How to Elect More Women: Gender and Candidate Success in a Field Experiment with Jessica Robinson Preece

Can political parties increase female representation in politics? This week’s WAPPP seminar featured Jessica Robinson Preece, Assistant Professor of Political Science at Brigham Young University and Co-Director of the Gender and Civic Engagement Lab. Professor Preece theorizes that political parties can affect both the supply of and the demand for female candidates. Women are recruited less often and less intensely to run for political office, despite the fact that recruitment may be especially important for women. Political parties’ recruitment efforts may affect the supply of women running for office. On the demand side, some voters hold stereotypes that can harm female candidates, particularly Republican women. However, elite opinion and rhetoric can shape these stereotypes. If political party leaders set a credible norm of gender parity in the party, they may be able to change stereotypes, influence voter behavior, and increase women’s representation in political office.

To test this theory, Professor Preece conducted an experiment in partnership with the Republican Party in a Republican-dominated state with very low levels of female representation. In this setting, the precinct caucus system selects delegates to the state nominating convention. Women comprise about 50% of caucus attendees, but historically only 20-25% of elected delegates.

The intervention in this case was an official letter to precinct chairs. Participants received one of four letters: a control letter asking chairs to ensure that people feel welcome; a supply letter encouraging chairs to consider and recruit women who would make good representatives to the convention; a demand letter that asked precinct chairs to read a paragraph at precinct meetings detailing the disparity in female attendees versus female delegates, along with a statement to the effect of “women here have good insights about the issues facing our families and communities, and our party would be stronger with more representation”; and a supply and demand letter that combined the language from both letters.

Of the precincts where the chair received the control letter, 37% elected at least one woman. In the supply and demand condition, 45% elected at least one woman. Proportionally, female delegates made up one-quarter of delegates from the control precincts compared with almost one-third of delegates from supply and demand precincts.

In the supply and supply and demand conditions, there was a statistically significant increase in the number of female candidates. In the demand and supply and demand conditions, the treatment successfully promoted norms of gender equality. Precinct chairs were more likely to say that there should be more women at the convention, and female candidates were significantly more likely to give a speech. If this treatment were to be replicated state-wide, it would bring an additional 230 women to the convention, a 25% increase!

Professor Preece conducted a replication study to test the external validity of the experiment. In partnership with YouGov, a nationwide sample of verified Republican voters received letters with either a control treatment, a demand treatment asking whether voters had heard Republican officials make statements encouraging Republicans to elect more women, a supply condition with extra female candidates on the ballot, or both in the supply and demand condition. Just as in the first experiment, voters in the supply and demand condition were far more likely to vote for a woman than those in the control condition.

In conclusion, political parties can increase women’s representation through active recruitment of female candidates and by setting norms and expectations of balanced gender representation. However, the largest effect occurs when parties do both. Significantly, quotas are not the only way to increase female representation, which is particularly important for increasing women’s representation in parties ideologically opposed to quotas.

Moving forward, Professor Preece wants to explore the rich observational data that came from the caucuses and the follow-up experiment. In particular, how to men and women in these entry-level elections present themselves as candidates? Both men and women talk about issues and qualifications, but in divergent ways. Male candidates are far more likely to talk about taxes and the deficit and their experience in business or military service. By contrast, female candidates talk significantly more about education and about being a homemaker or a parent. This difference in presentation influences who gets elected. In particular, the “double bind” is relevant for female candidates; female candidates with more feminine self-presentation are found to be more likeable but less competent, and voter choice tends to follow competence rankings. The “sweet spot” for voters is the supermom, who talks about being a mother while maintaining masculine self-presentation. We look forward to hearing more from Professor Preece going forward!