The Trouble with Categories in a Dynamic Account

Exploiting variation over time and across groups to gain inferential leverage is promising in theory, but fraught with difficulty in practice because categories are treated in static rather than dynamic terms. A theory of politicized context of gender is up against an imposing set of current analytical practices common in individual-level analyses of political behavior. In its most common form, analysis of political action in the mass public is accomplished by aggregating empirical observations taken from individuals. These data are typically quantitative responses to closed-ended questions in surveys from large-n samples. Among nationally representative data, the American National Election Studies (ANES), spanning presidential and midterm election years between 1948 and the present, are the most widely utilized data for studying voting behavior, candidate choice, and public opinion. Among the many virtues of ANES data is the ability to examine changes and continuities in behavior and attitudes over time, in large part due to the consistency of content and question wording in the survey instruments across time. Similarly, the sample sizes are large enough to analyze important groups of voters categorized by gender, region, partisanship, and race.¹

The analytical strategy of static categories in individuals as the unit of analysis not only impedes our ability to disentangle the roots of political inequality embedded in gender, it also creates other undesirable and unintended consequences. Regarding unforeseen outcomes, adherence to the perspective consistent with most individual-level behavioral accounts results in what has been described as the “puzzle of participation,” the observation of stagnant rates of political activity despite substantial increases in formal educational attainment in the mass public over time. The most important explanatory variable for political action at the individual level is some mix of the socioeconomic status (SES) duet of education and income. Anybody who analyzes survey data knows that educational attainment is the 800-pound gorilla in results of model estimation; depending on your version of the analogy, the gorilla sits on or eats up much of the explained variance. The problem with this most venerable of social scientific models is its application to explain change over time. If indicators of SES, particularly education, are critical antecedents to participation, and if formal education has risen dramatically and monotonically over time, why has political activity not increased in a commensurate rate?²

While there have been improvements in women's participation in terms of voter turnout as well as running for office and winning, most observers and analysts would agree that despite higher levels of education among women compared to men, inequality in the political sphere will likely persist for the same reasons that income inequality remains between men and women of identical educational attainment and credentials. Analyses structured around static categories and driven by aggregated individual-level data form the foundation for explanatory models, resulting in specific forms of substantive knowledge. Yet some of that knowledge is contrary to observed phenomena—no increase in political participation despite rising education levels in the mass public, and no greater political activity among women in the face of higher college attendance—creating at best a puzzle and at worst an embarrassment for scholarship in this tradition. Unable to accommodate a dynamic account of political context, the clunky square pegs of individual-level behavioral analysis cannot fit through a round hole of analysis.

Further complicating the picture of marrying individual-level analysis with a politicized context of gender is the issue of intersectionalty. When categories are constructed from static accounts—woman, African American, poor—analyses most often look for the independent effect of one category against another. Race may be a more important predictor than gender for one dependent variable, whereas class might overwhelm the effects of race for a different outcome. Results, indeed substantive knowledge structured by this analytical perspective, are meaningful when the default categories are assumed to be male, white, middle class. But why continue to assume this? What is the inferential utility of arguing, for example, that race is more visible or more important than gender? How would this generalization apply to women of color? These analytical strategies become less appealing when we take into consideration changes in the demographic composition of the U.S. population. Race in the United States is very much a moving target. Currently, one-third of the U.S. population considers itself to be something other than white, and with Latinos now the largest minority group in the country, we have traversed well beyond the black-white binary. The size of the “multiracial” population in the United States, a category of race only recently constructed by the federal government, reached 3% of the population in the 2000 census. Further, there is a significant tension between race and ethnicity, most clearly exemplified in the self-identification of the Latino and Hispanic population in the United States. These trends make manifest the notion that categories are fluid and porous, overlapping multiplicities at times and stubbornly stable in other contexts. The complexity of categories must be acknowledged and embraced and then integrated into the methodology of studying gender and political action, rather than perpetuated as mutually exclusive dummy variables.

In this regard, feminist scholarship offers an important theoretical alternative to static unidimensional categories. As Leslie McCall writes: “[F]eminists are perhaps alone in the academy in the extent to which they have embraced intersectionality—the relationships among multiple dimensions and modalities of social relations and subject formations—as itself a central category of analysis” (2005, 1771). Feminist scholars have long recognized the potentially damaging consequences of oversimplifying subjects and relationships, and Joan Scott articulated two decades ago (1986) the simple and powerful notion that relationships of power are manifested in the binary category of gender (Scott 1986). Scholarship by women of color, including the work of Gloria Anzaldua, Patricia Hill Collins, Kimberle Crenshaw, Angela Davis, Evelyn Nakano Glenn, bell hooks, Chandra Mohanty, Cherie Moraga, and Elizabeth Spelman, among others, pioneered theories of intersectionality, rejecting the separability of identity categories and analytical perspectives based in these distinctions. Recent essays in this journal (Beckwith 2005; Burns 2005; Hawkesworth 2005) review the relevant thinking on gender as a category of analysis, and provide important perspectives within the field of political science. Acknowledging a deconstructionist standpoint and recognizing the dynamism of categories are certainly important inasmuch as they caution us to refrain from reinforcing the simplification of categorical fictions. Yet it is these static and unidimensional categories that form the basis of evidence marshaled by political actors to advocate for public policies that have material influence and profound outcomes for women and men alike. While appealing on many levels theoretically, the elimination of conventional categories has important implications for the construction of substantive knowledge and for political action, potentially leaving a gap in the ability to provide an empirical or quantitative response to evidence in support of regressive policies that can further legitimize and exacerbate inequality.

Efforts continue to balance the theoretical robustness of perspectives suspicious of categories with a real-world policy imperative to enhance political equality among people classified by government into distinct categories (see, e.g., McCall 2005; Young 1994). Building on the work of scholars who advocate a relational perspective, McCall's work on inequality is exemplary as an approach to feminist intersectional analysis of intercategorial complexity that “requires that scholars provisionally adopt existing analytical categories to document relationships of inequality among social groups and changing configurations of inequality among multiple and conflicting dimensions” (2005, 1773). McCall (2001) demonstrates that particular social, economic, and political contexts reveal distinct configurations of inequality. She concludes that “having used traditional analytical categories as a starting point, classified individuals into those categories, and examined relationships of wage inequality among such groups of individuals, I arrived at the complex outcome that no single dimension of overall inequality can adequately describe the full structure of multiple, intersecting, and conflicting dimensions of inequality” (2005, 1791). Nancy Burns calls for a similar course of action: “[B]ecause gender is usually not an average experience, we are not going to be able to read the consequences of gender formation 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” (2005, 140).

This is hard work, but treating categories as static and unidimensional misrepresents circumstances of everyday life and produces substantive knowledge about political action of a particular kind. To develop and utilize a theory of a politicized context of gender, scholarship must treat categories dynamically, remaining suspicious of the homogenizing generalizations that go along with classification, while at the same time examining specific intersections of categories at particular points in time. If this were not difficult enough, a second roadblock in the form of a long-standing assumption about individual agency amid structural neutrality further slows the movement away from the mass production of square pegs.

Neutrality and Individual Agency

Beyond acknowledging the complexity of categories and fashioning analytical strategies to accommodate this shift, individual-level approaches need to reevaluate in a radical way another assumption driving the design of research questions and the interpretation of empirical results. The current state of knowledge in the participation literature based in individual-level approaches has been built substantially on an analytical triumvirate of individual-level synchronic data, the socioeconomic status model, and a mainstream definition of political participation as voluntary legal acts directed at government officials toward policy outcomes. Undergirding this perspective is the assumption that individuals have equal agency. Inferential models estimated with the data belie this bias: One more year of education or one more mobilization request will increase political participation among people regardless of the particularities of their social and political context. The starting position that makes this assumption defensible is that the system itself—whether political, social, or economic—is neutral, not favoring one or another for any particular characteristic. Put more concretely, if communication to a member of the U.S. Congress from a middle-class black man receives the same attention as one from a wealthy white contributor, then there may be good reason to proceed without examining further the assumption of equal agency. But if we are suspicious that political responsiveness varies systematically by race, class, gender, or some other category in which individuals exist, we must scrutinize the assumption and devise strategies to test the validity of the starting claim.

There is ample evidence within political science and other disciplines to document pervasive inequalities inside the political, economic, and social systems of the United States (see Frymer 1999; Marable 1983; Walton and Smith 2000). Yet there is surprisingly little sustained empirical and theoretical effort to illuminate more precisely when and how power and hierarchy structure opportunities and incentives to act in politics. There are certainly exceptions, chief among them John Gaventa's brilliant study of powerlessness and quiescence in Appalachia (1982). Similarly, important work in comparative politics, such as James Scott's Weapons of the Weak (1979), serves as a stark reminder to political scientists that one cannot assume all individuals to have equal ability and desire to influence the political system, and further, that the return on that investment in time and resources will be the same for all who take part.

Instead, agency or individual rights in the liberal democratic vernacular operate in both a social context of power relations and a structural context of democratic political institutions whereby actors deploy accumulated capital in pursuing their interests. Mary Hawkesworth makes a compelling argument identifying the roots of the “voluntarist” conception of politics within social contract theory: “Initially conceived by Hobbes, the voluntarist conception ties power to the voluntary intentions and strategies of individuals who seek to promote their interests. Within this frame, power is nothing other than ‘the present means to some future apparent good’ (Leviathan, Part I, Chapt. 10, p. 150)” (Hawkesworth 2005, 147–48). But the context of democratic politics itself is not neutral, not level, not fair. It matters who got there first and set up the rules of the game, and it matters that they were men. In this regard, a theorized political context must account for power that is manifested in the political state as institutional structure and practice. Civil society is inextricably linked to the state and political society, both as an extension of state hegemony and as an arena of potential counterhegemony. Thus, political action is a double-edged sword; it has potential for liberation and transformation, but it is also a uniquely powerful tool for the development of false consciousness. In this regard and as the literature in women and politics amply attests, more women in government does not always mean better government for women. As long as government—replete with gendered and discriminatory institutions—remains intact rather than transformed, populating it with diversity can at best alter outcomes incrementally. Is small change better than no change? Perhaps, but let us at least acknowledge it is small change.

Finally, we must also be mindful of the potential negative consequences of encouraging political action in an already existing system that structurally disadvantages groups of people. A memorable scene in the classic camp film What Ever Happened to Baby Jane? (1962) illustrates the significance of structural inequality (see Ambar 2006). In the film, two sisters, both former actors, live with each other in a crumbling Hollywood mansion. Blanche (played by Joan Crawford) was once a successful actress as an adult, but she is now in a wheelchair following a mysterious accident in which Jane was thought to have played a part. Jane (played by Bette Davis), a former child star long since gone from the spotlight, isolates and torments Blanche. In one scene Blanche is in her wheelchair in tears, telling Jane “You wouldn't be able to do these awful things to me if I weren't still in this chair,” to which Jane replies, “But you are, Blanche, you are in that chair.” Exploited, disenfranchised, discriminated populations—whole complex categories of people—are in a metaphorical wheelchair in the American political system, constrained by layer upon layer of disadvantage. Even though social scientists are fully aware of the situation, we still take the results of our SES models and implore members of disadvantaged groups to get up and walk; better yet, run! But why put the burden on Blanche to walk? Rather than focus on the individual as the thing that needs fixing—more motivation, higher resources, stronger democratic values—a higher level of scrutiny should be directed at the institutions and practice of the democratic system itself. In this regard, it is the public roots as well as the private roots of public action that need to be highlighted (see Mettler 1998, 2005). Thus, the task for scholars interested in gender and political action is to fashion innovative strategies both to account for complexities in categories populated by individuals and to discern how power and hierarchy structure the opportunities and rewards for political action among individuals.

What Has Equality Got to Do with It?

Getting otherwise square pegs through round roles has to have a substantive end, and in this case, it is the enhancement of equality. From the perspective of individual-level behavioral research, the strong normative prior assumption privileging equality of individual agency would force us to ask what we want from gendered political action. Let us take the plunge and ask ourselves the biggest counterfactual for U.S. politics: What would things be like if everyone were equal? What would gender look like? How should institutions be different? How could democratic practices and political culture work to enhance equality? In short, we need to begin by assessing the extent to which the political system as we know it is good enough to create equality across domains, categories, and time.

In the absence of this kind of a radical transformation of assumptions and research practices, individual-level research addressing political participation will continue to reproduce more of the same findings about the relevance of gender to political action. Small differences—inequalities across distinctive domains—can either be explained away, claimed as idiosyncratic, or downplayed as minor given the overall trend toward parity. A way to begin to whittle down the square edges is to engage in more “adding up” of differences by gender over time and across areas that we have until now neglected to cover. In a 2005 essay in this journal, Nancy Burns identifies the problem: “[G]ender 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” (p. 140). This is an important strategy, used with important substantive consequences by other social scientists to explain the slow progress of women in positions of power. Psychologist Virgina Valian (1998) argues that small differences in the perceptions of qualifications and the evaluations favoring men over women in the academic professions accumulate in the end to result in substantial disadvantage to women's advancement. Given smaller lab space or several thousand dollars a year less in salary, women faculty are encouraged by their department chairs not to “make a mountain out of a molehill,” but Valian shows just how those small differences aggregate to significant advantages to men in the long term: “Those small imbalances accumulate to advantage men and disadvantage women. The most important consequence of gender schemas for professional life is that men tend to be overrated and women underrated. Although most men and women in the professions sincerely hold egalitarian beliefs, those beliefs alone cannot guarantee impartial evaluation and treatment of others.”³

Valian's findings suggest that among the critical tools for improvement from theoretical models favoring static categories and assuming equal individual agency in neutral systems must be a shift in belief systems about what constitutes the right starting point for the articulation of research questions, the development of designs for data collection, and the interpretation of empirical findings. Transformation to models of political action embracing dynamic intersectional accounts must begin with greater self-recognition among scholars of the normative biases underlying their work, along with a commitment to develop creative methodologies to study the complex set of interactions of political action. Only then can the substantive knowledge generated from systematic research represent the lived experiences of women in political action.