for theoretical expansion. The forces revealed in my publication efforts corroborated the
view of the mining companies as flexibly adapting to government initiatives. Yet they also
showed that the government did not always turn a blind eye to the continuation of the color
bar, that the interests of the postcolonial state were not as homogeneous as I presented
them, and that social forces are the contingent outcome of social processes. In the positive
mode social science stands back and observes the world it studies, whereas in the reflexive
mode social theory intervenes in the world it seeks to grasp, destabilizing its own analysis.
V. THE EFFECTS OF POWER
In defending reflexive science and the extended case method, I am not laying claim to any
panacea. Just as there is an insurmountable hiatus between survey research and the positive
model it seeks to emulate, so a similar hiatus separates the extended case method and
the principles of reflexive science. Whereas the first hiatus is the result of context effects,
the second is due to the effects of power. Intervention, process, structuration, and reconstruction
are threatened by domination, silencing, objectification, and normalization. However,
the self-limitations of reflexive principles due to the ubiquity of power are no more
reason to abandon the extended case method than context effects are reason to abandon
survey research. The goal is to examine those limitations in order to take them into account
and perhaps even reduce them.
1. Domination
The intervening social scientist cannot avoid domination, both dominating and being dominated.
Entry is often a prolonged and surreptitious power struggle between the intrusive
outsider and the resisting insider.16
As I hunted through their records and participated in
high level negotiations I deceived the mining companies as to my true purpose. To penetrate
the shields of the powerful the social scientist has to be lucky and0or devious; the
powerless are more vulnerable. But even they have their defenses. Thus, in making my
way to the other side of the color bar, I had to use the pretext of a survey to make contact
with Zambian personnel officers and enlist the help of Zambian students to discover the
views of unskilled and semiskilled workers. But this introduced another layer of power
within the research team—my whiteness, with all its resources, and their blackness. The
students worked underground, in the smelter, tracklaying while I conducted interviews on
surface. There was no doubt that I was the “bwana” [boss], and they worked to rule,
delivering field notes but holding back their views. I was replicating the color bar within
the research team.
Nor do domination and resistance miraculously evaporate on entering the field. The
intervening social scientist faces two interrelated moments of domination, first as participant
and second as observer. As participants in sites invested with hierarchies, competing
ideologies, and struggles over resources, we are trapped in networks of power. On whomever’s
side we are, managers or workers, white or black, men or women, we are automatically
implicated in relations of domination. As observers, no matter how we like to deceive
16
James Clifford’s (1988: chapter 2) study of French anthropologist, Marcel Griaule, highlights the strategies
of power, the panoptic techniques of surveillance that the outsider uses in documenting the recalcitrant colonized.
Ethnography depends on an unabashed power struggle between observer and participant. Clifford contrasts this
with Griaule’s subsequent initiation into Dogon life by one of its chiefs. He becomes the interpreter of “authentic”
Dogon culture, an ambassador who would defend their interests in a colonial world. From willful resister and
liar the informant becomes colleague and teacher. But in neither case is there joint symmetrical construction of an
ethnographic portrait. Power suffuses both dramaturgies.
22 SOCIOLOGICAL THEORY
ourselves, we are on “our own side,” always there for ulterior reasons (Gouldner 1968).
Our mission may be noble—broadening social movements, promoting social justice, challenging
the horizons of everyday life—but there is no escaping the elementary divergence
between intellectuals, no matter how organic, and the interests of their declared constituency.
In short, relations of domination may not be as blatant as they were in the raw racial
and class order of the Zambian copperbelt, but they are nevertheless always there to render
our knowledge partial.
2. Silencing
This brings up the second face of power—silencing. Ruling ideology presents the interests
of the dominant class as the interests of all. The nationalist rhetoric of the Zambianization
report concealed diverse class and racial interests. How does one disclose this underlying
configuration of interests? As participant observers in various workplaces on and off the
mines, we registered the discordant voices of workers, expatriates, and Zambian successors.
This is the meat and potato of fieldwork. As I compiled our extended observations
made in different situations into a social process—the process of Zambianization understood
as forced succession—so these voices were reduced to, congealed into interests. I
was able to disclose the specific and conflictual interests that stood behind the rhetoric of
nationalism. But this new crystallization of interests inevitably excluded, marginalized,
and distorted other voices.
Thus, if I had been truer to the earlier Fanon of Black Skin, White Masks rather than the
later The Wretched of the Earth, I might have explored the formation of colonial subjectivities,
especially the Zambian successor, who is the prototype of Fanon’s “colonial Negro,”
caught up in a white world that rejects “him” as a racial inferior. If my own color had not
prevented it, I could have examined the way the colonial and postcolonial regimes induce
pathologies that incapacitate the successor and thereby reproduce the Manichean world of
white and black, turning African against African. Since silencing is inevitable, we must be
on the lookout for repressed or new voices to dislodge and challenge our artificially frozen
configurations, and be ready to reframe our theories to include new voices but without
dissolving into a babble.
3. Objectification
In the extended case method, the second extension from voices in social situations to
interests in social processes is followed by a third extension from interests in social processes
to the forces of social structure. Structuration involves locating social processes in
the context of their external determination. Thus, Zambianization followed the color bar,
despite being antithetical to nationalist ideology, because of the balance of external forces,
which appear all-determining. Objectification, that is hypostatizing social forces as external
and natural, is an inherent danger of this approach. There are simply limits to the
temporal and spatial reach of participant observation, beyond which we substitute forces
for processes.
Objectification is more than a methodological device, however; it also reflects the very
real power exercised by political, economic, and cultural systems over lifeworlds (Habermas
1987). But their power should not be exaggerated. First, forces are always the hypostatized
effects of concealed processes; that is, each system depends upon the shifting processes
of its own internal lifeworld. Second, lifeworlds—both those we observe directly and
those we reduce to forces—are themselves traversed by power, generating needs that escape
into the social sphere. Around such discursive need formation congeal social movements
THE EXTENDED CASE METHOD 23
that can dislodge systemic forces (Fraser 1989). Finally, systemic forces contain their own
contradictions, which burst forth unexpectedly as when my man in the ministry encouraged
a public attack on the mining industry’s conduct of Zambianization. Even as we
embrace objectification, we should be always prepared for subterranean processes to erupt
and break up the field of forces.
4. Normalization
Finally, reconstructing theory is itself a coercive process of double fitting. On the one side,
complex situations are tailored to fit a theory. The field site is reduced to a case, albeit one
that is anomalous vis-à-vis theory. On the other side, theory is then tailored to the case,
recomposed to digest the anomaly. This mutual fashioning creates an apparatus for reducing
the world to categories that can be investigated, sites that can be evaluated, people that
can be controlled.17
In order to assimilate Zambianization to a form of managerial succession, I expanded
Gouldner’s theory by introducing the distinction between natural and forced succession.
Usual attrition leads to “natural” succession but Zambianization was a forced succession.
In normalizing what was in effect a transfer of control I played straight into the hands of
the mining companies. Racial succession gave them the conceptual arsenal to discipline
their own managers. In his review of my book, Ben Magubane (1974: 598) picked up on
this normalizing effect of “succession” which overlooked the “intense but silent class
struggle of decolonization,” the fact that Zambia was being held to ransom by expatriates.
Magubane overlooked the other side of my analysis, the application of Fanon’s theory
of decolonization to the Zambian case, the extension beyond the microdynamics of Zambianization
to the class forces upholding the color bar. But here, too, normalization was at
work. It was astonishing to see how a refashioning of Fanon’s theory of postcolonialism
could be harnessed politically by the very forces it condemned. Yet one should not be
entirely surprised, given Marxism’s history as a tool of despotism.
There are formal features of Fanon’s analysis of colonialism, however, that do lend
themselves to adoption by multinational capital. He presumes, for example, the destruction
of precolonial cultures and thus the fragility of “local” or subjugated knowledges
(Lazarus 1993). I, too, gave scant attention to cultural contestation that drew sustenance
from beneath colonial regimes of power, modes of resistance discovered and celebrated by
subaltern and postcolonial studies. Challenging or tempering normalization would have
required embedding the analysis in perspectives from below, taking their categories more
seriously, and, in short, working more closely with those whose interests the study purported
to serve.18
These four power effects only add grist to the mill of postmodern critics. If context
effects demonstrate the impossibility of science, power effects show how dangerous and
self-defeating it is. But abandoning science altogether leaves power unaffected and the
hegemony of positive science untouched. Postmodernism’s dismissal of all science ignores
the pivotal distinction between positive and reflexive models.19
A self-critical positive
science concentrates on context effects but thereby obscures the functioning of power.
Constructing “detachment” and “distance” depends upon unproblematized relations of
power. A self-critical reflexive science, on the other hand, takes context for granted, but
17
The colonial encounter provides especially vivid examples of this close link between knowledge and power.
See, for example, Mitchell (1988) and Stoler (1995).
18
See, for example, Alain Touraine’s “action sociology,” which insists on social scientists working together
with participants in a social movement (Touraine 1983, 1988).
19
For a nuanced survey and evaluation of different approaches to “qualitative methods” which inclines toward
postmodern approaches but without being dogmatic, see Denzin and Lincoln (1994).
24 SOCIOLOGICAL THEORY
displays the effects of power so that they can be better understood and contained. The
limits of reflexive science lay the basis for a critical theory of society, by displaying the
limits of human freedom.
VI. THE IMPLICATIONS OF TWO MODELS OF SCIENCE
Methodological thinking can bring more than Weber claims, more than reflective understanding
of already proven practice. In codifying positive science, we subject it to immanent
critique, highlighting the gap between principles and practice. This directs our attention
not only to the possibilities of improving positive methods but also to formulating an
alternative conception of science. Table I summarizes my argument, describing the two
models of science and corresponding methods, and in each case points to the gap between
model and method. There is a circularity between the models: each takes as its own basis
the limits of the other. Positive science is limited by “context” which supplies the foundation
of reflexive science, while reflexive science is limited by “power,” the hidden premise
of positive science. Knowing the liabilities of each model-method we can work toward
their containment. If we accept this framework then we have to confront a new set of
questions and implications to which I must at least allude.
Technique, Method and Model
What is the relationship between techniques of data gathering and model-methods? Does
participant observation, that is the study of others in their space and time, have to follow
the extended case method and reflexive science? Does the interview, that is the study of
others in the interviewee’s space and time, have to follow survey research and a positive
model of science? In each case the answer is obviously no. The techniques of participant
observation and interviewing can be conducted according to either reflexive or positive
methods, as presented in Table II below.
Participant observation, conducted according to positive principles, becomes grounded
theory, which brackets involvement as bias and concentrates on deriving decontextualized
generalizations from systematic analysis of data (Glaser and Strauss 1967; Strauss 1987;
Becker 1958; Becker et al. 1961; Gans 1968). Here theory is the result and not the precondition
of research. Social scientists are outsiders and ethnographers are outsiders within,
strangers whose objectivity is vouchsafed by distance. Nonparticipant observation is preferred
over participant observation. In other words, reactivity is proscribed. To achieve
reliability ethnographers gather and analyze their data in a systematic fashion. Coding and
recoding field notes into emergent categories provides the prism for further observation.
Replication enters as a call for clarity in how categories are derived from data and is less
concerned with the replicability of data collection. It creates pressures to suspend context
so as to make cases comparable. Finally, to establish the representativeness of their results,
ethnographers should maximize variation within the field through constant comparison,
searching for extreme cases in what is called “theoretical” sampling.20
20
Elsewhere I have elaborated the distinction between the extended case method and grounded theory (Burawoy
et al. 1991: chapter 13). A contemporary exemplar of grounded theory is Martín Sánchez Jankowski’s
(1991) Islands in the Street—a ten-year study of 37 urban gangs in three metropolises. It is a remarkable,
sustained commitment to positivism. First, he constitutes himself as ethnographer and outsider. He tries to minimize
his own involvement, although this could never be complete, if he was to survive. Second, in seeking
general claims across the three cities about gang organization, business activities, patterns of violence, as well as
relations to community, to the criminal justice system, local politicians, and to the media, he has to standardize
his evidence and his categories, leading to thin rather than thick description, correlations rather than processes.
Third, in making the cases comparable he brackets the geographical and historical context—both the importance
of the specific urban context and changes during the ten-year period of the study. He homogenizes space and
time. Fourth, in building up his theory from the “ground” he systematically codes and classifies all the evidence.
He tends to reject (or sometimes endorse) other theories but rarely enters into sustained dialogue with them.
THE EXTENDED CASE METHOD 25
Table I. The Gap Between Principles and Practice of Science
POSITIVE SCIENCE REFLEXIVE SCIENCE
Positive
Principles Survey Research Method
Context
Effects
Reflexive
Principles Extended Case Method
Power
Effects
Reactivity Stimulus0Response Interview Intervention Extension of Observer to
Participant
Domination
Reliability Standardization Respondent Process Extension of Observations
over Time and Space
Silencing
Replicability Stabilization of Conditions Field Structuration Extension from Process to
Force
Objectification
Representativeness Sample to Population Situation Reconstruction Extension of Theory Normalization
26SOCIOLOGICALTHEORY
Just as participant observation can follow positive principles, interviews can follow the
precepts of reflexive science, in what I call the clinical method. The psychoanalytic variant
is a prototype, especially when the analyst is seen as reflexive anthropologist (Chodorow,
forthcoming). The relation between analyst and analysand is dialogic and interventionist.
Each reconstitutes the other. Second, the analyst tries to recover and work through situationally
specific experiences via dream analysis and free association. Process is the leitmotif
of psychoanalysis. The third element of structuration, that is locating psychological
processes in their wider social context, may not always be present. But here Fanon is an
exemplar. His brilliant essays on colonialism, which derive from clinical work in Algeria,
demonstrate the interdependence of psychic processes and economic, political, social, and
cultural contexts. Finally, the analyst works with a prior body of theory that is continually
evolving through attention to concrete cases. Theory is reconstructed.21
The clinical interview
not only instantiates the principles of reflexive science but thematizes its limitations:
domination of analyst over analysand, silencing of the past, objectification of personality
structures, while the theory itself is heavy on normalizing.
Extending to Historical Research
Can this binary view of science be extended to techniques other than interviewing and
participant observation? What does it mean to extend reflexive science to historical research?
This is no place to offer an extensive treatment, but let me illustrate with the comparison
of Skocpol’s and Trotsky’s approaches to the study of classical revolutions (Burawoy
1989). They are both concerned with a comparison of successful and failed revolutions.
Beyond that their approaches are diametrically opposed—the one following positive principles
and the other reflexive principles. Where Skocpol situates herself outside history to
discover the necessary conditions of revolution, Trotsky stands at the center of history to
reconstruct Marx’s theory of revolution. Where Skocpol standardizes revolutions in order
to discover the three universal factors that make for their success, Trotsky makes every
revolution distinct in revealing its defining social processes. Where Skocpol develops a
single explanation of revolution that spans three centuries as though historical time were
of no importance, Trotsky shows how the movement of world history—combined and
uneven development of capitalism on a world scale—sets off different processes for each
revolution. In the one case detachment, factor analysis, decontextualization, and induction;
in the other case intervention, process, structuration, and reconstruction. Once more
we have two models of science and two methods.
Again I have chosen Skocpol and Trotsky to highlight the contrast between positive and
reflexive methods. But one need go no further than Max Weber’s analysis of the origins of
21
Feminists have also explored this clinical or dialogical approach to interviewing. See, for example, Oakley
(1981) and De Vault (1990).
Table II. Four Methods of Social Science
Models of Science
Techniques of Research Positive Reflexive
Interview Survey Research Clinical Research
Participant Observation Grounded Theory Extended Case Method
THE EXTENDED CASE METHOD 27
capitalism for an illustration of the extended case method. In asking what it means to be a
scientist in a disenchanted, rationalized world and then asking where that world came
from, he is placing himself within history. Virtual participation gives him the psychological
processes linking Calvinist predestination to the spirit of capitalism which he then
locates within a broad array of historical forces, including the rise of a legal order, systematic
accounting, and wage labor. Throughout he was engaging with and building upon
materialist theories of the origins of capitalism. Of course, historians are usually less
self-conscious in their methodological precepts, and their work cannot be so easily divided
into one or other model of science. The purpose here, however, is to open up the imagination
to different ways of doing social science rather than abandoning science altogether
when the 4Rs seem out of reach.
Industrial and Craft Modes of Science
Having established two models of science we must now ask what are the criteria for each
model that distinguish between “good” and “bad” science—science well executed and science
badly executed. The regulatory principles of positive science—reactivity, reliability,
replicability and representativeness—define a procedural objectivity, a process of gathering
knowledge. We can call it an industrial mode in which process guarantees the product.
Conception is separated from execution and engineers define each task in the division of
labor so as to assure the quality of the final product. In the corresponding view of science,
theory is separated from research practice, which is carried out according to predefined procedures.The
prototype of the industrial mode is survey research where different tasks are parcelled
out in a detailed division of labor—the researcher, the designer, the interviewer, the
respondent—ordered by a bureaucratic structure. The interviewer and the respondent are
subordinated to the schedule, constructed by the researcher. The purpose is to obtain as
accurate a mapping of the world by delineating the procedures for gathering knowledge.
The regulatory principles of reflexive science—intervention, process, structuration, and
reconstruction—rely on an embedded objectivity, “dwelling in” theory. Here we have a
craft mode of knowledge production in which the product governs the process. The goal of
research is not directed at establishing a definitive “truth” about an external world but at
the continual improvement of existing theory. Theory and research are inextricable. The
extended case method is thus a form of craft production of knowledge wherein the conceiver
of research is simultaneously the executor. The individual participant observer carries
out all the tasks of the research process in collaboration with her subjects. The research
process is not arbitrary but it cannot be reduced to a set of uniform procedures. The weight
of evaluation lies with the product, whether reconstruction pushes theory forward or merely
makes it more complex, whether reconstruction leads to more parsimonious theories with
greater empirical content, whether reconstruction leads to the discovery of new and surprising
facts.
To put it another way, following Weber we can distinguish an objectivity based on
formal rationality—what I have called procedural objectivity—from one based on substantive
rationality—or what I have called embedded objectivity. We can even go so far as
to say that underlying our two models of science are two different theories of action—
instrumental action on the one side and communicative action on the other.
The coexistence of two models of science with their own regulative principles—their
own notions of what is a good and bad science, that is, their own notions of objectivity—
has profound consequences for evaluating any given piece of research. It means that we
should be careful not to level positive criticisms at reflexive methods nor reflexive criticisms
at positive methods. It is as inappropriate to demand the extended case method
28 SOCIOLOGICAL THEORY
follow the 4Rs as it is to impose intervention, process, structuration, and reconstruction on
survey research. One cannot dismiss the extended case method because the practitioner
alters the world she studies, because her data are idiosyncratic, because she extends out
from the local to the extralocal, or because she only has a single case! The extended case
method simply dances to another tune. Listen to the tune before evaluating the dance.
A Tale of Two Handmaidens
The coexistence of two models of science has important repercussions for the way we
think of methodology. Since, conventionally, there is only one model of science which,
moreover, usually remains invisible, method and technique are rolled into one.22
In this
monocratic scheme, methodological thinking concentrates on the relative virtues of techniques.
Some are ecumenical and argue that one chooses the technique or combination of
techniques appropriate for the problem being investigated (Sieber 1973). Others claim that
some techniques are superior to others. Thus, in the heyday of the Chicago School, participant
observation of the detached, male, professional sociologist reigned over social
surveys, sullied by their association with muckraking women reformers (Bulmer 1984;
Fitzpatrick 1990; Deegan 1988; Gordon 1992). Only later as quantitative sociology asserted
itself did survey research come to be regarded as more objective and scientific than methods
based on participant observation. In the struggle for disciplinary hegemony, each
technique tried to demonstrate its own superiority by calling attention to the biases of the
other. The elaboration of a binary view of science, however, turns the debate away from
techniques, toward the explication of methods, tied to alternative models of science.
With only one model of science, techniques may vie for a place in the sun. With two
models of science, any given method may deploy a contrapuntal method as its subordinate
complement. Survey research suffers from context effects that can best be studied and
minimized with reflexive methods. To minimize interview, respondent, field, and situation
effects, survey researchers use clinical or extended case methods. Reflexive methods become
the handmaiden of positive methods.23
Can positive methods also be the handmaidens of
reflexive science? Here, too, the answer would seem to be affirmative. The extended case
method embeds social processes in the wider array of social forces. The latter are constituted
as external to the observer and therefore can be studied with positive methods. Max
Weber, after all, depended on the empirical generalizations he developed in Economy and
Society in order to undertake the extended case analysis of the rise of capitalism in The
Protestant Ethic and the Spirit of Capitalism. In extending out from the processes of
Zambianization I made use of surveys that portrayed miners as a social force, bent on
protecting their privileged status as an aristocracy of labor. Just as reflexive methods can
serve survey research, so positive methods can serve the extended case method.
Impediments to Science: From Context to Power
It might be argued that the choice between positive and reflexive methods turns on the
problem being studied—positive methods are more appropriate to the study of enduring
22
Even the very best of methodological texts compound these different levels. Charles Ragin’s (1987) comparison
of “variable” analysis and “case study,” while overlapping with some of the distinctions between survey
research and the extended case method, assumes there to be a single model of science, one that we all share and
therefore not requiring explication.
23
Burgess (1927:114) writes: “The case-study method was first introduced into social science as a handmaiden
to statistics.” He was referring to such early sociologists as LePlay who used monographic studies to prepare the
basis for greater statistical studies. But, Burgess continues, there is nothing inherently unscientific about the case
study “provided that it involves classification, perception of relationships, and description of sequences” (117).
He, of course, sees these as two techniques for getting at the truth and not two methods, corresponding to two
visions of social science.
THE EXTENDED CASE METHOD 29
systemic properties, while reflexive methods are better attuned to studying everyday social
interaction; positive methods for the objective and reflexive for the subjective. Such an
instrumental view of method misses deep differences between the two conceptions of
science that orient us to the world we study—to stand aside or to intervene, to seek detachment
or to enter into dialogue. Usually, it is not the problem that determines the method but
the method that shapes the problem. Our commitment to one or the other model of science,
it turns out, endures across the problems we choose to investigate.
We should ask then whether there are broad factors predisposing one to adopt one or the
other model of science. Can we turn the extended case method on itself and locate each
model historically? As we have seen, the challenge for positive methods is to minimize or
control for context. Survey research becomes the less problematic the more interviews are
indeed stimuli unaffected by the character of the interviewer, the more respondents interpret
questions in identical ways, the more external conditions remain fixed, and the more
situations do not produce different knowledges. Survey research most closely approximates
positive goals the more the specifics of situations and localities are destroyed. It
works best in a reified world that homogenizes all experience, when—to use Habermas’s
(1984, 1987) vocabulary—the system colonizes the lifeworld. Positive science realizes
itself when we are powerless to resist wider systems of economy and polity. Some analyses
of the information society, postmodernity, and space-time distanciation do indeed suggest
that we are moving toward a contextless world, made for the social survey.
Reflexive science, on the other hand, takes context and situation as its points of departure.
It thrives on context and seeks to reduce the effects of power—domination, silencing,
objectification, and normalization. Reflexive science realizes itself with the elimination of
power effects, with the emancipation of the lifeworld. Even as that utopian point may be
receding, the extended case method measures the distance to be travelled. In highlighting
the ethnographic worlds of the local, it challenges the postulated omnipotence of the global,
whether it be international capital, neoliberal politics, space of flows, or mass culture.
Reflexive science valorizes context, challenges reification, and thereby establishes the
limits of positive methods.
REFERENCES
Abbot, Andrew. 1992a. “What Do Cases Do?” Pp. 53–82 in What is a Case?, edited by Charles Ragin and
Howard Becker. Cambridge: Cambridge University Press.
———. 1992b. “Of Time and Space: The Contemporary Relevance of the Chicago School.” Sorokin Lecture,
delivered at the Southern Sociological Society, New Orleans, Louisiana.
Asad, Talal (ed.). 1973. Anthropology and the Colonial Encounter. London: Ithaca Press.
Bates, Robert. 1971. Unions, Parties, and Political Development. New Haven: Yale University Press.
Bauman, Zygmunt. 1987. Legislators and Interpreters. Cambridge: Polity Press.
Becker, Howard. 1958. “Problems of Inference and Proof in Participant Observation.” American Sociological
Review 23:652–60.
Becker, Howard, Blanche Greer, Everett Hughes, Anselm Strauss. 1961. Boys in White. Chicago: University of
Chicago Press.
Bhaskar, Roy. 1979. The Possibility of Naturalism. Atlantic Highlands, New Jersey: Humanities Press.
Bischoping, Katherine and Howard Schuman. 1992. “Pens and Polls in Nicaragua: An Analysis of the 1990
Preelection Surveys.” American Journal of Political Science 36(2):331–50.
Blum, Linda. 1991. Between Feminism and Labor: The Significance of the Comparable Worth Movement. Berkeley:
University of California Press.
Bourdieu, Pierre. 1977. Outline of a Theory of Practice. Cambridge: Cambridge University Press.
———. 1990. The Logic of Practice. Stanford: Stanford University Press.
Bourdieu, Pierre and Loïc Wacquant. 1992. An Invitation to Reflexive Sociology. Chicago: University of Chicago
Press.
30 SOCIOLOGICAL THEORY
Bourgois, Philippe. 1995. In Search of Respect. New York: Cambridge University Press.
Bulmer, Martin. 1984. The Chicago School of Sociology. Chicago: University of Chicago Press.
Burawoy, Michael. 1972a. The Colour of Class: From African Advancement to Zambianization. Manchester:
Manchester University Press for Institute for Social Research, University of Zambia.
———. 1972b. “Another Look at the Mineworker.” African Social Research 14:239–87.
———. 1974. Constraint and Manipulation in Industrial Conflict—A comparison of strikes among Zambian
workers in a clothing factory and the mining industry. Institute for African Studies, Zambia.
———. 1979. Manufacturing Consent. Chicago: University of Chicago Press.
———. 1982. “The Written and the Repressed in Gouldner’s Industrial Sociology.” Theory and Society 11:
831–51.
———. 1989. “Two Methods in Search of Science: Skocpol vs. Trotsky.” Theory and Society 18:759–805.
Burawoy, Michael et al. 1991. Ethnography Unbound. Berkeley: University of California Press.
Burgess, Ernest W. 1927. “Statistics and Case Studies.” Sociology and Social Research 12(2):103–20.
Chodorow, Nancy. Forthcoming. The Power of Feelings. New Haven: Yale University Press.
Cicourel, Aaron. 1964. Method and Measurement in Sociology. New York: Free Press.
———. 1967. “Fertility, Family Planning and the Social Organization of Family Life: Some Methodological
Issues.” Journal of Social Issues 23(4):57–81.
———. 1982. “Interviews, Surveys, and the Problem of Ecological Validity.” The American Sociologist 17(1):
11–20.
Clark, Herbert, H. and Michael F. Schober. 1992. “Asking Questions and Influencing Answers.” Pp. 15–48 in
Questions about Questions, edited by Judith Tanur. New York: Russell Sage Foundation.
Clifford, James. 1988. The Predicament of Culture. Cambridge, Mass.: Harvard University Press.
Clifford, James and George Marcus. 1986. Writing Culture. Berkeley: University of California Press.
Converse, Jean, M. and Howard Schuman. 1974. Conversations at Random. Ann Arbor, Michigan: Wiley.
Deegan, Mary Jo. 1988. Jane Adams and the Men of the Chicago School, 1892–1918. New Brunswick, New
Jersey: Transaction Books.
Denzin, Norman and Yvonna Lincoln (eds.). 1994. Handbook of Qualitative Research. Thousand Oaks, Calif.:
Sage Publications.
De Vault, Marjorie. 1990. “Talking and Listening from Women’s Standpoint: Feminist Strategies for Interviewing
and Analysis.” Social Problems 37(1):96–116.
———. 1991. Feeding the Family: The Social Organization of Caring as Gendered Work. Chicago: University
of Chicago Press.
Emigh, Rebecca. 1997. “The Power of Negative Thinking: The Use of Negative Case Methodology in the
Development of Sociological Theory.” Theory and Society 26:649–684.
Epstein, A.L. 1958. Politics in an Urban African Community. Manchester: Manchester University Press.
Escobar, Arturo. 1995. Encountering Development. Princeton: Princeton University Press.
Fanon, Frantz. [1952] 1968a. Black Skin, White Masks. London: MacGibbon and Kee.
———. [1961] 1968b. The Wretched of the Earth. New York: Grove Weidenfeld.
Fantasia, Rick. 1988. Cultures of Solidarity. Berkeley: University of California Press.
Fitzpatrick, Ellen. 1990. Endless Crusade: Women Social Scientists and Progressive Reform. New York: Oxford.
Forsyth, Barbara H. and Judith T. Lessler, “Cognitive Laboratory Methods: A Taxonomy.” Pp. 393–418 in Measurement
Errors in Surveys, edited by Paul B. Biemer et al., New York: John Wiley, 1991.
Fraser, Nancy. 1989. Unruly Practices. Minneapolis: University of Minnesota Press.
Gans, Herbert. 1968. “The Participant Observer as Human Being.” Pp.300–17 in Institutions and the Person:
Papers Presented to Everett C. Hughes, edited by Howard Becker, Blanche Greer, David Reisman and Robert
Weiss. Chicago: Aldine.
Garbett, Kingsley. 1970. “The Analysis of Social Situations.” Man 5(2):214–27.
Garfinkel, Harold. 1967. Studies in Ethnomethodology. New York: Prentice Hall.
Geertz, Clifford. 1973. The Interpretation of Cultures. New York: Basic Books.
———. 1983. Local Knowledge. New York: Basic Books.
Giddens, Anthony. 1984. The Constitution of Society. Berkeley: University of California Press.
———. 1992. The Consequences of Modernity. Stanford: Stanford University Press.
Glaser, Barney and Anselm Strauss. 1967. The Discovery of Grounded Theory. Chicago: Aldine.
Gluckman, Max. 1958. Analysis of a Social Situation in Modern Zululand. Manchester: Manchester University
Press for The Rhodes-Livingstone Institute.
———. 1961a. “Ethnographic Data in British Social Anthropology.” The Sociological Review 9:5–17.
———. 1961b. “Anthropological Problems Arising from the African Industrial Revolution.” Pp. 67–82 in Social
Change in Modern Africa, edited by Aidan Southall. Oxford: Oxford University Press for International African
Institute.
THE EXTENDED CASE METHOD 31
———. (ed.) 1964. Closed Systems and Open Minds: The Limits of Naivety in Social Anthropology. Chicago:
Aldine.
Gordon, Linda. 1992. “Social Insurance and Public Assistance: The Influence of Gender in Welfare Thought in
the United States, 1890–1935.” The American Historical Review 97(1):19–54.
Gouldner, Alvin. 1954. Patterns of Industrial Bureaucracy. New York: Free Press.
———. 1973. “Sociologist as Partisan.” Pp. 27–68 in For Sociology: Renewal and Critique in Sociology Today.
London: Allen Lane.
———. 1970. The Coming Crisis of Sociology. New York: Basic Books.
Habermas, Jürgen. 1984. The Theory of Communicative Action, Volume One: Reason and the Rationalization of
Society. Boston: Beacon Press.
———. 1987. The Theory of Communicative Action, Volume Two: Lifeworld and System: A Critique of Functionalist
Reason. Boston: Beacon Press.
Haney, Lynne. 1996. “Homeboys, Babies, Men in Suits: The State and the Reproduction of Male Domination.”
American Sociological Review 61(5):759–78.
Haraway, Donna. 1991. Simians, Cyborgs, and Women. New York: Routledge.
Harding, Sandra. 1986. “The Instability of Analytical Categories of Feminist Theory.” Signs 11(4):645–64.
———. 1990. “Feminism, Science, and the Anti-Enlightenment Critiques.” Pp. 83–106 in Feminism0
Postmodernism, edited by Linda Nicholson. New York and London: Routledge.
Hondagneu-Sotelo, Pierrette. 1994. Gendered Transitions: Mexican Experiences of Immigration. Berkeley: University
of California Press.
Hyman, Herbert, et al. 1954. Interviewing in Social Research. Chicago: University of Chicago Press.
Jankowski, Mártin Sánchez. 1991. Islands in the Street. Berkeley: University of California Press.
Johnston, Paul. 1994. Success While Others Fail: Social Movement Unionism in the Public Workplace. Ithaca,
New York: ILR Press.
Kanter, Rosabeth. 1977. Men and Women of the Corporation. New York: Basic Books.
Katz, Jack. 1983. “A Theory of Qualitative Methodology: The Social System of Analytical Fieldwork.”
Pp. 127–48 in Contemporary Field Research, edited by Robert Emerson. Prospect Heights, Illinois: Waveland
Press.
Keller, Evelyn Fox. 1983. A Feeling for the Organism. New York: W.H. Freeman.
———. 1985. Reflections on Gender and Science. New Haven: Yale University Press.
Knorr-Cetina, Karin D. 1981. “Introduction: The micro-sociological challenge of macro-sociology; towards a
reconstruction of social theory and methodology.” Pp. 1–48 in Karin Knorr-Cetina andAaron Cicourel, Advances
in Social Theory and Methodology. Boston and London: Routledge and Kegan Paul.
Kuhn, Thomas. 1962. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
Lakatos, Imre. 1978. The Methodology of Scientific Research Programs. Cambridge: Cambridge University
Press.
Lamphere, Louise, et al. 1993. Sunbelt Working Mothers: Reconciling Family and Factory. Ithaca, New York:
Cornell University Press.
Lazarus, Neil. 1993. “Disavowing Decolonization: Fanon, Nationalism, and the Problematic of Representation in
Current Theories of Colonial Discourse.” Research in African Literatures 24(4):69–98.
Liebow, Elliot. 1967. Tally’s Corner: A Study of Negro Streetcorner Men. Boston: Little Brown.
MacLeod, Jay. 1987. Ain’t No Making It: Leveled Aspirations in a Low-Income Neighborhood. Boulder, Colorado:
Westview Press.
Magubane, Bernard. 1974. Review of The Colour of Class on the Copper Mines. American Journal of Sociology
80(2):596–98.
Milkman, Ruth. 1987. Gender at Work. Urbana and Chicago: University of Illinois Press.
Mishler, Elliot, G. 1986. Research Interviewing: Context and Narrative. Cambridge, Mass.: Harvard University
Press.
Mitchell, Clyde. 1956. The Kalela Dance. Manchester: Manchester University Press for Rhodes-Livingstone
Institute.
———. 1983. “Case and Situation Analysis.” Sociological Review 31:187–211.
Mitchell, Tim. 1988. Colonizing Egypt. Cambridge: Cambridge University Press.
Oakley, Ann. 1981. “Interviewing Women: A Contradiction in Terms.” Pp. 30–61 in Doing Feminist Research,
edited by Helen Roberts. London: Routledge and Kegan Paul.
Polanyi, Michael. 1958. Personal Knowledge: Towards a Post-Critical Philosophy. Chicago: University of Chicago
Press.
Popper, Karl. 1963. Conjectures and Refutations. London: Routledge and Kegan Paul.
Powers, Brian. Forthcoming. Making Marginality: How High Schools Reproduce Inequality in the Inner City.
New Haven: Yale University Press.
32 SOCIOLOGICAL THEORY
Ragin, Charles. 1987. The Comparative Method. Berkeley: University of California Press.
Ray, Raka. 1998. Fields of Protest: A Comparison of Women’s Movements in Two Indian Cities. Minneapolis:
University of Minnesota Press.
Rorty, Richard. 1979. Philosophy and the Mirror of Nature. Princeton, New Jersey: Princeton University Press.
Sanders, Lynn. 1995. “What is Whiteness?” American Politics Workshop, University of Chicago.
Schaeffer, Nora Cate. 1991. “Conversation with a Purpose—Or Conversation? Interaction in the Standardized
Interview.” Pp. 367–91 in Measurement Errors in Surveys, edited by Paul P. Biemer, et al. New York: John
Wiley.
Schuman, Howard, and Stanley Presser. 1981. Questions and Answers in Attitude Surveys. New York: Academic
Press.
Sewell, William. 1992. “A Theory of Structure: Duality, Agency and Transformation.” American Journal of
Sociology 98(1):1–29.
Sieber, Sam. 1973. “The Integration of Fieldwork and Survey Methods.” American Journal of Sociology 78(6):
1335–59.
Smith, Dorothy. 1987. The Everyday World As Problematic: A Feminist Sociology. Boston: Northeastern University
Press.
Smith, Vicki. 1990. Managing in the Corporate Interest. Berkeley: University of California Press.
Sniderman, Paul and Thomas Piazza. 1993. The Scar of Race. Cambridge: Harvard University Press.
Somers, Margaret and Gloria Gibson. 1994. “Reclaiming the Epistemological ‘Other’: Narrative and the Social
Construction of Identity.” Pp. 37–99 in Social Theory and the Politics of Identity, edited by Craig Calhoun.
Oxford: Blackwell.
Stacey, Judith. 1990. Brave New Families. New York: Basic Books.
Stinchcombe, Arthur. 1980. “Erving Goffman as a Scientist.” Unpublished Lecture, Northwestern University.
Stoler, Ann. 1995. Race and the Education of Desire. Durham: Duke University Press.
Strauss, Anselm. 1987. Qualitative Analysis for Social Scientists. New York: Cambridge University Press.
Suchman, Lucy and Brigitte Jordan. 1990. “Interactional Troubles in Face-to-Face Survey Interviews.” Journal
of the American Statistical Association 85(409):232–41.
Susser, Ida. 1982. Norman Street: Poverty and Politics in an Urban Neighborhood. New York: Oxford.
Tanur, Judith (ed.). Questions about Questions. New York: Russell Sage Foundation.
Thomas, Robert. 1985. Citizenship, Gender and Work. Berkeley: University of California Press.
Touraine, Alain. 1983. Solidarity: Poland 1980–81. New York: Cambridge University Press.
———. 1988. The Return of the Actor. Minneapolis: University of Minnesota Press.
Van Velsen, Jaap. 1960. “Labour Migration as a Positive Factor in the Continuity of Tonga Tribal Society.”
Economic Development and Cultural Change 8:265–78.
———. 1964. The Politics of Kinship. Manchester: Manchester University Press for the Rhodes Livingstone
Institute.
———. 1967. “The Extended Case Method and Situational Analysis.” Pp. 29–53 in The Craft of Urban Anthropology,
edited by A.I. Epstein. London: Tavistock.
Van Maanen, John. 1988. Tales of the Field: On Writing Ethnography. Chicago: University of Chicago Press.
Whyte, William Foot. 1943. Street Corner Society. Chicago: University of Chicago Press.
Willis, Paul. 1977. Learning to Labour. Westmead, England: Saxon House.
Winch, Peter. 1958. The Idea of Social Science and Its Relation to Philosophy. London: Routledge and Kegan
Paul.
THE EXTENDED CASE METHOD 33