An Investigation into the Differences in Predicted Election Performance Based on Candidate Gender

    There has been much discussion in the lead-up to the 2018 election about the influx of female Democratic candidates. Take Kentucky’s 6th District, for example. Republican Representative Andy Barr won the 2014 and 2016 elections by margins of 20% and 22.2%, respectively. While he is running again as an incumbent this year in a district that leans heavily Republican, we only predict him to win by a 1% margin this year. The ORACLE gives Barr a 56.3% chance of winning; while we still predict Barr to win, we expect his margin of victory to be much smaller than in past elections, in large part due to his opponent. Amy McGrath served two tours in the Afghanistan War as a fighter pilot in the Marine Corps. Polls have her neck-and-neck with Representative Barr in what has previously not been a contested district. (Please see https://polistat.mbhs.edu/states/Kentucky/6/ for more information.) The case of Amy McGrath may be representative of a trend toward success for women in congressional elections, but it does not address the real question about gender equality in the electoral process: Do female candidates truly see the same amount of success as male candidates? While McGrath is an inspiring war hero, would her campaign be more successful if she were a he? Our goal is to determine how well women are predicted to fare as candidates in our model compared to men.

    Our first step was to take our data for all 435 races and mark the gender of the Democratic and Republican candidates. In the 435 races, there were 252 male Democratic candidates, 179 female Democratic candidates, 346 male Republican candidates, and 51 female Republican candidates.

    Because we were interested in how candidates performed, the 43 races lacking a candidate from both parties were discounted. We calculated the average percentage of Democrat vote share for all four candidate gender combinations: Democrat male and Republican male; Democrat male and Republican female; Democrat female and Republican male; Democrat female and Republican female. We also found the proportion of districts that Democrats are predicted to win for each combination. Please note that all data come from our predictions made on October 17.

    As seen above, in both the percentage of vote share and proportion of wins, women are expected to perform worse than men in both parties. For example, Democratic males are predicted to win 52.2% and 60.2% of the vote when against Republican males and females, respectively, while Democratic women are predicted on average to get 50.2% and 57% of the vote when running against male and female Republicans. The discrepancy for Republican men versus women is even greater - our model predicts Republican men to garner on average 47.8% of the vote against Democratic men and 49.8% against Democratic women, while Republican women are predicted to receive 39.8% and 42.8% against Democratic men and women, respectively. The same trends can be seen in our predicted proportion of winners for each gender combination.

    We wondered exactly why these patterns were present. Do voters just prefer candidates who are men? That may be part of it, but we found another explanation for this trend that was based on the data.

    The Blair Partisan Index (BPI) measures the lean of a district based on past elections. A negative BPI means that a district leans Republican, while a positive BPI means that a district leans Democrat. The further from 0, the larger the lean. As seen above, female Democrats typically run in districts that lean more Republican than do male Democrats, and female Republicans typically run in districts that lean much further Democratic than do male Republicans. This explains why we are predicting women to typically perform much worse than men of the same party, but it brings up yet another question: why are women running in harder districts than men?

    While where candidates choose to run could be a factor, there are no data in our model to support this. Instead, we found that difference in average difficulty of districts between men and women is because there are so few female incumbents compared to the total number of female candidates. 97 out of the 228 male Democratic candidates that we observed are incumbents, while only 39 out of the 163 female Democrats are incumbents (42.5% compared to 23.9%). The same is true for Republicans: 30 out of 51 female Republicans are incumbents, while 225 out of 340 male Republicans are incumbents (55.8% compared to 66.2%). While incumbency is an advantage in itself, incumbents unsurprisingly run in easier districts. The BPI in districts is tied closely to the incumbency of the district. The average BPI of districts with Democratic incumbents is 16.7%, while it is -11.8% for districts with Republican incumbents.

    The fact that there are few women in Congress now coupled with the comparatively large number of female challengers this year means that a lower proportion of them are incumbents. The elections they compete in are thus typically more difficult. Because there are so many female Democratic challengers, a comparatively tiny proportion of Democratic women are incumbents this year. This largely explains why they are predicted to do worse. However, when we looked only at performances of based on incumbency, the results were quite unexpected.

    We decided to look at the races broken down by who the incumbent is -- Democrat or Republican. We found that, when comparing Democratic females to Democratic males, the females actually perform better. They have a higher predicted vote share regardless of which party has an incumbent or the gender of their opponent. Female Republicans, on the other hand, perform worse than their male counterparts; We predict that all Democrats -- males and females, incumbents and non-incumbents -- will have higher vote shares in races against Republican females than in races against Republican males.

    Currently, 19.4% of Congresspeople are female, and the percentage has been increasing since 1978. We tend to predict that females will receive lower vote shares than males mainly because there are currently so many more male incumbents running than female incumbents. “The Year of the Woman” implies that representation of men and women in congress will become dramatically more equal , which will still not be the case this year. However, as more women win races, the incumbency advantage they already possess will become more prevalent.This year, there was an increase in the success of women in primary elections. Importantly, we have found that women do not perform worse than men running in similar situations. Thus, continued success will eventually translate into a Congress more evenly divided by gender, but for now, more men than women enjoy the incumbency advantage and we have yet to see “The Year of the Woman.”