Women and Compromise: A DW-NOMINATE Analysis of Congresswomen and Congressmen

Following the end of the government shutdown, many senators, political analysts, and news hosts stated that congresswomen were instrumental in shepherding through a resolution to the standoff. This was reported by the New York Times, the Huffington Post's Laura Bassett, and even my local Star-Telegram. While the progressive feminist in me was more than happy to accept this nascent conventional wisdom at face value, the political psychologist in me wanted to test it empirically. What follows is are my analyses and results of the differences in polarization and (when converted into absolute scores) ideological partisanship between men and women of both parties. Although the media focused on women in the Senate, I focused on the House of Representatives for my analyses as that chamber provides larger numbers of both men and women to find statistically significant effects.

(Note: Skip to the results if you're not as interested in the finer details of coding and preparing the data. For any policy wonks who are, feel free to email me for more information and I'll reply as quickly as doctoral responsibilities will allow. Lastly, I started from 1967 because it was near the end of the Realignment of the South and it would be harder to find significant results based on sex in previous Congresses. There simply weren't enough women elected yet. The same is true for my race/ethnicity and party voting unity analyses, which aren't reported in this article.)

Method. Using the DW-NOMINATE datasets available at VoteView.com and Congressional information on women who have served in the House of Representatives since 1967, I coded all the women in the Excel spreadsheet as '2' and men as '1' (parties are already coded as 100 for Dem and 200 for GOP). I removed any codes for presidents or third-party outliers (only a handful in over 10,000 rows in Excel) and brought the dataset into SPSS.

Polarization Measure. The distances between the parties on the 1st-dimension variable in the DW-NOMINATE dataset represents polarization (see Keith Poole, 2008). It ranges from -1 (Liberal) to +1 (Conservative), with higher scores reflecting party loyalty as a function of members' voting behavior (Hare and Poole, 2013).
Partisanship Measure. I computed the partisanship measure by multiplying all Democrats' polarization scores by -1. This allowed for direct comparisons between parties and sexes.

As illustrated in Figure 1 (open here), Republican women were less partisan than any other PartyxSex group from 1967-1994 (and statistically the lowest of all in 1987 and 1989). From 2003-2012, however, Republican women have been more partisan than Democrats regardless of sex.

Republican men were less partisan than Democrats, regardless of gender, from 1973-1978. From 1991-2012, however, Republican men have been more partisan than Democratic men, and they have been more partisan than Democratic women from 1997-2012.

Finally, Democratic women have been more partisan than Democratic men from 1999-2012 (but this wasn't statistically significant in 2005).

In summary, Republican men and women have been more partisan than Democratic men and women for the past decade or so, and Democratic women more than Democratic men.

So, Are Women Less Partisan? (YES)
Putting parties aside, is it the case that Congresswomen are less partisan than Congressmen? As illustrated in Figure 2 (open here), women in the House of Representatives were statistically less partisan than men in the 112th (2011-2012) Congress, F(1, 440) = 4.71, p = .031. (Note: the degrees of freedom suggest a total above 435 due to some House seats being replaced during the ongoing term, such as Suzan DelBene replacing Jay Inslee.)

Figure 3 (open here) shows an interaction across parties such that Democratic men and Republican women are less polarized than Democratic women and Republican men.

Figure 4 (open here) shows more polarization between Democratic men and women than between Republican men and women (indicating that Republicans are closer in partisanship across sex).

In short, the data supports the conventional wisdom that women are more likely to compromise than men.

Of course, there are several caveats and limitations to these results. First, I stated that the data 'supports the conventional wisdom' but doesn't confirm it. There is no statistic of how constructively politicians compromise and whether or not congresswomen have more integrative negotiations. That strong an assertion would require a more controlled study, though I doubt any politicians will sign-up for a laboratory experiment anytime soon. Sex may moderate the relationship between partisanship and compromise, or ideology may mediate the relationship between sex and compromise.

The next caveat is that both parties have become more polarized over time, so even if Congresswomen are less partisan than men, there is still more partisanship than there was for either party or sex 10-20 years ago. In addition, this data does not reflect the proportion of moderates (DW-NOMINATE scores of -.25 to .25) among men and women. While Democrats have not increased in partisanship over time to the same degree as their Republican counterparts, the proportion of moderate women may be similar across parties. Given that this categorization depends on a statistical cut-point, it can't be ascertained by simply viewing the graphs.

Lastly, differences in racial/ethnic diversity between parties must be assessed to ensure they are not contributing to these results. To understand how this would affect political negotiations, consider this example: a person comfortable with diversity could work with a racially diverse or homogeneous group equally well, whereas a person uncomfortable with diversity can only work with a homogeneous group. Given that House Democrats today are more diverse than they have ever been, and Republicans are less diverse in this 113th Congress than they have been at any time since in 1991, the inability of Republicans to understand or care for issues important to minority groups may further impede their ability to compromise with a group as increasingly diverse as the Democratic Party, regardless of sex.

This disconnection from minorities is not only true of Conservatives in the U.S., but Conservatives in the U.K. as well. Thus, having a top-two voting process in which candidates must earn votes from voters across the ideological spectrum may help reduce polarization to the extent that Republicans would have to appeal to minority voters, which is something they never have to do when addressing their own base. Of course, if voting laws prevent a portion of racial/ethnic minorities, college students, single women, and divorced (newly single) women from voting, then Republicans may still thrive off running to the wing as opposed to running to the center.

Perhaps a combination of more women in leadership AND the Top-Two Primary system is a solution for America's ongoing political depression. All the more reason to look forward to our first female President in 2016.

Acknowledgement: I would like to thank Dr. Keith Poole for assisting in the interpretation of the DW-NOMINATE data (and for making these amazing datasets publicly available).

Inquiries: I'm happy to provide Bonferroni pairwise comparison data, year-by-year, to any political analysts or professional policy wonks who are interested. Email: Jarryd.Willis@gmail.com