This project develops a theory of social definition of gender , the meaning society ascribes to being a man or a woman, and its influence on partisan attachments of men and women and the resulting partisan gender gap. The gender gap in partisanship is a dynamic process. I address the dynamic nature of the gender gap in partisanship between 1953 and 2003 by tracing its evolution as a function of the social and political context captured by the social definition of gender. I argue that how society conceptualizes the meaning of gender continually reshapes not only the social, but also the political fabric of the United States giving rise to shifts in the social definition of gender. These shifts produce new policies for women and men, resulting in the reshuffling of partisan attachments. Additionally, this project integrates the theoretical work on the social construction of gender into empirical research by developing empirical measures of the social definition of gender.

Many theories exist to explain change in aggregate party balance: partisan realignment and generational replacement (Campbell et al. 1960; Nie, et al. 1979; Miller and Shanks 1996); retrospective evaluations of candidates (Downs 1957; Fiorina 1978); political and economic events (Erikson et al. 2002). The empirical work on the gender gap allows changing trends to have differential impacts on men's and women's partisanships, but treats sex as a constant. Treating sex as a constant limits our understanding of the dynamics underlying the gender gap in partisanship. The goal of the dissertation is to integrate the theoretical work on gender into our understanding of partisan change.

Two questions serve as the organization of the dissertation: First, what has caused changes in the social definition of gender? Second, how has the social definition of gender affected the gender gap in partisanship? I test the effect of the social definition of gender and competing theories on the formation of the gender gap.

The concept of the social definition of gender is operationalized using public opinion. I use James Stimson's dynamic algorithm, a form of principle component factor analysis, to combine multiple question series into a single time series of the progressiveness of American's gender attitudes. Survey questions pertaining to the roles and expectations of men and women are collected from iPoll and the National Election Studies. The partisanship data has been gathered from Gallup surveys.

Changes in the social definition of gender continually remold the social and political landscape of the United States. However, the connection between changes in social identities and the behavior resulting from those changes is often overlooked. This project makes that connection through understanding the impact the dynamics of social definition of gender has on men's and women's political alignments. Additionally, this project enhances our understanding of what causes changes in macropartisanship and provides an explanation for the dynamics of the partisan gender gap.