8
spending. Democrats, women, and nonblack respondents are more likely to prefer female
candidates, whereas men are more likely to prefer male candidates. Respondents who do not seek a
change in “politics as usual” are also more likely to prefer male candidates. Thus, the baseline
preference is partly a consequence of voters’ stereotypes. Not only are stereotypes widely held, but
they shape voters’ predispositions towards candidates.
Vote Choice
Next I turn to the main empirical analysis. Is the baseline gender preference related to the
vote in male-female races, but not in male-male races?
9
I use logistic regression to estimate the
effect of the baseline gender preference on voting for the Republican candidate, controlling for
party identification. I also include the respondent’s vote in the gubernatorial election in order to
control for other factors leading to support for Republican candidates.
[Table 3 about here]
Recall that the baseline preference should predict the vote in the male-female race but not
the male-male races. The results generally conform to the hypothesized pattern of results. As
expected, and regardless of what control variables are included in the model, a voter’s underlying
predisposition to support female or male candidates is not systematically related to the vote in
contests featuring two male candidates (see Table 3). In the absence of a female candidate, the
baseline preference does not shape voting behavior.
9
Race politics scholars have used similar techniques to compare the determinants of the vote in biracial contests with
voting in contests featuring two white candidates (Citrin, Green, and Sears 1990).
Convention