![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
American women remain vastly under-represented in positions of political power, and researchers have argued that this representational deficit is partially due to a political candidacy gap. In this article, we investigate the importance of gender identity at a formative moment in the development of political ambition. More specifically, we contacted thousands of local party leaders, donors, and activists to solicit advice for one of two student clubs: one women’s group and one gender-nonspecific group. Our experiment suggests that overt gender identity can lead to greater guidance opportunities – particularly from Republican women. Moreover, while the advice offered to the women’s club was far more likely to be gendered, the advice given to this group did not include a greater emphasis on the challenges associated with a political career. These findings suggest that organizational gender identity can meaningfully affect political mentorship experiences and, ultimately, the candidate pipeline.Footnote1
Acknowledgments
We are indebted to the following members of the Lafayette College Gov Lab: Alex Hunter, Azalea Danes, Caitlin Mullooly, Cameron Downey, Claire Brassil, Claire Fedor, Dylan Gooding, Eleanor Griffiths, Evan Corn, Evan Fell, Henry Morgan, Jillian Updegraff, Katie Butler, Kelly Mwaamba, Leah Bartlett, Maddie Holden, Makaila Wilson, Nora Sweeney, Olivia Lattanzi, Ryan Motto, Ryan Stratton, and William McAlpine.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1. This research was approved by the Institutional Review Board of Lafayette College on May 12, 2021 (Proposal Number: AY2021–52). Our pre-analysis plan (AsPredicted #75677) was registered just before our recruitment e-mail was distributed, on September 28, 2021. Replication data will be made available on the Harvard Dataverse.
2. We do not explore intersectional gender gaps in this article, but it is worth noting Shames (Citation2017) finds that nonwhite women were considerably less likely to consider a bid for public office when compared to men in their race/ethnic group. At the same time women in these categories had higher self-expressed levels of ambition.
3. As Sanbonmatsu (Citation2010) points out, we do not see observational evidence to support the claim that stronger parties will necessarily lead to greater women running for and winning elected positions of power.
4. In other words, it is unlikely that young women genuinely interested in gaining some advice from their local party organizations would lead with an offer to be entered into a lottery for a 25 USD Amazon gift card.
5. For examples of impermissible use of FEC data, please see the list of advisory opinions (AO) on their website: https://www.fec.gov/updates/sale-or-use-contributor-information/
6. CallTimeAI donors made up 31,440 contacts in our potential sample list. We were unable to locate a comparable list of Republican donors, which significantly affected the partisan balance of our sample.
7. See https://www.nfrw.org/history for more details on the NFRW.
8. See https://www.nfdw.com/history for more details on the NFDW.
9. In an ideal world, we would have also collected contact information on men engaged in similar forms of party activism. We restricted this sub-sample to women for several reasons. [1] Convenience: the structure of these organizations allowed us to have some sense of geographic variation and an understanding of potential missing organizations; [2] Practical Concern with Response Rates: we assumed that women dedicated to women’s organizations may be more likely to take our survey (which offered no material incentives); [3] Scope Conditions: this project involved very time-intensive data collection labor, and using these organizations allowed us to gain a foothold in an important population of interest without spending years on additional contact information collection. We do believe that future work should expand upon our steps taken here.
10. Their survey work was fielded several years before our own, but the book was not published until the year following our own survey work.
11. While we lack ex ante measures of gender identity to estimate differential response rate for all of our sub-samples, we can provide some descriptive numbers about responses to a demographic question at the end of our survey. Approximately 44% of our sample of Democratic donors identified as women. About 49% of our sample of Democratic party leaders identified as women. By contrast only 23% of our sample of Republican party leaders identified as women. Unsurprisingly, 100% of our sample of party activists in the two women’s organizations identified as women.
12. Democratic donors, local Democratic Party leaders, local Republican Party leaders, members of the National Federation of Democratic Women, and members of the National Federation of Republican Women.
13. The real student organizations were the Kirby Government and Law Society and Lafayette College Women in Law. Alterations to the real names were made with the (enthusiastic) consent of each student organization.
14. By recruiting participants to assist college students we may have unintentionally invited both differential response rates across parties and the nature of the responses we did receive. Recent efforts to target higher education institutions and the polarizing effects that follow merely underscore this potential problem. Future research should generalize from this starting point, and researchers may consider shifting to an even earlier point in the candidate pipeline by studying political advice offered to high school students or young professionals in careers unlikely to be stereotyped as disproportionately associated with the Democratic Party.
15. We include “might or might not” in our pre-analysis plan (and subsequent analysis) because we believed this would allow us to capture a tentative openness to support. As we show later, the disaggregated results indicate strong support is the driving force behind our findings.
16. Our pre-analysis plan specified simple linear models, which is what we report below. We prefer linear models for their ease of interpretation and familiarity to readers, but nearly every analysis in the article does not depend upon regression analysis at all – let alone the choice of linear vs non-linear alternatives – because our treatment is randomly assigned. Randomly assigned treatments help to obviate the need for matching algorithms or control variables intended to approximate an experimental design. In each of these instances, simple t-tests provide differences in proportion results that are identical to those presented in our paper. The only case in which modeling may be necessary is our interaction models testing Hypothesis 3. Rerunning that analysis using logit regression returns results consistent with the linear models presented below. We also report the results of randomization inference analyses for the primary results in this article in Online Appendix 3. The results are the same as those presented here.
17. It is worth noting that we do not evaluate the quality or utility of the advice provided by our respondents. On the whole, our impression was that the advice offered was thoughtful and genuine, but we make no claims about the exact benefit of additional advice rendered.
18. We had 14 respondents indicate that they do not identify as a man or a woman, but instead, in “some other way.” However, a closer look at these respondents suggests that the overwhelming majority of these – if not all of them – selected this option in protest of a question that allowed for more than two answer choices. For example, one respondent wrote, “seriously – some other way? Good grief. A WOMAN!” This respondent was coded as a woman, given that they self-identified as a woman in an open-ended response.
19. While it is outside of the scope of this study, we argue that a qualitative research design focused on deep interviewing and campaign strategy documents would better allow us to evaluate the possibility of a post-gender approach adopted within the Democratic Party.
20. This pattern may reflect several plausible processes. It is possible that Republican women are eager to encourage would-be candidates that can help close the gender gap in their own party, which is larger than the gap in the Democratic Party. Alternatively, Republican women are more dismissive of gendered barriers to professional success than Democrats (Winter Citation2023), which may lead Republican women to also dismiss the misperception of an average gender penalty at the ballot box. This interpretation, however, runs contrary to the primary findings on hostile sexism and candidate opposition in Winter (Citation2023).
21. This problem may actually be more severe than our results suggest, given that respondents in our study almost certainly are predisposed to offer advice than those that choose not to participate in the study.
Additional information
Notes on contributors
Julia Cassidy
Julia Cassidy is a recent Lafayette College graduate (Class of 2022), a former EXCEL scholar, and the inaugural Gov Lab Manager. Her research focuses on US elections and the gender gap in American political representation. Julia is a Legal Assistant at Giordano, Halleran & Ciesla, PC.
Andrew J. Clarke
Andrew J. Clarke is an Assistant Professor in the Department of Government & Law at Lafayette College and a Faculty Affiliate at the Center for Effective Lawmaking. He teaches and researches on topics relating to the distribution of political power within the United States, with a particular emphasis on legislative politics, intraparty factions, and American political institutions.