Female Candidates in the 2018 U.S. Midterm Election

*This BBL seminar is co-hosted by RIETI and Tohoku University.

Date January 10, 2019
Speaker Barry BURDEN (Professor of Political Science at the University of Wisconsin-Madison)
Moderator ONO Yoshikuni (Faculty Fellow, RIETI / Professor, Tohoku University)

Professor Burden will analyze the 2018 U.S. midterm elections with a focus on the role of female candidates in the #MeToo era, the importance of the Midwest "rust belt" states to swings in election results, and the influence of President Trump's foreign policies in voter decision making. Exploring further the impact of female candidates in recent elections, he will present the results of survey experiments designed to test the conditions under which American voters prefer male or female candidates for office, testing several prominent theories of gender bias in voting behavior.


Introduction and results of the 2018 election in the United States

Barry BURDEN's Photo


The voter turnout for both Republican and Democrat parties in the United States' 2018 midterm election reached an almost unprecedented 50 percent. The Democratic Party gained control of the U.S. House of Representatives as a result, for the first time in eight years. These results are typical in that the Presidential party lost seats in the midterm elections. The loss in seats can be considered the cost of being in power of the ruling party. While the Republican Party gained seats in the Senate, this does not refute the trend as a majority of typically Republican states had Senate elections. The Democratic Party also gained Governor seats as well as seats in State Legislatures.

Social groups and issues in the election

The preliminary estimated data from the 2018 midterm election voter turnout by key demographic groups defined by education, race, ethnicity, place of habitation, and age compared to the 2014 midterm election show significant differences. There was a higher proportion of college educated voters for the Democratic Party in 2018, especially women. Also, younger voters showed a large increase in voter turnout and support for the Democratic Party. Since 2006, young voters have shown strong support for the Democratic Party. This age difference between Democratic and Republican voters is historically significant, and the Republican Party stands to lose more voters as its more elderly voters age.

Voter issue priorities

Republican voters prioritized issues of the economy, illegal immigration, taxes, and gun policy. Whereas Democratic voters focused on issues of healthcare, treatment of women, income and wealth disparities, gun policy, and climate change. The Democratic candidates benefited from these issues because Democratic voters were more energetic in the election and their issues gained prominence. Voters were quizzed on President Trump's trade, immigration and foreign policies as they exited polling stations. Regarding trade policy, only a minority of voters found the policy to be helpful, of which these voters were predominantly Republican voters, while the majority, who thought the policy had no impact or hurt the economy, voted for Democratic candidates. Similarly, immigration policy and foreign policy also showed findings that suggested these policies negatively affected Republican Party performance on the midterm elections.

Polls of whether foreign countries treat the United States fairly or not in terms of trade demonstrate that public opinion has been shifting with voters increasingly viewing the United States as being fairly treated, in contrast to the Trump administration's position on the topic. This trend has increased gradually from 1994 to 2018 where the proportion reached almost 50%. The change in public opinion of foreign countries treatment is largely due to Democratic voters, whose opinions also shifted significantly since President Trump's election, and this partisan divide is a new phenomenon.

The mobilization of women in Democratic politics

2018 was dubbed "the year of the woman" due to the increased presence of power exercised throughout society. 1992 was first called the year of the woman. This is a comparison of those assessments.

The 2018 elections were different in terms of issues relating to women voters, candidates, and women in general. After the Trump inauguration, a women's march was held in Washington which galvanized political opinion, encouraged and organized candidates for office, and highlighted concerns related to the Trump administration. Additionally, the last two years has seen the rise of the #MeToo movement, where cases of sexual harassment, assault, and discrimination have been made public to combat the perpetrators, and this combined with the response to Trump's election and policies have inspired women to enter politics. The dramatic increase in the overall number and number of female Democratic candidates seeking to run against Trump in 2020 is a remarkable indication of this trend.

Before 1992, there was no difference between Democratic and Republican parties in the likelihood of women running for office. This changed significantly in 1992, especially in the Democratic Party where the number of women running for election in the Democratic Party was approximately 2:1 compared to the Republican Party. This is partly due to a controversial U.S. Supreme Court nomination which involved allegations of sexual harassment. Another factor was that healthcare was a key issue for women in that election. In addition, there was a change in the zeitgeist of U.S. politics where a large number of incumbents left office or were defeated, leaving room for women to enter the arena. As a result, the Democratic Party became identified with women and in 2018 gained a ratio of more than 3:1 in female candidates compared to the Republican Party, where the number of women has not changed considerably. Furthermore, women running for the Democratic Party are more likely to win than Republican women, and have thus Democratic women have gained more seats. Of the new Republican seats gained in the 2018 midterm election, only two are women, and the racial demographic profile of the party is almost exclusively white. However, of the new Democratic seats, the demographic profile has almost as many women as men, is younger, and more diverse in terms of race and ethnicity. This demographic difference in the parties has only emerged in the last 20 years.

Experimental Analysis

Women are increasing in numbers of those running for and gaining office, but the numbers of women represented in office is still not on par with men. We do not know whether voters respond differently to male and female candidates. We conducted experiments inside public opinion surveys to try to understand how voters respond to female candidates in a general sense. In the major public office seats, women represent a ratio of 1:4 or 1:5 compared to men, and the office of the President has never been held by a woman.

The reasons are unclear. Voters may have a bias against women in office. Survey research shows that there is increasing support for well-qualified female candidates for President. When asked whether they would vote for a qualified female for President, the figures have risen from less than half of the population in the 1950s to over 90 percent who currently replied in the affirmative. However, there is a disparity between these survey responses and a similarly worded survey question. When respondents were asked whether the United States is ready for a woman president, only 60 to 80 percent of respondents replied in the affirmative., indicating that the responses to the more direct question may have been somewhat dishonest. Survey respondents may feel pressure to give socially-appropriate answers, not to be seen as discriminating against a female candidate. These inconsistencies in survey results led to the use of experiments.

Candidate characteristics in the experiment

Voters may have a predisposed baseline gender bias for having either a woman or man in public office. In addition, voters might be biased against male or female candidates because of stereotypes they hold. It is difficult to distinguish between a baseline bias and the influence of a stereotype by simply looking at real world data. Herein lies the benefit of experiments which can disentangle these factors. Conjoint experiments were conducted which randomize candidate characteristics of hypothetical candidates to test the independent effects of individual characteristics such as male, female, experienced, inexperienced, etc. The experiment was conducted on the general public so it generalizes well to the U.S. population. It analyzed responses regarding hypothetical candidates running for the office of the President as well as hypothetical candidates for Congress, for both primary and general elections. General elections are dominated by partisanship which guides voter's choices. However in a primary election, because of the fact that the candidates are from the same political party, the voters would rely on other factors to shape their decisions, possibly considering gender, race, or age. The participants in the experiment are told the candidate's sex, age, race, ethnicity, family background, popularity, personal traits, political party, policy stance and experience in office. These are characteristics that may lead to a voter stereotype, however in the experiment the participants are directly informed of the nature of these characteristics for each candidate. As a result, characteristics such as leadership or experience, which may have been attributed to a particular gender stereotype, are challenged through randomly-chosen, contrasting examples. The participants see a table of two candidates and a list of characteristics. The characteristics are chosen at random by the computer program. Subsequently, the participants are asked who they would vote for.

Statistical analysis

The characteristics are plotted against their statistical effect on the likelihood that a participant would vote for a hypothetical candidate, showing either a positive or negative effect of each characteristic for the hypothetical candidate. The effect of being a female candidate compared to being a male candidate controlling for all other characteristics shows a statistically significant negative result, a 2 percentage point penalty. This result appears to be a pure gender bias, as stereotypes have been accounted for in this experiment.

During the experiment, half of the participants were told that they were examining candidates for presidential elections while the other half were told they were candidates for congressional elections. The analysis shows that there is no statistically significant benefit or detriment to being a female congressional candidate. However, female presidential candidates suffer a statistically significant, two and a half percentage point loss. This is interesting when considering that Hilary Clinton lost the general election by less than a percentage point in the Electoral College vote in three states.

The research on the reasons for the differences seen in the response to female presidential candidates and female congressional candidates is ongoing, but there is speculation that the fact that voters have prior experience of female members of Congress, as it is comprised of approximately 25 percent female members, provides a reference for the voters and thus the choice is a comfortable one. However, voters choosing a female presidential candidate are faced with a somewhat unprecedented situation, and the uncertainty in relevant details may cause issues for 2.5% of voters.

Barriers for female Presidential candidates

The experiment is also useful for examining the effect in primary elections where all candidates have the same party affiliation and partisanship is not a discriminating factor, versus general elections where there may be differences between voters with different party affiliations, and also observed the effect for candidates of different party affiliations and voters who have identified themselves as members of one of the two major parties or as independents. In the data, only Republican voters discriminate against female candidates running in primary elections. The data is statistically significant by two and a half percentage points. On the other hand, Democrats seem to have a positive bias in favor of female candidates in primary elections, however this effect is not statistically significant. In general elections, only independent voters discriminate against female candidates, by three and a half percentage points, which is statistically significant. This could be due to independents having no political party as a guide to direct their voting. Also in general elections, for Republican or Democratic registered voters, the decision is easier assuming they vote with their party affiliation as the candidate's party is printed on the ballot and becomes the sole determining factor. As a result the data for the Republican and Democratic female candidates in general elections is not statistically significant. Thus, there are two hurdles for female candidates. The first is female candidates running in Republican Party primary elections, and the second is independently aligned voters who have no guide from a political party to rely on, who discriminate against female Presidential candidates.


Q: Regarding the effects of candidate gender compared by office, is there any difference among the Congress, Senate and Governors?

Although this experiment only analyzed two offices, presidential and congressional, there are good reasons to expand and investigate other offices because the results show no conclusive reason why female Presidential candidates are penalized. In addition to the theory of female presidential candidates being unprecedented in the United States, the office of the President may also be unique in that it requires executive decision-making, which is a stereotype of male candidates, whereas legislative offices, such as Congress, do not. The office of the governor is a possible future subject for study which may shed more light on this idea due to the executive nature of the position. However, in fact, preliminary experiments into the gender differences in gubernatorial elections seem to suggest a slight positive bias for female candidates.

Q: In reference to various popular theories of the female support for Hilary Clinton, do female voters and male voters have different bias towards the gender of a candidate?

The theory of gender affinity suggests female voters align with female candidates, notably when the information on a candidate is lacking. That being said, in our experiments there is no evidence of this. In fact, male voters show preference for male candidates; however, female voters show no preference for either gender. The overall gender bias against female candidates running for President comes only from male voters. In the 2016 election, female voters did not vote for Hilary Clinton at a higher rate than they have for previous female Democratic candidates. In fact, a majority of white women voted for Trump. Therefore Hilary's advantage among women must have come from women of other races.

Q: There may be a bias against female candidates for presidential elections because of height. In some cultures, such as Japan and the United States, strength, musculature, and height, are regarded as synonymous with ability to govern, whereas this may not be the case in countries such as the United Kingdom and Germany, which have previously elected female political leaders. What is your opinion of this theory?

There is research suggesting that taller candidates in the United States are seen more favorably, and physical strengths seem to be seen as signs of mental strengths in terms of leadership. The 2016 election was unique in that it pitted a male candidate against a female, and markedly, the male candidate, Trump, is taller than average. Some noteworthy situations that reinforced the candidate's understanding of the importance of height were the town hall-style debate where Trump intimidatingly loomed over Hilary, and Trump accusing Hilary of weakness when she fell sick during the election campaign. There is also research showing that taller voters are more conservative than shorter voters. Additionally, Hilary criticized Trump for avoiding the Vietnam War, implying that he was weak, perhaps to diffuse the height differential. I agree that height was a factor in the election. The studies on height in politics to this point have all been conducted on the basis of two male candidates. In the 2018 election there were many races of two women candidates, so it would be interesting to see research on two female candidates to ascertain if height is still a factor when both are women.

Q: Regarding the experimental procedures, does the category of independents which show bias towards female presidential election candidates contain both third party supporters and non-aligned voters?

In the United States, historically voters self-identify as Republican or Democrat rather than being official members. As such the question was worded to ascertain self-identification. In our study, the third party supporters did not represent a significant number of respondents and the results do no change whether grouped with independents or not. Additionally, the experiment did not contain hypothetical candidates of non-major party affiliation. That being said, it would be interesting for further research to analyze this if there was a larger pool of participants.

Q: Do you think economic fluctuations have an impact on female candidate bias?

President Trump is a member of the populist movements around the world that especially court economically disgruntled voters, including those that support Brexit, increasing tariffs, etc. In fact, those are quite controversial and largely unpopular ideas for the majority of voters in the U.S., although his supporters are very vocal in supporting him. The economy has been strong since he has been President, which has given him support despite his other unpopular views. However, at present, the U.S. stock exchange is in turmoil, the government has been shut down, and Trump has threatened to fire the chairman of the Federal Reserve. A continued uncertain economic situation will cast doubt on Trump's political future. If the economy continues to falter, Trump's only popular point with the average voter will be in doubt, and it will encourage Democratic candidates to run against him and other Republicans. This opens opportunities for female candidates, because the simple fact is that the large majority of the women who have entered politics in the U.S. have done it through the Democratic Party. The Republican Party will have to make significant changes if they want to increase the proportion of women voters.

Q: How many U.S. midterm election candidates emphasized the fact they are women as part of the election?

Hilary Clinton emphasized her role as a woman in the 2016 general election which is a new strategy compared with her previous campaign. The 2018 midterm elections saw many women elected, but also younger and more diverse candidates in general. The recently elected New York State congresswoman who replaced a Democratic incumbent, had more liberal political leanings. There are questions as to whether she was successful because of her liberal policies or her identity as a young, Hispanic woman helped her in the election. I think both aspects played a role. Overall, there are many factors including not only the candidates themselves but also their opponents, their particular circumstances and policies which seem to dictate whether they stress their gender or not.

*This summary was compiled by RIETI Editorial staff.