Gender Turnover and Roll Call Voting in the US Senate

Abstract

Most studies looking at the roll call voting behavior of female legislators have investigated this phenomenon at the state legislative level and for the US House of Representatives. Very little research has looked at the impact of gender on the policy records of US senators. With the number of female senators continuing to increase it is now possible to undertake such an analysis. This study examines the influence of gender in predicting the roll call voting behavior of US senators across several recent congresses. To unearth gender effects, it employs a longitudinal design based on turnover in the Senate, which holds constituency constant while allowing gender and party to vary. The results indicate that male and female senators representing the same state compile very similar voting records on the basic left/right policy dimension. However, when votes on issues of concern to women are examined, female senators tend to be more supportive than the male senators they replaced, and male senators tend to be less supportive than the female senators they replaced.

Keywords:

• women in Congress
• Senate roll call voting
• gender and representation

Acknowledgments

Support for this research was provided by the Center for Legislative Studies at Bridgewater State University. The author wishes to thank Tracy Osborn and the anonymous reviewers for their helpful comments on earlier drafts of this article.

Notes

1. This observation should not be interpreted as suggesting that previous studies of gender and legislative representation in the US House have not made a valuable contribution to understanding this topic. It is merely meant to convey that examining the descriptive representation of women in the Senate can build on what is already known based on the existing body of research.

2. The same relationship exists for the US Senators as well. Examining a t test for the difference of means on one widely used measure of state ideology (Berry et al. 1998) indicates that the states represented by women over this period were about three points more liberal than the states represented by male senators (p < .05). This measure ranges from 0 to 100 with higher values representing more liberal states.

3. Of course while the geographic boundaries of the district may remain constant there are inevitably some changes in the composition of the population from one year to the next. However, this approach most closely approximates legislators representing the same constituency since any year-to-year changes are likely to be minor.

4. One potential problem with employing the turnover model in the Senate is that, unlike members of the House, senators stand for election every six years. Some studies have indicated that senators may tend to alter their voting habits as the date of the election becomes imminent (e.g., Wright and Berkman 1986). However, as previous research has demonstrated, the ideological positioning of most members of Congress tends to be very stable over time (Poole 2007; Poole and Rosenthal 2007).

5. The list of senators serving in each state over this period was collected from http://www.senate.gov/pagelayout/senators/f_two_sections_with_teasers/states.htm.

6. Another possibility is to construct a composite index of votes on women's issues for each Congress over this period. However, establishing an unbiased set of criteria for what votes should be included in such an index is problematic. Relying on the scorecards of other women's group is constrained by data availability and the fact that most groups keep scorecards on a session-by-session basis rather than for the entire Congress. Some of the women's groups relied upon in previous efforts to create an index of women's issues have not published scorecards for all of the years covered in the present study (see Norton 1999).

7. However, Swers (1998; 2002) also supplements the votes on the AAUW scorecard by including votes on issues of concern to the Congressional Caucus on Women's Issues in her analyses. There was no such caucus in the Senate over this period.

8. The percentage of support for the AAUW position excludes those votes for which members were not present.

9. For additional documentation of the votes included by the AAUW in its scorecard, refer to http://www.aauwaction.org/VoterEd/cvr.htm

10. The average number of votes included in the AAUW scorecard was 7.6 per Congress over the time period covered in this study.

11. These issues are likely to be problematic for any other measure that estimates support for women's interests as well.

12. In all of these cases, a male senator replaced a male senator.

13. In one case, an Independent replaced an Independent: In the 110th Congress, Senator Bernard Sanders replaced Senator James Jeffords of Vermont. This case of turnover was coded as no partisan change.

14. Another possibility in attempting to control for the effects of partisanship is to simply examine cases of gender turnover within each party. However, this approach is problematic on both practical as well as theoretical grounds. Eliminating cases of partisan turnover further reduces an already limited sample size. In addition, the number of cases of gender turnover within each party is quite small, especially for Republicans. There are only six cases of a female Republican senator replacing a male Republican senator. These numbers make it very difficult to estimate models that are statistically reliable. Another reason it is advisable to include partisan turnover in the model is that it allows for a comparison of the relative impact of each variable. More specifically, it provides an estimate of how much influence gender has on roll call voting behavior stacked up against the most powerful predictor of roll call behavior: a senator's party affiliation.

15. In total, 12 of the 104 cases of turnover involved an instance where a new senator was appointed as a replacement.

16. Furthermore, the independent variable controlling for margin of victory was not statistically significant in either model specification. This result is consistent with the findings of Schwindt-Bayer and Corbetta's (2004) model of House turnover.

17. However, Frederick's (2010) analysis of roll call voting in the Senate employing a pooled regression model with a much larger number of cases found that female Republican senators may be significantly more liberal on the basic-left continuum and on a metric of support for women's issues.