19
the proportion of the legislator’s total sponsorship and cosponsorship agenda that is
devoted to women’s issues and the subset of feminist and social welfare issues, I utilize
Ordinary Least Squares (OLS) regression models. To address the fact that the
sponsorship data is highly skewed toward zero since the majority of members do not
sponsor any women’s issue bills, I employ robust standard errors in the bill sponsorship
models.
3
The independent variables employed in the regression analyses in Tables 3
through 6 draw on the vast congressional research concerning the factors that motivate
legislators’ policy decisions. Since party affiliation is one of the most reliable guides to
how members of Congress approach issues (Rohde 1991; Cox and McCubbins 1993), I
created variables for Republican men and women and Democratic men and women.
Dividing men and women by party allows me to assess the possibility that differences
attributed to gender are better explained by the fact that more women in Congress are
Democrats and Democrats are viewed as more active on social welfare issues. Since I
expect that Democrats are more supportive of these proposals, the models in Tables 3
through 6 include the variables for Republican men and women and Democratic women.
Democratic men are the comparison category. Therefore, a positive and significant
coefficient for Democratic women would indicate that being a Democratic woman is an
important predictor of the agenda-setting activities of members of Congress and
Democratic women are even more likely to devote their scarce resources to the pursuit of
women’s issue initiatives than are Democratic men. Additionally, the division of gender
by party allows me to examine the differential impact that the shift in party control and
therefore majority/minority status has on men and women of the same party. To further
examine the impact of identity on legislative choices, I also include variables for African-
Convention