Edlund, L. and Pande, R. (2002). "Why Have Women Become Left-Wing? The Political Gender Gap and the Decline of Marriage", Quarterly Journal of Economics 117(3): 917-962.
This article demonstrates a particular possible basis for the gender gap in American voting: the decline of marriage. The authors find a relationship between marriage and the Democratic vote for both genders: being married makes women less Democratic and men more so, while being unmarried has the opposite relationship. They claim that this relationship remains robust even after controlling for labor force participation, "women's issues" (abortion in this study), and social and religious values. The study is predicated on some pretty rough stereotypes: that women are more discriminating in partner selection and that women have default property rights in children, which leads men to contract for parental rights and sex in the form of marriage. They use an economic model to explain the income dynamics in the choice to marry or not marry, demonstrating that the existence of the gender gap is determined by the number of men for whom nonmarriage increases their income sufficiently to make them oppose redistribution versus the number of women for whom nonmarriage decreases their income sufficiently to make them support redistribution. They argue that the sequence of allegiance shifts shown by Reagan Democrats and Clinton Moms demonstrates their model's empirical significance. They also test the robustness of their results by checking the correlation of Democratic affiliation with preferring redistributive policies. Finally, they check for race as a confounder and for various statistical artifacts. They conclude by cross-checking using a different set of survey data.
My take: Let's start with the notion that children are property, and that the only thing women receive in marriage is resources, and the only thing men get is access to their kids and sex. This is a remarkably gender-essentialist approach to marriage, which has been steadily rejected by society. Tellingly, the authors immediately back off this conception, noting that "it is ill-suited to explain out-of-wedlock fertility, a family form that involves children, possibly cohabitation, but not marriage; or polygamy, a family form that involves marriage and children, but not necessarily cohabitation." I'm also concerned about how they operationalize their income control variable, as it presumes an income division that, while probably empirically correct, isn't explained in their literature review, nor argued for by their theory. I'm also concerned about their statement that "relative to the base year, there is no gender gap except in 1972 until 1980." This could equally as easily be described as "we found an effect that began in 1972, but noise drowned it out in 1976." But that isn't science, so I can't complaint too loudly.