By Christopher Achen
The following post is written by Christopher Achen, Roger Williams Straus Professor of Social Sciences and Professor of Politics at Princeton University. Achen was the first president of the Political Methodology section of the American Political Science Association.
What does diversity have to do with political methodology? Not much, it might be thought. Of course, it is nice to have a wide range of people at methods panels and at the Summer Meetings. We’re glad to encourage a diversity of entry-level people to join us. But in the end, we may say, the central aspect of our professional lives is that we do science. It is really just about the work. And the whole point of the Society for Political Methodology is that we know how to do the work and how to train new people to do it, too. That is what constitutes us as an academic enterprise.
That perspective on the business of political methodology certainly captures one aspect of who we are. But taken as a full description, it seems to me quite mistaken. It fails to see our mission whole. In the end, it is naïve, both about politics and about the science of politics. And diversity is what it does not really understand.
The lived experience of being a woman is different from that of being a man. The experience is not different in every respect, of course, but it is different. The same is true of being African-American, Asian-American, or Latino rather than being an American of European ancestry. Like gender or sexual orientation or social class, race shapes our life to an important degree, whether we acknowledge it or not. And in shaping our life course, race helps determine what we know, how we think, and what seems to us valuable and important. Our subculture’s presuppositions–whether wise or foolish, ignorant or profound–are “obvious” truths to us, and they usually go unquestioned, even by people with doctorates.
It follows that any field of study as intimately involved with human life as political science needs diversity if it is to be intellectually reputable. First, political science needs diversity in what it studies. The largest group of scholars in contemporary political science are white males, as I am. Most of us have little experience with working class life, and the great majority of us are straight. In consequence, some topics get more attention than they should, and others less. Too often, we cannot see certain topics because our eyes are blinded.
In my view, the long struggle for women’s enfranchisement has received less study by political scientists than it should, to take just one example. More generally, the study of gender, sexual orientation, social class, and race are frequently marginalized into separate courses or even separate departments, offering a convenient rationale for not doing what we ought to be doing if we were serious about politics, which is to mainstream those subjects in American politics courses. When such topics are omitted, students in our courses get a bowdlerized version of political life.[1]
Similarly, political methodologists often look past key inferential problems in the discipline because we know too little about the issues that raise them. Our students, copying what we do rather than what we say, frequently replicate our style in their applied work. By contrast, Harold Gosnell, the great pioneer of political methodology, spent the latter part of his career working with African-American political scientists on the substantive and methodological challenges of studying black politics in an era when African-Americans were poorly represented in surveys and many could not vote. The result was that he published something consequential about the topic before most other white political scientists even realized that the field of study existed.
Including the politics of neglected or marginalized groups in our coursework need not, and should not, result in ideological one-sidedness. Most of us are Democrats, and we need to guard against partisan bias in our teaching. It is not progress to replace one kind of blindness with another. But neither is it acceptable to set aside from our teaching the political lives of entire groups of people. The full range of American life needs to be taught, and to be taught with the full range of political perspectives we should bring to every topic. And these same topics need to be recognized in methods courses and connected to the statistical issues we teach for precisely the same reasons.
Second, political science needs diversity in the set of scholars studying it. In any society, not all of politics will be captured by the cultural norms and shared understandings of dominant groups. As sociologists have long understood, powerful sectors of society attempt to make their sectarian views normative, and they often succeed. This effect is no less true in academic life than in national economic and political life.
The result is that a set of highly talented but narrowly based scholars may sometimes fail even to get the facts right, as in the case of Thomas Jefferson’s relationship with his slave mistress, Sally Hemings. The overwhelming consensus of (white) Jefferson scholars was that no such relationship had existed. The testimony of those who claimed to be Jefferson’s mixed-race descendants was largely set aside or explained away—until DNA evidence showed that they were very likely correct. Nearly all the “experts” had been wrong.
The evidence for Jefferson’s paternity occasioned a great deal more surprise among whites than it did among many African-Americans. Black people came to the question with a shared memory of inter-racial relationships under slavery. Here again, ancestry matters.
“Ah,” we political methodologists will be tempted to tell ourselves, “this is other people’s problem. Humanities types—they’re different from us. By the nature of their methods, they will fall into all sorts of prejudicial errors. We, on the other hand, do science. There is a right and a wrong answer. “Eleven” has the same meaning in every culture. There is a clarity about judging good work in mathematical fields. More than most, we know how to be fair. We are trained scientists.”
Apart from the anti-humanities prejudice, much of that self-described professional identity is valuable. Scientific training does have many admirable consequences. Yet professional narrowness can also blind us to conceptual failures that are obvious to those outside our field. Consider how we treat race and ethnicity in our partisanship, turnout, and vote choice equations—either when we are doing applied work or when we are producing methodological innovations. Nearly all the time, those explanatory factors enter only as dummy variables, with no interaction terms. Since whites constitute (still) the large majority in samples of American citizens, white respondents will be the primary determinants of the other coefficients. What we are saying, then, when we enter race and ethnicity only as dummies, is that Americans of African, Latin American, or Asian descent behave just like whites in every respect but one—their intercept terms differ. White behavior is taken as fundamental, and other groups are thought to differ only in the simplest way. But that gets the substantive science wrong. Just a few minutes of obvious statistical tests suffice to demonstrate that the constant-coefficients assumption is nearly always wrong. This is a problem in purely methodological explorations, too: We often forget that it is hard to learn much about the value of a proposed new estimator when the substantive model under test is brutalizing the data.
The simple fact is that people with different histories often have different coefficients. And who first pointed that out to me? An African-American political scientist, to whom the blunder of our usual procedures was much more obvious than it was to me.
The same kinds of mistakes occur in much else we do. We are familiar with the general result that better educated Americans are less likely to be pro-life. But at one point some two decades ago, I was surprised to find in survey data that among Roman Catholics who attend services regularly, the more education they had, the more pro-life they became—the opposite of the usual effect. Thus simply putting an education variable into an equation explaining abortion attitudes makes Protestant notions normative. It ignores the different perspective of devout Catholics, and no doubt those of many other religious groups as well.
When I finally had a chance to describe my finding during a talk at Georgetown University, a Catholic institution, much of the room nodded: It was not news to them. But few of us have the kind of colleagues who could help us understand the diversity of American religious belief and experience. Most political scientists are not specialists in religion and politics. Often we have only our own experience with religion to go on, which may be thin or non-existent, and in any case is necessarily narrow. We rely on those who know more, but there are few scholars with the relevant background and expertise in most departments. Yet religion is central to much contemporary politics, and so we often write about it anyway, hoping that our lack of depth will go unnoticed. Often it does: The reviewers are not well informed either. The resulting mistakes in our professional journals are all too obvious to religion-and-politics scholars. Here, too, our lack of diversity harms the science.
Political methodologists have done very little to help us think about how to model the variegated impact of religious diversity in a country where many denominations and sects are quite small and thus scarce in our national samples. Yet the topic is crucial to political science. Here as elsewhere in political science, the important methodological advances are those that break through a bottleneck impeding the progress of applied researchers. In those circumstances, marketing methodological innovation to our substantive colleagues is easy. But to make sales, ya gotta know the territory. We often don’t. Our narrowness shows. No sale.
Political methodology and formal theory remain overwhelmingly white male enterprises. Once a field becomes monochromatic, or nearly so, then self-reinforcing mechanisms set in. The jokes, the small talk, the food preferences—all send subtle, or not so subtle, signals about who is welcome. In a subfield not famous for its practitioners’ social skills, male insecurity can lead to clumsy combative behavior that makes the atmosphere even colder. The cumulative effect can be depressingly powerful. One need not spend much time talking to women political scientists who have attended past Summer Methods meetings to hear dreadful stories of dismissive or belittling remarks, stories that are not told with nearly the same frequency about other political science conventions.
I believe that we have gotten better on all counts. Certainly the racial and gender diversity of the annual meeting is broader than it once was. But there remains much to do to make the field of political methodology genuinely welcoming to a variety of backgrounds and perspectives. Among other things, we need more tenured professors in the field who reflect the country that America has become. How might that be done?
One of the biggest obstacles, in my view, is the notion that we do not have to be intentional about diversity in hiring. This is the view that we know what constitutes good work, and we know it when we see it. I have often encountered this view among natural scientists, though not only among them. When I was an undergraduate at Berkeley, one physics professor was interviewed in the student paper on the topic of how to reach undergraduates with physics ideas. He said, “I don’t teach students. I teach physics. Some of them can get it, and some of them can’t.” That was a man ignorant about human beings.
One can understand when academics without broad graduate training across the social sciences lack sophistication about cultural hegemony: Often they have never heard the relevant ideas. And even with that training, the ideas of our own kind of people inevitably seem completely correct to ourselves. Attention to what we do not already know, or to the skills that we personally are not good at, can seem a waste of time, the kind of thing that weaker minds would be drawn to, people who cannot master what we are good at. Besides, other kinds of people are different, and sometimes one has to work harder to communicate and make friends across social boundaries. It all seems hard, somehow. Better to think that we are fine as we are, and that making a department diverse is just lowering standards. But as I have already noted, the resulting scientific blunders are very much in evidence, and very much the responsibility of that blinkered point of view.
The first step is to admit that you have a problem. And what is that problem? It is, as one sociologist puts it, that “the people you know are a really bad sample.” One high university official at another institution told me that year after year, departments had reported to him that there were no qualified minorities available to hire. What they meant was that they knew of none. Knowing that, this official eventually blocked other hiring slots until departments put in the time to call people different from themselves and ask them about talented individuals among their friends and colleagues. “You’d be amazed at how many highly qualified people they were able to find,” the official said. And departments who diversify soon see that, once hired, these same new faculty members became prominent and respected members of their departments, so that their colleagues brag about having hired them. And of course, having hired a diverse faculty, majority scholars will have colleagues who can help them avoid the bad social science endemic to narrow groups.
To get a diverse faculty, one needs a diverse graduate student pool. And all graduate students need training. That is where the Society for Political Methodology comes in. In a profession where most scholars will use quantitative tools at some point in their careers, and in which no one department can possibly offer all the courses that students might need, the summer meetings of this Society are a crucial place for students to expand their methodological horizons and acquire some inspiration and mentoring from those who have preceded them. Methodology is tough: We all have to step up to a rigorous set of standards. But the trick is to offer that opportunity in a way that is welcoming, neither off-putting nor dismissive. We don’t just teach statistics; we teach students.
Much has been done in recent years to ameliorate these subcultural deficiencies. As the meetings have grown larger, having smaller events for those not in the majority can be very helpful. But an equally important goal has to be a broad shift in how we think about ourselves. Yes, we take mathematics seriously. But no, we do not think that our math skills define who we are professionally, nor do they establish a single hierarchy that will determine all our professional decisions. After all, by the mathematical standards of theoretical statisticians or real mathematicians, everyone in political methodology and formal theory is a hopeless mediocrity. The argument for us as a field is that we have strong applied math skills, and we have serious substantive interests and real insight into politics, backed by deep political science graduate training. Our research requires both, and therefore our methodological graduate training does, too. Bright prospective graduate students will want to go where they can get an integrated understanding of both politics and methods.
In political methodology and adjacent areas of political science, all of us bring something to the table. Some of us will be better at mathematics than at depth of political understanding, and others will be the reverse, but all of us will need a good deal of both. The extremes in political methodology are likely to be weak, but the broad middle kingdom should have many mansions. That is what political methodology needs to help students build. And if that is so, then no one gender, no one ethnic or racial group, and no subcultural framework can be normative. We need to be open to everybody and genuinely welcoming to everybody—not just tolerant, but genuinely warm and collegial toward all.
That is going to require even more changes than those we have already made. But it is what a modern science of politics demands. My generation has made enough foolish mistakes already. Meeting the scientific needs of the twenty-first century demands that we broaden both our membership and our intellectual vision. For all sorts of reasons, that is the right thing to do.
[1] Needless to say, foreign scholars have a great deal to teach us, too, but the growing internationalization of political science and political methodology is too large a topic to be discussed in this essay.