Stephen Colbert is one of many commentators with suggestions for fixing the GOP’s “women problem.”

One might disagree about the causes or consequences, but a lot of people seem to be certain that the Republican Party has a Women Problem. Women are more likely to be Democrats than Republicans, and they more often cast their votes for Democratic candidates. And Republican officeholders display a particular (although hardly exclusive) talent for genderrelated snafus with electoral consequences.

What is the Republican Party to do? Both practitioners and political scientists suggest one potential solution: Get more Republican women on the ballot and into elected office. On average, Republican women are perceived as more moderate than Republican men and might therefore appeal more to independent and Democratic voters in general elections. Women in office from both parties are more likely than their male counterparts to sponsor and pursue legislation with particular relevance to women, and one might hope (though some counter-evidence exists) that female candidates would less frequently commit gender-related gaffes. If the solution is this simple, why aren’t more Republican women running for and holding office? Why do women compose only 8 percent of the Republican House caucus, compared to 29 percent of the Democratic caucus (a gap that holds in other offices as well)? Our research explores one explanation for this party gap in women’s representation — by using a new method to examine the supply of female candidates for each party.

Who runs for Congress? Not your average Jo(sephina). New representatives are typically middle-aged, highly educated, strong partisans, working in high status occupations. However, determining how many men and women with these profiles are present in each party is complicated. Even large surveys like the General Social Survey contain too few individuals with these characteristics to draw conclusions about their presence in the broader population. So, we devised a somewhat unusual use of (nonparametric) regression, modeling the probability of being female as a function of the characteristics mentioned above. This enabled us to look at the proportion of women among those with most of these likely candidate characteristics and make small extrapolations to the proportion of women in the population with all of these desired characteristics. For more details on this methodology, please see our published paper in the new open-access journal Research & Politics and for even more details see the online appendix.

Our analysis demonstrated a few things about the supply of female candidates for the Democratic and Republican parties, which are highlighted in these figures. The figure on the left displays our estimates for the percent women in the pool of likely candidates — middle-aged, strong partisans, with high levels of education and occupational prestige — in each party. The figure on the right displays the actual trend in the percentage of women among newly elected U.S. House representatives over time.  There are three big points to take away from these figures.

First, the supply of women with likely candidate characteristics differs by party. While the parties had a similar and small proportion of women in their candidate pools at the beginning of the period we examine, by 2012 there were many more women in the Democrats’ pool (56 percent) than the Republicans’ pool (26 percent). Now, in terms of sheer numbers of individuals, there is ample supply of qualified women candidates in the United States to fill all of the congressional ballot spots in both parties many times over. However, the larger proportion of women in the Democratic candidate pool relative to the Republican candidate pool means that if the parties pay no special attention to candidates’ sex, the Democrats will end up selecting more women. Parties and primary electorates care about many attributes of candidates other than gender, so the percentage of women among selected candidates is likely to continue to reflect the percentage of women among Democrats and Republicans with these other attributes, resulting in significant gaps in women’s candidacies by party.

Second, this gap is unlikely to change in the near future. By looking at highly educated, professional, strong partisans in our data who are younger than typical candidates, we can project the likely pool of candidates a little bit into the future. While uncertainty is always greater when trying to predict the future, our model suggests the gap in the percentage of women in each party’s pool of candidates in upcoming elections is likely to remain at least as large as it is today.

Finally, the supply of women isn’t the only reason for women’s lower representation in elected offices. There is another big gap revealed in these graphs — between the supply of women candidates on the left and the actual percentage of women elected to office on the right. For example, although we estimate a 56 percent female Democratic pool and 26 percent female Republican pool in 2012, women composed only 31 percent of new Democratic representatives and 9 percent of new Republican representatives that year. So, it’s not just the Republicans who have a women problem — in fact, both parties are electing far fewer female  candidates than we would expect, at least given our definition of likely candidate characteristics.

One possible explanation is that we are estimating too broad a definition of the characteristics that potential candidates hold, and women are more rare in the pool of candidates defined by a more stringent set of characteristics.  Other research helps identify some other explanations for this difference, including gender gaps in recruitment, political ambition or perhaps even voter biases. But whatever the explanation, it’s clear both parties have some work to do if they want their caucuses to look more like the populations they represent.

* Melody Crowder-Meyer is an assistant professor of politics at Sewanee: The University of the South. Benjamin Lauderdale is an associate professor in the Department of Methodology at the London School of Economics and Political Science.