What About Mozart? What About Murder?

Reasoning From Cases

Howard S. Becker

• Publisher: University of Chicago Press
• Year: 2014
• Print ISBN: 9780226166490, 9780226166353
• eBook ISBN: 9780226166520
• Topic: Social Sciences
• Subtopic: Anthropology
• Index: Social Sciences

In 1963, Howard S. Becker gave a lecture about deviance, challenging the then-conventional definition that deviance was inherently criminal and abnormal and arguing that instead, deviance was better understood as a function of labeling.  At the end of his lecture, a distinguished colleague standing at the back of the room, puffing a cigar, looked at Becker quizzically and asked, "What about murder? Isn't that really deviant?" It sounded like Becker had been backed into a corner. Becker, however, wasn't defeated! Reasonable people, he countered, differ over whether certain killings are murder or justified homicide, and these differences vary depending on what kinds of people did the killing. In What About Mozart? What About Murder?, Becker uses this example, along with many others, to demonstrate the different ways to study society, one that uses carefully investigated, specific cases and another that relies on speculation and on what he calls "killer questions," aimed at taking down an opponent by citing invented cases.

Becker draws on a lifetime of sociological research and wisdom to show, in helpful detail, how to use a variety of kinds of cases to build sociological knowledge. With his trademark conversational flair and informal, personal perspective Becker provides a guide that researchers can use to produce general sociological knowledge through case studies. He champions research that has enough data to go beyond guesswork and urges researchers to avoid what he calls "skeleton cases," which use fictional stories that pose as scientific evidence. Using his long career as a backdrop, Becker delivers a winning book that will surely change the way scholars in many fields approach their research.

1 First Look

After the Civil War in the United States, after Abraham Lincoln proclaimed the emancipation of black slaves and the Congress and the several states passed the Fourteenth Amendment to the Constitution, guaranteeing full civil rights to all Americans, regardless of race—after those things happened, as soon as African Americans were able to vote, they voted overwhelmingly for candidates of the Republican Party and continued to vote that way for many years. Everyone knew the reason: Abraham Lincoln had been a Republican, and the Democrats had opposed him and his forward-looking policies with respect to race, so no African American person in his (women didn’t have the vote for a lot of this time) right mind would vote otherwise. That relationship between race and voting for Republicans persisted for a long time. Until it didn’t.

Until Franklin D. Roosevelt brought the Democratic party to power in 1932 and kept it there long enough to pass major legislation that changed the social and economic position of poorer people, a group that included most black citizens. The resulting relationship between race and voting for Democrats has persisted for a long time and looks as if it will be as permanent as the relationship between race and voting Republican once looked. Or will be until it isn’t.

Similarly, after the post–World War II boom subsided, the United States changed in many ways. Factory and other blue-collar workers, who Roosevelt had also converted into persistent Democratic voters, stopped voting for the party in such large numbers, and the relation between class and party, which had seemed so permanent, stopped being so, and in a few years blue-collar voters started voting Republican in large numbers and the Reagan years began.

These correlations exhibited all the strength anyone could require to use them as building blocks of sociological thinking. But they disappeared and their opposites replaced them in a relatively short time. Were the research methods and theoretical strategies that produced those so-quickly falsified causal connections wrong? Were political scientists using bad data or faulty analytic techniques? More likely, were these supposedly invincible conclusions about race and class and voting so tied to historical circumstances that you couldn’t be sure their validity would last until the next election? Was there something wrong with the way of thinking that supposed that specific, for the most part isolated, facts about people could predict with such certainty what they would do in a specific situation like an election?

Yes. Something was wrong. I think about these things because, as a working social scientist, doing research on specific questions that interest me and I hope will interest others, they bring up practical problems I have to solve. (Discussion of related, more abstract questions relevant to the philosophy of science, epistemology, or more abstract versions of sociological theory can easily be found elsewhere, for instance Hedström and Swedberg 1998; Hedström and Ylikoski 2010; Hedström and Bearman 2009; and Ragin 1987). Many things that I study change over time—people’s experiences with drugs, for instance, which I take up in chapter 3, or how ordinary musicians, the kind who play in bars and for parties, can play together competently without any written music before them or any prior rehearsal (Faulkner and Becker 2009)—and I study them in an inclusive way, trying to learn as much as I can about what affects what I’m interested in, seeking the detailed understanding of social phenomena that results from studying them close up, finding out as much as I can about them. Close observation invariably shows that, even in the most ordinary situations, more than a few easily measured variables are at work and that everything in the situation has some effect on what happens next. If any one of those things isn’t there or, better put, is there in a different degree or in a different form, the result (the next events that happen) will differ. As a corollary, everything left out of the analysis or datagathering, perhaps because you aren’t aware it’s present, perhaps because it’s too hard to find out about, let alone measure, is still there, at work, having its effects. I want to avoid the fate of researchers who relied too heavily on a relatively few easily observed facts to do their explanatory work, so I have to not only learn about all these other elements (or variables; the name isn’t important) but incorporate them, systematically, into my explanations of what I’ve studied.

That insistence doesn’t fit well with much contemporary thinking about how social facts or events occur and develop, which instead works by measuring the connections between measured things rather than explaining how those connections produce the results we want to understand. So I rely on what have often been called case studies, in-depth studies of particular situations, organizations, or kinds of events. (The essays Charles Ragin and I collected in What Is A Case? [1992] contain important discussions of these matters; I won’t summarize them here.) Many experts have explained the logic of reasoning from collections of correlations between variables (see, especially, the illuminating essay by Passeron and Revel [2005]). This book offers explanations, arising from my own research experiences and those of others, of the logic of reasoning from cases. How do you get from the detailed knowledge of one case to more general ideas about how society, or some part of it, works? To explain that, I have to introduce a few more, not very complicated, ideas.

First, a simple observation. Everything present in or connected to a situation I want to understand should be taken account of and made use of. If it’s there, it’s doing something, however unimportant that thing seems, no matter how unobtrusive it is. Focusing on a sharply defined and narrowly delimited research question leads us to ignore everything else, or write it off as random error, or in some other way stop paying attention to it. I think that’s a mistake and instead look for a way to build what might otherwise be left out into my thinking about what I’m studying.

Another simple observation. The things I study don’t happen all at once, so I build the idea of change or process into my thinking about them. When some hitherto stable relationship turns up missing in what I’m studying, I don’t treat that as the unfortunate failure of a theory I’m testing, but rather as an opportunity to learn about some parts of the process I hadn’t seen until then. Not, as computer people say, a bug, but a feature.

I also know that what I’m studying at the moment connects to other things outside the framework I’ve built for my work and that, seen from another vantage point, the things I’ve left out could well be the center of my analysis. I try not to mistake my deliberate choices of focus for ineluctable aspects of reality.

As a result, my work doesn’t produce timeless generalizations about relations between variables. It results instead in the identification of new elements of a situation, new things that can vary in ways that will affect the outcome I’m interested in, or new steps in a process I thought I’d understood until a result different from what I expected occurred. I can use these new elements of organization and process to direct my next inquiry. For me, that’s the way social science works. I use the in-depth study of specific cases to produce new questions whose answers, in particular cases, can help me and others understand what’s going on in the social world. (For a somewhat different, but related, view of how this works, see Vaughan’s essays [2004, 2006, 2009] on analogical reasoning).

Many people think the object of sociological research and theorizing is to simplify our understanding of social life by finding the underlying laws that govern its operation. I think, contrariwise, that the object is to find out the nature of, and make a place in our thinking for, everything that observably contributes to producing the results I’m interested in. I want my analysis, my theory, to contain everything I need to describe and account for what my case study has forced me to see.

Many social scientists take nuclear physics as the model of the kind of theory they want their own work to resemble. I find a more realistic and useful model in some of the life sciences. In physiology, for instance, the reality we have to explain contains innumerable cases of the things that interest us (e.g., human bodies and their components), but, unlike the things physicists or chemists study, none of them are just alike, the way samples of copper or oxygen are or can be made to be. We have to explain how an underlying mechanism, like a circulatory system (whose fundamental design doesn’t vary much among individual specimens), produces widely differing results (blood pressure, for instance) depending on the activity of all the other systems feeding into it (that’s the substance of the input-output machines, or black boxes, I discuss in chapter 3 and again in chapter 6).

Like physiology, sociology explains how an underlying mechanism produces a great variety of experiences, depending on all the other processes whose results feed into the process producing those results (the way, for instance, drug users’ ideas about what will happen affect what does happen when they take a drug).

If you think sociology should produce a simple model that explains everything, you won’t find this way of working attractive. If you think a functioning scientific community thrives less on piling up conclusions than on creating a continuous flow of new problems to solve (which I take to be one of the messages of Thomas Kuhn’s [1970] description of scientific activity), this approach will keep us busy long into the future. It’s not just the complexity of social life that guarantees that, but also the fact of historical change, which keeps producing new forms of collective activity that provoke new ideas, new research problems, and new categories of elements whose variation will be at work in these new forms.

The chapters that follow take up a variety of questions that arise when you work this way, always looking for new elements to add to the explanatory scheme and finding them in the careful inspection of the details of specific cases, reasoning from the details of a case to a more general idea. Each chapter uses specific cases, mostly work I’ve done and reported on in the past, which exemplify one or another way of doing that, and explain how I did it. The specific cases have an interest of their own, but the emphasis is on what’s to be learned from them about this way of working, and how to do it fruitfully.

2 What’s Happening Elsewhere: Reasoning from a Case to the World

Empirical cases, studied in depth, lead us (if we pay attention to their details), to important social processes and the details of social organization that produce them. A few illustrative cases from my own experience introduce a detailed analysis of Everett Hughes’s classic article (1949), which linked race, ethnicity and the processes involved in industrialization in a general account of social change in the modern world, based on his own intensive study (1943) of a town in Quebec. International comparisons play an important role in theories of industrialization and elsewhere, and his 1949 article embodies a useful analytic strategy.

When sociologists look at other countries, they hope to see something different from what they see at home. But they also want to use what they see elsewhere to enlarge their understanding of events and organizations at home. Sometimes, more ambitiously, they hope to learn something about all countries, about countries in general, so they compile data on all the countries there are, relying for the most part on statistical data gathered by international organizations and polls. They compare countries to one other and to international averages and ranges, seeing which ones score high or low on such variables as health, wealth, political freedom, and other topics of major theoretical and political concern. Other researchers hope to learn about the generic character of certain forms of life through intensive studies of several relevant cases.

Comparing countries has a long history in sociology and related social sciences. Historically oriented social scientists have traditionally used what’s conventionally called the comparative (or comparative-historical) method to understand societies and social change at the “macro [i.e., macroscopic or large-scale] level.” They have compared, as in Edwards’s pioneering The Natural History of Revolution (1927), societies that experienced a violent revolution, to see what is common to that kind of event. More recently, Skocpol’s States and Social Revolutions (1979) compared revolutionary events and outcomes in Russia, China, and France and became a model for a succeeding generation of such studies, using archival and secondary materials to produce historical interpretations oriented toward distinctively sociological comparisons.

After World War II, the United Nations and its ancillary organizations (UNESCO, WHO, FAO, and many others) made possible and gave impetus to a new kind of cross-national research when they collected, archived, summarized, and analyzed information and then distributed the data and results widely. Sociologists, economists, political scientists, and others suddenly had massive amounts of quantitative data, useful not only for the administrative purposes for which they were gathered but also for research focused on topics of interest in their disciplines. This produced the field of “development,” the study of how countries that had not yet industrialized and modernized along Western European and North American lines fared as that process moved forward. United Nations statistics made possible, for perhaps the first time, large-scale research on a variety of topics in that general area.

So comparative sociology often takes the form of cross-national comparisons, comparing the kinds of things that happen to, and in, whole societies. Researchers who envision society as operating according to laws that specify how things like revolutions are caused by antecedent conditions, deduce possible solutions to these problems from theories, themselves deduced from more general principles or by induction from a mass of already studied cases. They try to establish, with modern statistical techniques, relations between variables describing whole societies, mostly numerically—demographic data on years of school completed, as a proxy for education; percentages of the population belonging to various religious communities; age and income distributions; political party affiliations; data on aspects of governmental forms; share of the vote political parties of differing orientations got in the last election; the incidence of various medical conditions—which might account for variation in the variables they’ve used to measure the development and modernization they want to explain.

Some anthropologists hoped that similarly testing hypotheses on larger samples would help them escape the problem of the inevitable specificity of their findings and the consequent lack of general laws that had always plagued their field. Studies of individual societies produced provocative and interesting findings. But did such findings occur universally? Margaret Mead’s research in Samoa had exploded the theory that the hormonal changes of adolescence necessarily produced the stormy emotional lives of Western adolescents, by showing that Samoan adolescents, who experienced the same hormonal changes, didn’t have those problems. But maybe one case didn’t count for so much. Wouldn’t it be better to test her idea on a larger number of societies? Such concerns led George Peter Murdock to create the Human Relations Area Files (HRAF), a heroic effort to summarize and catalog everything anthropologists had discovered and published about all the societies they had studied over the years, so that generalizations linking variables in a variety of topical areas could be checked against real data on a much larger number of societies than was possible with the single case studies that characterized almost all anthropological research (Lemov 2006, 147–69).

Political scientists and sociologists who worked on problems in societal development, and anthropologists seeking universal laws with the help of the HRAF, modeled their work on the quasi-experimental methods that dominated both psychology and economics, comparing the values of quantified variables in a large sample of societies. Though such procedures were a second-best substitute for the rigorous experimental controls of the sciences they wanted to emulate, no one knew any better way to do it, so that’s what researchers did. If their research produced the numbers the theory said it should, they took that to prove that the hypothesis they wanted to test was correct. In the anthropological cases, the variables might be quite simple, something like the presence or absence of a trait like cross-cousin marriage. With the more complex data available for nation-states, researchers used correlation coefficients or still more sophisticated measures computed from the primitive numbers. Researchers wanted to accept or reject ideas about the co-occurrence of variables, correlations that would provide evidence in favor of one hypothesis or another.

As the work progressed, these scholars thought, they would eventually be able to formulate laws modeled on what they took (not necessarily accurately) to be the kind of general laws developed by physical scientists, operating in the same way everywhere, not subject to local variation. In principle, they hoped to arrive at a social scientific “theory of everything,” like the one physicists seemed always on the verge of creating, which explained all the variation in the social world being studied. They knew—all scientists know this, though they don’t always say so—that this goal, in principle reachable, couldn’t ...

Table of contents

• Cover
• Copyright Page
• Title Page
• Dedication
• Contents
• Acknowledgments
• 1. First Look
• 2. What’s Happening Elsewhere: Reasoning from a Case to the World
• 3. Reasoning from Analogy
• 4. Black Boxes: Using Cases to Study Input-Output Machines
• 5. Complicating and Combining Black Boxes: Where Is the Value in Art?
• 6. Imagining Cases
• 7. Where Do You Stop?
• 8. IOU’s, Promissory Notes, and Killer Questions: What About Mozart? What About Murder?
• 9. Last Words
• References
• Index