"An instant classic.…Becker's stories and reflections make a great book, one that will find its way into the hands of a great many social scientists, and as with everything he writes, it is lively and accessible, a joy to read."—Charles Ragin, Northwestern University
An excerpt from|
Tricks of the Trade
How to Think about Your Research While You're Doing It
Howard S. Becker
Undergraduates at the University of Chicago, when I was a student there, learned to deal with all difficult conceptual questions by saying, authoritatively, "Well, it all depends on how you define your terms." True enough, but it didn't help us much, since we didn't know anything special about how to do the defining.
I stayed at the University of Chicago for my graduate training and so met Everett C. Hughes, who became my adviser and, eventually, research partner. Hughes was a student of Robert E. Park, who could be considered the "founder" of the "Chicago School" of sociology. Hughes taught me to trace my sociological descent, through him and Park, back to Georg Simmel, the great German sociologist who had been Park's teacher. I am still proud of that lineage.
Hughes had no love for abstract Theory. A group of us students once approached him after class, nervously, to ask what he thought about "theory." He looked at us grumpily and asked, "Theory of what?" He thought that there were theories about specific things, like race and ethnicity or the organization of work, but that there wasn't any such animal as Theory in general. But he knew what to do when a class or a student got into a tangle over what we thought of as "theoretical" questions, like how to define ideas or concepts. We would wonder, for instance, how to define the concept of "ethnic group." How did we know if a group was one of those or not? Hughes had identified our chronic mistake, in an essay he wrote on ethnic relations in Canada:
Almost anyone who uses the term [ethnic group] would say that it is a group distinguishable from others by one, or some combination of the following: physical characteristics, language, religion, customs, institutions, or "cultural traits." (Hughes  1984, 153)That is, we thought you could define an "ethnic" group by the traits that differentiated it from some other, presumably "nonethnic," group; it was an ethnic group because it was different.
But, Hughes explained, we had it backwards. A simple trick could settle such a definitional conundrum: reverse the explanatory sequence and see the differences as the result of the definitions the people in a network of group relations made:
An ethnic group is not one because of the degree of measurable or observable difference from other groups; it is an ethnic group, on the contrary, because the people in and the people out of it know that it is one; because both the ins and the outs talk, feel, and act as if it were a separate group. (Hughes  1984, 153-54)
So French Canadians were not an ethnic group because they spoke French while other Canadians spoke English, or because they were usually Catholic while the English were usually Protestant. They were an ethnic group because both French and English regarded the two groups as different. The differences in language, religion, culture and the rest we thought defined ethnicity were important, but only because two groups can treat each other as different only if "there are ways of telling who belongs to the group and who does not, and if a person learns early, deeply, and usually irrevocably to what group he belongs." The heart of the trick, which can be applied to all sorts of other definitional problems (for example, the problem of deviance, to which I'll return later in the book), is recognizing that you can't study an ethnic group all by itself and must instead trace its "ethnicity" to the network of relations with other groups in which it arises. Hughes says:
It takes more than one ethnic group to make ethnic relations. The relations can no more be understood by studying one or the other of the groups than can a chemical combination by the study of one element only, or a boxing bout by the observation of only one of the fighters. (Hughes  1984, 155)
The tricks that make up the content of this book help solve problems of thinking, the kind of problems social scientists usually see as "theoretical." Defining a term by looking for how its meaning arises in a network of relations is just the kind of trick I'm talking about, but it's not the usual way of settling theoretical questions. Social scientists typically discuss "theory" in a rarefied way, as a subject in its own right, coordinate with, but not really related to, the way we do research. To be sure, Merton's two classic papers (Merton 1957, 85-117) outline the close relations he thought theory and research ought to have to one another, but students studying for examinations used those ideas more than working researchers ever did. Hughes, who oriented his own methodological work to the practical problems of finding out about the world, always threatened to write "a little theory book," containing the essence of his theoretical position and somehow different from the nuggets of sociological generalization scattered through his essays and books.
Hughes's students, me among them, all hoped he would write that theory book, because we knew, when we listened to him and read his work, that we were learning a theory, though we couldn't say what it was. (Jean-Michel Chapoulie  analyzes the basic ideas of Hughes's sociological style perceptively.) But he never wrote it. He didn't, I think, because he didn't have a systematic theory in the style of Talcott Parsons. He had, rather, a theoretically informed way of working, if that distinction conveys anything. His theory was not designed to provide all the conceptual boxes into which the world had to fit. It consisted, instead, of a collection of generalizing tricks he used to think about society, tricks that helped him interpret and make general sense of data. (The flavor is best conveyed in his essays, collected in Hughes  1984.) Because his theory consisted of such analytic tricks rather than a Theory, students learned it by hanging around him and learning to use his tricks, the way apprentices learn craft skills by watching journeymen, who already know them, use them to solve real-life problems.
To repeat and amplify, a trick is a specific operation that shows a way around some common difficulty, suggests a procedure that solves relatively easily what would otherwise seem an intractable and persistent problem. The tricks that follow deal with problems in several areas of social science work, which I've roughly divided under the headings of imagery, sampling, concepts, and logic.
My descriptions of the tricks frequently consist of extended examples that might serve as exemplars in one of the Kuhnian senses, as models you can imitate when you run into a similar problem. I've been guided in this preference for examples, as opposed to general definitions, by my experience in teaching. When I taught the sociology of art, at a time when I was writing what became the book Art Worlds (Becker 1982), I was eager to share with students my theoretical framework for understanding art as a social product. But, of course, to fill out the class hours I told a lot of stories. One of my best lectures was on the Watts Towers, the incredible construction an Italian immigrant mason made in Los Angeles in the 1930s, and then left to take care of itself. I told his story and showed slides of the work. I meant it as a limiting case of the social character of an art work. Simon Rodia, who made the Towers, really did it all himself, with no help from anyone, no reliance on art theories or ideas or art history or art supply stores or museums or galleries or any organized art anything—and I explained how the work exhibited that independence and showed how you could see the marks of most works' dependence on all that stuff in the way they were made. To me, the point was the way the marginal case explained all the other cases. It was chastening, therefore, when students later told me that the thing they really remembered from that course was the Watts Towers. Some of them, with the story in mind, remembered the point I had been at such pains to make with the Towers too, but most of them just remembered the fact of the Towers' existence, the story of this crazy guy and his crazy art work. That taught me that stories and examples are what people attend to and remember. So there are plenty of both here.
(Some readers will note that many of my examples are not exactly up-to-date, not the latest findings or ideas. I've made that choice on purpose. It surprises me how much good work of the past is forgotten, not because it isn't good, but because students have never heard about it, never had their attention drawn to it. So I have often picked my examples from work that is thirty, forty, even fifty years old, in hope of giving it a deserved new life.)
These tricks, then, are ways of thinking about what we know or want to know that help us make sense of data and formulate new questions based on what we've found. They help us get all the good we can out of our data by exposing facets of the phenomenon we're studying other than those we've already thought of.
Sociologists of science (e.g., Latour and Woolgar 1979 and Lynch 1985) have shown us how natural scientists work in ways never mentioned in their formal statements of method, hiding "shop floor practice"—what scientists really do—in the formal way they talk about what they do. Social scientists do that too, using a workaday collection of theoretical tricks when they're actually doing social science, as opposed to talking about Theory. This book deals with what are often thought of as theoretical problems by cataloguing and analyzing some tricks social scientists use, social science's shop floor practice. I'll describe some of my favorites, as well as some I learned from Hughes, noting their theoretical relevance as I proceed. I've occasionally given them names to serve as mnemonics, so you'll encounter such creatures as the Machine Trick, the Wittgenstein Trick, and many others.
Calling this book Tricks of the Trade creates some ambiguities that should be cleared up right away. The phrase has several potential meanings, most of which I don't intend. Some may hope that I'm going to pass on tricks of getting along in academia: how to get a job, how to get tenure, how to get a better job, how to get your articles published. I'm always willing to discuss such things. My unconventional academic career, in which I spent many years as what used to be called a "research bum" before finally entering academia as a full professor, might have given me some special insights that come with marginality. But times change and the economic and political situation of universities has changed sufficiently that I doubt I any longer have any inside information on those chancy processes. In any event, academia isn't the trade I have in mind. (Aaron Wildavsky  covers a lot of that ground.)
Others may think I mean technical tricks of writing or computing or "methods" or statistics (though not many expect statistical tricks from me). I've told what I know about technical writing tricks elsewhere (Becker 1986b), and probably have a similar collection of folkloric tips on other areas of social science practice to pass on. But those, while they are tricks of our social science trade, are too specific, not generalizable enough to warrant lengthy discussion. They are appropriately handed on in the oral tradition.
So I am talking about the trade of sociologist or (since so many people do work that I think of, imperialistically, as sociology even though they themselves think they are some other breed of social scientist or humanist) about the trade of studying society, under the aegis of whatever professional title suits. The tricks I have in mind are tricks that help those doing that kind of work to get on with it, whatever professional title they use. As a result, I have been somewhat carefree in using "sociology" and "social science" interchangeably, even though that occasionally creates ambiguities with respect to disciplines on the margin, like psychology.
The word "trick" usually suggests that the device or operation described will make things easier to do. In this case, that's misleading. To tell the truth, these tricks probably make things harder for the researcher, in a special sense. Instead of making it easier to get a conventional piece of work done, they suggest ways of interfering with the comfortable thought routines academic life promotes and supports by making them the "right" way to do things. This is a case where the "right" is the enemy of the good. What the tricks do is suggest ways to turn things around, to see things differently, in order to create new problems for research, new possibilities for comparing cases and inventing new categories, and the like. All that is work. It's enjoyable, but it's more work than if you did things in a routine way that didn't make you think at all.
Clifford Geertz has given a good description of the work these tricks are supposed to do:
What recommends them ["figurations" describing an ethnographic result], or disrecommends them, is the further figures that issue from them; their capacity to lead on to extended accounts which, intersecting other accounts of other matters, widen their implications and deepen their hold. We can always count on something else happening, another glancing experience, another half-witnessed event. What we can't count on is that we will have something useful to say about it when it does. We are in no danger of running out of reality; we are in constant danger of running out of signs, or at least of having the old ones die on us. The after the fact, ex post, life-trailing nature of consciousness generally—occurrence first, formulation later on—appears in anthropology as a continual effort to devise systems of discourse that can keep up, more or less, with what, perhaps, is going on. (Geertz 1995, 19)
Every section of the book thus takes up the theme of convention—social convention and scientific convention—as a major enemy of sociological thought. Every subject we study has already been studied by lots of people with lots of ideas of their own, and is further the domain of the people who actually inhabit that world, who have ideas of their own about what it's about and what the objects and events in it mean. These experts by profession or group membership usually have an uninspected and unchallenged monopoly of ideas on "their" subject. Newcomers to the study of the subject, whatever it is, can easily be seduced into adopting those conventional ideas as the uninspected premises of their research. The estimable activity of "reviewing the literature," so dear to the hearts of dissertation committees, exposes us to the danger of that seduction.
So we need ways of expanding the reach of our thinking, of seeing what else we could be thinking and asking, of increasing the ability of our ideas to deal with the diversity of what goes on in the world. Many of the tricks I describe are devoted to that enterprise.
The book's sections concern major aspects of the work of social science research. Imagery deals with how we think about what we are going to study before we actually start our research, and how our pictures of what that part of the social world is like, and what the work of the social scientist is like, get made. It discusses the various forms imagery about society takes, and suggests ways of getting control over how we see things, so that we are not simply the unknowing carriers of the conventional world's thoughts.
Sampling, the next section, recognizes that our general ideas always reflect the selection of cases from the universe of cases that might have been considered. It takes up the question of how we choose what we actually look at, the cases we will have in mind when we formulate our general ideas explicitly. It suggests the necessity of choosing cases in ways that maximize the chance of finding at least a few that will jar our ideas, make us question what we think we know.
Concepts, the third section of this book, takes up the making of our ideas. How shall we put together what we learn from our samples in the form of more general ideas? How can we use the world's diversity, which our efforts to improve our imagery and sampling have delivered to us, to create better, more useful ways to think about things?
Finally, Logic suggests ways of manipulating ideas through methods of more or less (mostly less) formal logic. This section borrows heavily from materials already constructed and diffused by others (notably Paul Lazarsfeld, Charles Ragin, and Alfred Lindesmith—an unlikely trio). A major theme here, borrowed from Ragin, is the usefulness of focusing on a diversity of cases rather than on variation in variables. (That shorthand will be explained in "Logic.") I don't apologize for my borrowings, except to say that I've taken only from the best and given credit, as best I can remember, for what I've taken.
Readers will soon discover, so I might as well confess, that there is a certain arbitrariness in where topics are discussed. Most topics could have been (and sometimes are) taken up in more than one place. The section headings are only rough guides to the section contents. The ideas are not a seamless web of logically connected propositions (don't I wish!), but they are an organic whole. That is, they all pretty much imply one another. The book is a network or web rather than a straight line.
The sections seem to have a kind of rough chronological order, too. You might think that researchers naturally begin their work by having images of various kinds about what they are going to study and then, on the basis of those images, develop ideas about what to study and how to choose cases (in other words, how to devise sampling schemes). You might think further that, having picked the cases to be studied and having studied them, researchers then develop concepts to use in their analyses, and apply logic in the application of those concepts to their cases. You might reasonably think all that because most of the books on theory building and methods of research specify such an order as the "right way." But if you did, you'd be wrong. The various operations have that kind of logical connection among themselves—imagery, in some sense, certainly underlies and seems to dictate a kind of sampling—but that doesn't mean you do them in that order, not if you want to get any serious work done.
Serious researchers repeatedly move back and forth among these four areas of thought, and each area affects the others. I may choose my sample in a way that takes into account my image of what I'm studying, but I will surely modify my image on the basis of what my sample shows me. And the logical operations I perform on the results of some part of my work will probably dictate a change in my concepts. And so on. There is no sense imagining that this will be a neat, logical, unmessy process. Geertz again:
One works ad hoc and ad interim, piecing together thousand-year histories with three-week massacres, international conflicts with municipal ecologies. The economics of rice or olives, the politics of ethnicity or religion, the workings of language or war, must, to some extent, be soldered into the final construction. So must geography, trade, art, and technology. The result, inevitably, is unsatisfactory, lumbering, shaky, and badly formed: a grand contraption. The anthropologist, or at least one who wishes to complicate his contraptions, not close them in upon themselves, is a manic tinkerer adrift with his wits. (Geertz 1995, 20)None of the tricks of thinking in this book have a "proper place" in the timetable for building such a contraption. Use them when it looks like they might move your work along—at the beginning, in the middle, or toward the end of your research.