Having argued against the compassion and self-interest explanations for the gender gap, we now wish to move toward an alternative explanation. Previous research has shown that men are on average more willing than women are to endorse what Jim Sidanius and colleagues call group-based inequality or hierarchy. In contrast to our earlier findings on compassion differences, GSS survey evidence confirms that this difference between men and women exists and is related to the partisan gender gap. We use several measures of preferences regarding income equality from the GSS. The most direct measure available, from the 1984 GSS, asks respondents to rate (on a five-point scale) the acceptability of income inequality across social groups. We also use related questions from additional survey years which focus more on the particular situation in the United States.

What we found were clear and consistent differences between white men and white women across all measures of attitudes toward income equality. On the generic measure of acceptability of income differences, a gender gap existed among men and women of all races. The gap was consistently in the direction expected and present at all income levels, although it was statistically significant only among white respondents in the highest third of family income and among black respondents in the lowest third of income.

On measures that were not generic but, rather, were linked to American political issues, we see the racialized gender gap emerge again. More white women than white men supported a larger role for the government in reducing income inequality in America, and the gap remained significant regardless of individual income. Fewer white men than women expressed concern about the size of the income gap in America, and the gap was consistent with respect to the year of the survey and the income level of the respondent. Only the question asking about the income tax on the rich showed any significant variation in the gap over time or with respect to income level. In addition, also as expected, black males and females shared similar attitudes toward the contemporary income gap in America and the role of government in reducing income inequality. But the crucial point here is that, when asked about income inequality in a general, generic way rather than in the American context, men were less egalitarian than women, and this gender difference held across racial lines.

Extending our analysis, we examined the association between these attitudes toward income inequality and Democratic partisanship and vote choice, controlling statistically for compassionate attitudes (which fortunately were measured directly in the 2002 and 2004 GSS) as well as demographic variables. We find that more egalitarian attitudes predict Democratic partisanship and vote choice in both elections. As would be expected from our critique of the compassion theory, empathy has no discernible impact on either vote choice or partisanship. Moreover, the inclusion of attitudes toward income inequality in effect eliminates the gender gap; once we account for the impact of attitudes toward income inequality, gender no longer has a statistically significant association with partisanship or vote choice. We replicated the analysis with 1984 ANES data, and find that these results also hold for Democratic partisanship in 1984 and vote choice in 1980.  

A static gender difference in predispositions toward inequality cannot, however, explain the contingent, changing gender gap. For this, we turn to discuss political and social identity. We suggest that gender gaps emerge when men identify politically with groups that benefit from status quo inequality or group-based hierarchy. Among groups that are disadvantaged by existing inequalities, men’s identification with their group will dampen their typically higher levels of support for inequality, and thus prevent the emergence of gender gaps.

What does this mean for contemporary American gender gaps? First, we suggest that predispositions toward social and economic equality (unlike compassion) will be associated with Democratic partisanship and issue attitudes for those Americans, and that this association will be higher among those who identify with group(s) that are the “winners.” Second, we suggest that political identities will have complex but predictable effects on gender gaps—particularly on men’s responses to hierarchy and inequality. In addition, if we are correct about the role of identity, then we should see gender gaps emerge among African Americans who do not strongly identify with their racial group—those who lack a strong sense of what Michael Dawson in 1994 called “linked fate.”