Women Don’t Run? Election Aversion and Candidate Entry

Women and men are equally likely to volunteer as group representatives when chosen randomly, but women are less likely than men to run in an election.


Women are underrepresented in public office, holding under twenty percent of seats in Congress and under 15 percent in state legislatures. Closing this gender gap is essential to ensure that women’s interests are represented, increase legitimacy, and strengthen decision-making by making diverse perspectives available to legislative groups. When women do run for office, they are just as likely as men to win—but women are less likely to enter an electoral race in the first place. While this gap may have complex explanations including gender differences in ambition, past positions, and party politics, this study tests whether the election process itself dissuades women from running for office, hypothesizing that women may be more election averse than men.   The authors tested women’s and men’s willingness to volunteer when the representative was randomly selected, versus their willingness to run when the representative was chosen through an electoral campaign. By comparing willingness to volunteer against willingness to run, the study untangles the effects of ambition, confidence, and risk aversion from the effects of the electoral process.


When completing an addition task with aligned individual and group incentives, women and men were equally likely to volunteer to represent the group, but women were more averse than men to run in an election to represent the group.

  • Women and men were equally qualified to be group representative based on individual scores on the addition task. Women and men also reported similar beliefs about other group members’ abilities and willingness to volunteer or run for group representative.
  • When a group representative was randomly selected from willing group members, women and men were equally willing to volunteer: about 70% of both women and men volunteered when there was a financial cost and benefit to doing so, and over 80% when there was none.
  • When a group representative was selected through an election process, men were just as willing to run as they had been to volunteer, while women were significantly less willing to enter an electoral race—unless the race was cost-free and completely truthful.
    • When election campaigns were conducted through individually composed chat messages where candidates chose how to represent themselves, women were significantly less willing than men to run regardless of whether there was a financial cost and benefit to entry (50% vs. 72.5%) or not (60% vs. 78%).
    • When election campaigns were conducted through a truthful disclosure of each candidate’s individual score in a prior round of the task, women were somewhat but not significantly less willing than men to run when there was a financial cost and benefit to entry (55.3% vs 72.1%), but equally willing to run when there was no cost (80% and 82.5%).

In short, women are more averse than men to elections, for reasons distinct from gender differences in ambition, confidence, or risk aversion. Women may be dissuaded from candidacy by personal costs, lesser trust in the truthfulness of campaigns, and greater aversion to lying. These findings suggest that elections could attract a broader pool of qualified candidates if electoral processes were more truthful and carried less personal cost.


The University of Pittsburgh Experimental Economics Laboratory recruited 350 undergraduate students (177 women) to participate in largely gender-balanced 20-person sessions. Participants could see one another but interacted anonymously via computer. In groups of 5, they completed three rounds of an addition task, with incentives for correct answers. In Part 1, they received incentives and feedback for individual performance. In Part 2, participants first decided whether to volunteer for random selection to be the group representative. They received incentives for both individual and group-representative performance, with two variations: an individual cost for volunteering and benefit for being selected, or no individual cost or benefit. In Part 3, participants first decided whether to run for election as group representative, with two campaign variations: a brief message composed by each candidate, or disclosure of each candidate’s true Part 1 score. Participants then voted for a group representative and were compensated for both individual and group representative performance and the same cost-benefit variations as Part 2. In Part 4, participants guessed other group members’ performance rankings and decisions to volunteer or run for group representative to assess confidence levels. In Part 5, participants chose between a series of riskier and less risky paired lotteries to assess risk preferences.

Related GAP Studies