How do we find gender in data from individuals?

I should start by saying what I mean by gender. I take gender to be, in part, the “values, norms and demands the female human being—precisely because she is female—comes up against in her encounter with the Other” (Moi 1999, 79). And, in part, it is what women and men make of the systematic way social interactions, structures, and institutions are organized around gender.

Finding gender in data requires understanding a few important features of its social organization (Gurin 1985; Jackman 1994). These features are easiest to see in contrast with race. These features of social organization—largely enabled by an intimacy that is usually absent in modern racial formations in the United States—affect the forms the “values, norms and demands” I just mentioned take.

By contrast with race, gender is both more in the open and more invisible. It is more in the open in the sense that ordinary Americans are often more comfortable making essentialist claims about gender than they are about race (Burns and Kinder 2003). And it is more invisible in that distinctions of sex are more naturalized, less questioned, than distinctions of race. It is also more invisible because its hierarchy is made through often-subtle cumulation of often-small advantages across a host of different institutional spaces—at work, in the family, in school, in religious institutions (Burns, Schlozman, and Verba 2001). Its cumulation is subtle because of the range of seemingly disconnected institutions where it operates, because intimacy conditions the size and nature of the advantages, because intimacy often makes it hard to see the disadvantages in the first place, and because it is often easy to explain away the disadvantage by the arrangements that are taken as necessary for childbearing and childrearing.

The ways gender is more invisible than race make it harder to see gender at the individual level and especially in the cross-section than it would otherwise be, but these differences with other social formations are, to my mind, research opportunities. I will take up each of the aspects of invisibility in turn.

First, gender arrangements are often naturalized, seen as the way things must be. That makes trouble, for example, for understanding elite-mass linkages because there is often simply no worked-out language people use to talk about gender and politics. Of course, social movements have tried to make some kinds of language—especially languages of structural disadvantage—commonplace (Goffman 1977). Despite these efforts, politicians often seem unable to find words when faced with gender issues. That lack of words opens up a host of comparative research questions: What, for example, are the consequences for ordinary understandings of policy when elites are articulate about disadvantage and when they simply do not know what to say? Does it matter that—to my mind—ordinary and elite Americans are practiced with languages of race and not at all practiced with languages of gender? And, as Erving Goffman asked, can we explain “the way in which these differences were (and are) put forward as a warrant for our social arrangements, and most important of all, the way in which the institutional workings of society ensured that this accounting would seem sound” (1977, 302)?

Second, gender happens in a host of social institutions. It is made and remade across these institutions in ways that build linkages across institutions. So an advantage or disadvantage that comes from a gender formation in one space can have far-reaching consequences by shaping outcomes in other places. The cause of any gender advantage or disadvantage, then, might be proximate or it could be quite distant. Without taking simultaneous account of the host of institutions in which women and men operate, scholars are not likely to understand the causes and scope of disadvantage.

Third, intimacy makes trouble in a number of ways. Because of it, there is probably not as much systematic violence as with some other hierarchies, and so the hierarchy sometimes works more subtly. By working often through psychological intimidation, coercion, and acquiescence, gender hierarchies are recipes for the morselization of experience, for enabling people to explain any individual outcome as the product of individual and idiosyncratic circumstance and not as a consequence of large-scale structural forces like discrimination. To be visible, these cumulated wrongs must be added up—either over institutions or over time. A single snapshot can miss them unless that snapshot is viewed in the context of a structural account of disadvantage. Otherwise, disadvantage may be hard to see and easy to explain away. Without one of these two approaches—adding up or setting within a structural account—disadvantage, even disadvantage that is perpetrated with violence, can seem like a choice. (In some sense, this is the burden of Catherine MacKinnon's arguments about difference and dominance [MacKinnon 1987].)

Of course, as I mentioned earlier, gender is also more out in the open than race, and that too has consequences for research. In work that we have done, Donald Kinder and I found that when ordinary Americans build folk theories of gender—when they do find language for gender—their theories often sound old-fashioned by comparison with the theories they construct for race (Burns and Kinder 2003). The terms of the debate center on essentialism—its acceptance or its rejection. With race, by contrast, the theories are elaborate, multifaceted, and about structural or cultural difference. The ordinary people we talked to in our work were quite comfortable saying that gender differences in a range of different outcomes are part of God's plan for women and men. This sort of essentialized language is nearly invisible when these same people are talking about race.

When we investigate gender in any setting, I think we need to have these, and I am sure other, features of gender formation squarely in mind. These features should shape our research questions, our research designs, and our strategies of analysis. And the comparative leverage they give us—because they demand that we think about the specificity of gender formations and the ways the formations we care about are different from other forms of social organization—will strengthen our insights and our contributions.

I should conclude with a small set of ideas that grow out of what I have just said, a set of ideas that have affected how I think about studying gender in the individual-level data I so often put to use in my own work.

The first two points are ones I have already mentioned: first, that gender is easier to see over space and time, after the researcher does the work of adding up the many often-small wrongs through which gender inequalities are manifest; and second, that the influences of gender often come from the ways it shapes people's lives in institutions outside the ones we study, and it is the researcher's job to link these institutions.

The third point grows out of the point I started to make about the potential for specific gender formations. It is unlikely that there is only one way gender is arranged in the United States. And so it is unlikely that we'll succeed in our research if we pretend that all women or all men share a vast quantity of life experiences. In the end, there are only a few ways that scholars have succeeded when they have treated gender as an average experience (I am thinking here of MacKinnon [1987] and Jackman [1994]). There are two consequences of this third point for our analyses. First off, we have to theorize and model the ways gender works homogeneously and heterogeneously, not because heterogeneity is a goal in itself, but rather because we will get the story wrong if we focus solely on the things that all women or all men share. Instead, we might think of the things that some women or some men share at some times and places (see, for example, Young 1994). And, second, because gender is usually not an average experience, we are not going to be able to read the consequences of gender formations from a single coefficient on whether the person is a woman or a man. Instead, we will want to structure our analyses to pinpoint gender in a pattern of coefficients that represent the paths, the experiences, the mechanisms through which gender formations operate. Of course, that means we must explicitly theorize those paths, those experiences, those mechanisms in the very first place.

In the end, I am excited about the places gender scholarship is going, and I am thrilled at the ways gender scholars are working to develop the theoretical tools, the research designs, and the analytical tools that enable them to carry their rich understandings of gender to data and—because they are the data with which I mostly work—to individual-level data in particular, where I think the task is an especially challenging one (Burns 2002).