Women comprise only 20% of Congress, and there are stark partisan divides–of 21 female senators, 16 are Democrats while only 5 are Republican, and in the House there are 62 Democratic women, but only 22 female Republican House members. One possible explanation for such under-representation is voter bias against female candidates. While voters may prefer shared-gender candidates for either instrumental or affective reasons, if there is a psychological benefit from voting along gender lines, this affective motivation could take the form of bias against female candidates that serves to perpetuate the under-representation of women in politics. I employ a self-affirmation treatment in an experimental voting vignette to isolate this psychological payoff. Conducted among male and female Republicans and Democrats, subjects are faced with a trade-off between voting either along identity or policy lines. By substituting for the boost to self-integrity that would otherwise be gained through in-group voting, the self- affirmation treatment is shown to reduce in-group voting among men but has no effect among women. More specifically, Republican men are significantly less likely to vote for the male candidate when self-affirmed. An experimental boost to self-worth (self-affirmation) reduces in-group bias among Republican men by about 20 percentage points. Reductions in the extent to which self-worth is tied to one’s group attachments drive this reduction in shared-gender voting among Republican men.