Research Papers & Projects
Who Wins Where? Gender Stereotypes of Qualifications and Offices in Entry-level Elections
Voters often rely on candidate gender to infer candidate traits, yet candidate qualifications such as career background can also trigger gender stereotypes. How might these qualifications help or hinder women candidates? I argue that the answer depends on stereotypes about specific elected offices. Rather than preferring masculinity overall, I propose that voters favor matched stereotypes, preferring feminine qualifications in feminine-coded races (e.g., school board) and masculine qualifications in masculine-coded races (e.g., mayor).
Testing this theory, I conduct a conjoint experiment varying hypothetical candidates' occupational backgrounds and family roles. I find that Democratic respondents favor women with feminine qualifications in school board races, while Republican respondents reward men with masculine qualifications in mayoral races and stay-at-home mothers in school board races. However, only the effect of masculine-coded qualifications for women candidates differs significantly between parties. These findings suggest that differing partisan preferences may create distinct electoral opportunities for men and women. Link.
Funded by the Stanford Center for American Democracy and Institute for Research in the Social Sciences and the Elections, Public Opinion, and Voting Behavior (EPOVB) Section of the American Political Science Association
Trickle Up Representation? How Gendered Backgrounds Affect Candidate Pipelines
Women’s underrepresentation is often studied in state and national elections, yet political careers begin much earlier. Because early stages of the political pipeline shape the candidate pools available at higher levels, conclusions about women’s candidacy and electoral success are vulnerable to selection effects. Using comprehensive data on contested local elections in California, I examine how gendered candidate characteristics structure candidacy and electoral success across school board, city council, and mayoral elections. I find that women and men enter local politics through different occupational pathways, with women disproportionately concentrated in education, caregiving, and other feminine-coded backgrounds. Consistent with a matched stereotypes framework, candidate backgrounds are rewarded differently across offices, with educator and caregiving signals proving most advantageous in school board contests and government experience more valuable in mayoral races.
I also examine how these local dynamics shape political advancement. Women school board winners are significantly less likely than comparable men to seek additional office and are roughly half as likely to run for state legislative office. These findings suggest that women’s underrepresentation emerges through a gendered political pipeline rather than a single electoral barrier.
The Masculine/Feminine Double Bind: Audio-Visual Gender Cues and Candidate Evaluation
While most studies find no aggregate penalty against women candidates, recent research on intra-party elections suggests women face a gendered “double bind” in self-presentation: voters prefer women who fit traditional feminine profiles yet also reward masculine-coded traits associated with political leadership. However, this research relies heavily on text-based survey experiments that lack more implicit gendered cues.
To better approximate holistic candidate evaluation, I conduct a survey experiment incorporating manipulated audio, visual, and biographical cues of masculinity and femininity. Contrary to expectations, I find relatively weak and inconsistent evidence that voters systematically reward masculine self-presentation or penalize feminine presentation. Instead, the clearest pattern is a modest yet persistent disadvantage for women candidates among Republicans; the reverse holds true for Democrats. However, male Republican respondents do appear to prefer candidates with stereotypically masculine self-presentation. These findings suggest that gendered candidate evaluations may be more conditional and context dependent than previously thought, while also underscoring the importance of carefully designing candidate vignettes and subgroup analyses. Link.
Recipient of APSA Elections, Public Opinion, and Voting Behavior Section John Sullivan Award for best paper by a graduate student at the previous APSA Annual Meeting, 2022
Funded by the Stanford Center for American Democracy and the Institute for Research in the Social Sciences
Misperceptions, Falling in Line, or Electoral Threat? Negative Attitudes Toward Mail-In Voting
With Alena Smith
In the past two elections, many conservative Americans have expressed hostility toward votes cast by mail, typically citing concerns about election fraud. Other sources of concern might stem from the threat of losing more elections to the Democrats or falling in line with elite rhetoric, especially that of Donald Trump. Using summaries of previous work in political science, news reporting, and different remarks by Donald Trump as treatment conditions, we test whether hostility toward mail-in voting is reduced by a) correcting misperceptions about the security of election outcomes of mail-in voting procedures, b) establishing key elites' acceptance of mail-in voting in certain circumstances, c) limiting the perceive electoral threat of mail-in voting, or d) some combination of the above.
We find that information treatments alone shift neither favorability toward mail-in voting nor likelihood to vote by mail in the future. However, Republican voters are responsive to information treatments when coupled with pro-mail-in voting statements; when elites are not signaling opposition to vote-by-mail, voters find the information about security and electoral threat more credible and shift their opinions accordingly. However, Republican voters are less likely to oppose restricting vote-by-mail in the electoral threat condition specifically, suggesting that fear of losing elections, not security, may primarily explain opposition to the franchise.
Funded by the Stanford Center for American Democracy, the Institute for Research in the Social Sciences, and the Stanford Center on Philanthropy and Civil Society
Welcome to the Jungle? The Effect of Voting System Reforms on Descriptive Representation
Amid rising dissatisfaction with the two major parties, several states have reformed how candidates are elected. Though often meant to curb polarization or reduce one-party dominance, these reforms may unintentionally shape who runs for and who wins elected offices. This study examines how “top-two” or “jungle” primaries affected women’s descriptive representation in the state legislatures of California, Washington, and Alaska. I assess how these changes influenced both the supply of women candidates and their electoral success, using descriptive analysis and the synthetic control method.
Findings show modest but consistent effects: under California’s jungle primary, women ran less often and won fewer seats than they would have under the prior system. These effects are most pronounced in the Assembly, a common entry point to elected office. These results caution that institutional reforms, though well-intentioned, can produce unintended consequences, shaping political opportunity structures and slowing progress on descriptive representation. Policymakers and citizens should consider these impacts when evaluating or proposing electoral system changes.
A New Evaluation of the Impact of Combining Probability and Non-probability Sample Data
With Jon Krosnick
In recent years, survey researchers have begun to explore the possibility of "sample blending", wherein a questionnaire is administered simultaneously to a probability sample selected randomly from a population and also to a non-probability sample of people who volunteer to complete questionnaires without compensation but have not been selected using any purposing method. A great deal of research shows that probability samples continue to yield highly accurate characterizations of populations, whereas non-probability samples yield notably less accurate measurements. Sample blending involves weighting a non-probability sample to match a probability sample using a handful of variables, with the intent that the weighting will eliminate the inaccuracy of the non-probability sample and yield an effectively larger sample size for much lower cost than would be incurred by collecting exclusively probability sample data.
This paper tests the effectiveness of a variety of weighting approaches applied to datasets collected from large probability and non-probability national samples who answered the same long and elaborate questionnaire, which afford opportunities for different analyses.