Skip to main content

Is Running Enough? Reconsidering the Conventional Wisdom about Women Candidates


The conventional wisdom in the literature on women candidates holds that “when women run, they win as often as men.” This has led to a strong focus in the literature on the barriers to entry for women candidates and significant evidence that these barriers hinder representation. Yet, a growing body of research suggests that some disadvantages persist for Republican women even after they choose to run for office. In this paper, I investigate the aggregate consequences of these disadvantages for general election outcomes. Using a regression discontinuity design, I show that Republican women who win close House primaries lose at higher rates in the general election than Republican men. This nomination effect holds throughout the 1990s despite a surge in Republican voting starting in 1994. I find no such effect for Democratic women and provide evidence that a gap in elite support explains part of the cross-party difference.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8


  1. As a result, I do not discuss the extensive literature on barriers to entry or theorize specifically about how party might interact with gender at the candidate entry stage of the electoral process. These questions, while important, are independent of my research design, which focuses exclusively on the women who have chosen to become candidates for office (and thus surpassed the barriers to entry).

  2. Anastasopoulos (2016) also uses an RDD on congressional primary elections with male and female candidates, but finds no evidence of a ‘gender penalty.’ Why do our results differ? First, I analyze outcomes separately for Republican and Democratic women, motivated by recent work suggesting the effects of gender may vary by party (Sanbonmatsu 2002b; Palmer and Simon 2008; Elder 2008, 2012; Pearson and McGhee 2013; Swers and Thomsen 2014; Thomsen 2015). Second, our results differ in the specific type of effect being estimated (Cattaneo et al. 2016). This difference arises from a choice about which races to include and is discussed in further detail in the methods section of this paper.

  3. The increase in the number of female candidates during this cycle is largely attributed to the highly salient allegations of sexual harassment made by Anita Hill during Clarence Thomas’s confirmation hearings, along with an unusually high number of open seats (Palmer and Simon 2008, p. 37).

  4. As Koch (2000) notes, such a heuristic could lead to a greater perceived ideological gap between female candidates in the Democratic Party and general election voters than it would between female candidates in the Republican party and general election voters, perhaps suggesting there might be a ‘gender bonus’—operating through ideology—for female Republican candidates who reach the general election. However, such a heuristic-driven gap might also explain why women are more likely to be nominated as Democrats in the first place.

  5. My paper builds on Pearson and McGhee (2013) in that the estimates from the RDD are based on women and men who ran in districts that are comparable, on average. As a result, the selection of women into districts should not bias my estimates.

  6. Throughout this study, I have limited the data exclusively to primary elections where a man and a woman were the top two finishers. This is necessary to ensure that each observation has a proper counterfactual. If primaries with only women as the top two finishers were included, for example, the estimates could not be interpreted as the effect of nominating a female candidate as opposed to nominating a male candidate as this would not always be a plausible alternative outcome. This issue cannot be avoided by comparing each female candidate to the male candidate with the highest vote share. For intuition, consider a three candidate primary with two women who placed first and third. If the third-place female candidate is included in the analysis, that observation would be coded as the female candidate losing the primary, yet the general election outcomes used as the dependent variable would be from a case where a female candidate was the nominee. To account for this, I subset to races where a man and a woman finished in the top two. However, an alternative way to ensure a proper counterfactual would be to subset to primaries in which only one woman entered the race. In that case, each observation would be a female candidate and all comparisons would be between her and the male candidate with the highest vote share. In practice, since the number of primaries with two female candidates is relatively low, the samples that result from using these two methods are quite similar (particularly around the primary threshold) and so the results are identical regardless of which is chosen.

  7. Specifically, Lawless and Pearson (2008), citing Gaddie and Bullock (2002), note that “up until the 1970s, nearly half of all congresswomen were elected following the deaths of their husbands” (p. 67). In turn, starting the analysis in 1972 decreases (although certainly does not eliminate) the probability that a woman is elected in this scenario. This is important because these particular situations could confound estimates of a female candidate’s nomination with an observationally equivalent ‘coattail’ effect.

  8. For example: if the female candidate wins the primary, the dependent variable would be the share of the vote that the female candidate received in the general election. If the female candidate loses the primary, the variable is coded as the share of the vote that the male candidate receives in her stead. Thus, this is fundamentally a comparison between female candidates who narrowly win or lose, and the outcomes that occur as a result.

  9. These covariates, which are all common predictors of candidate and/or party success, include Presidential Vote Share at T-1 and indicator variables for Incumbent, Presidential Election Year, and Decade (by redistricting cycle).

  10. Although somewhat common in the literature, I do not present estimates using third or fourth-order polynomials as RDD’s using high-order polynomials can be highly misleading and are more prone to type 1 errors (Gelman and Imbens 2014).

  11. This imbalance is not a sorting violation as the fundamental comparison for the design is between the women who win—and get nominated—and the women who lose—and see a man nominated instead. Moreover, even if this particular imbalance is present in the data, it would only make the finding that Republican women lose at higher rates more unlikely.

  12. The plots are zoomed in to provide a better view of the discontinuity. However, doing so means that a small number observations at the extremes are not depicted.

  13. For example, the first vertical line represents a bandwidth of 5%. For this bandwidth, I run a regression of the form in Eq. 1 using only the observations for which the primary margin is between −2.5 and 2.5%. I then construct confidence intervals using the maximum of either the conventional or robust standard errors (Angrist and Pischke 2009, p. 296) and repeat this procedure for bandwidths up to 25%.

  14. Full regression results for the vote share models can be found in Tables 3 and 4 in Appendix 1.

  15. Full regression results for the vote share models can be found in Tables 5 and 6 in Appendix 1

  16. Given the evidence in the literature on the types of districts that Republican women tend to run in—less women-friendly than Democratic women, less Republican-friendly than Republican men—it is worth considering whether these findings differ for women in more or less partisan districts. To test this, I included an interaction between the indicator for a female candidate winning the primary and lagged presidential vote share in a new set of models. The results, which are in the supplementary materials, are generally mixed. While there is some evidence that the effect attenuates in highly Republican districts, this pattern is sensitive to model specification. There is, however, evidence that the effect generally holds in both democratic-leaning and moderate districts, where the majority of the women in my sample are running.

  17. It is important to note that no evidence of a negative effect for Democrats is not necessarily evidence of no effect, as a test of equality of the coefficients across parties does not consistently reject the null hypothesis of no difference. The result of this test does not undermine the estimated effects for the Republican models; rather it means that we cannot be certain that the Democratic and Republican estimates are not in fact equal.

  18. While the exact cause of this difference in general election variance is unclear, it does not appear to be a function of incumbency: only three candidates within a 15 percent bandwidth are incumbents, all of them women.

  19. The results do not change if I instead use a continuous ‘year’ measure in the interaction (see Fig. 9). However, I prefer the simpler approach as interactions of this kind can be inconsistent for the RDD effect (Calonico et al. 2016) and running the models in the dichotomous manner allows me to verify the results by splitting the sample and estimating the effect on all races after 1990 absent the interaction. Doing so yields results that are identical to those presented.

  20. To simplify the construction of 95% confidence intervals, all of the predicted differences were estimated via simulation.

  21. Riding the wave of Newt Gingrich’s Contract with America, the Republicans seized control of Congress in 1994 by winning an additional 52 seats in the House. They retained control of the House until 2007, when the Democrats elected Nancy Pelosi as the new speaker.

  22. Specifically, I use the yearly campaign finance data provided in bulk by the Center for Responsive Politics at Note that a free login is required in order to access the data.


  24. Thus, this adjustment is important because it yields a standardized outcome measure that—unlike the raw data—should be generally comparable across years.

  25. The BCRA is commonly referred to as the McCain-Feingold Act. It restricted the use of soft money in federal campaigns and also increased the contribution caps for certain types of donors. For more details, see:

  26. A number of the primary candidates have no reported contributions in FEC data during the primary stage. I code all such individuals as having received $0 in contributions during the primary. Dropping these races from the data does not alter the results substantively, but does increase the variance of the estimates.

  27. Jenkins (2007) provides evidence for the hypothesis that women face additional barriers when fundraising, showing that women need to work harder than men to raise the same amount and that they often need to rely on a greater range of funding sources. This evidence, however, is not inconsistent with the theory of elite support put forward in this paper; indeed, a lack of support from party leaders and strategic donors would explain why women need to work harder to raise campaign funds in the first place.

  28. That is, if the findings in my paper are correct, more than 50% of Republican candidates in the general election will need to be women in order to achieve equal representation within the Republican party.


  • Anastasopoulos, L. (2016). Estimating the gender penalty in House of Representative elections using a regression discontinuity design. Electoral Studies, 43, 150–157.

    Article  Google Scholar 

  • Angrist, J. D., & Pischke, J.-S. (2009). Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton: Princeton University Press.

    Google Scholar 

  • Anzia, S. F., & Berry, C. R. (2011). The Jackie (and Jill) Robinson effect: Why do congresswomen outperform congressmen? American Journal of Political Science, 55(3), 478–493.

    Article  Google Scholar 

  • Arceneaux, K. (2001). The gender gap in state legislative representation: New data to tackle an old question. Political Research Quarterly, 54(1), 143–160.

    Google Scholar 

  • Ban, P., Llaudet, E., & Jr. Snyder, J. M. (2016). Challenger quality and the incumbency advantage. Legislative Studies Quarterly, 41(1), 153–179.

    Article  Google Scholar 

  • Broockman, D. E. (2014). Do female politicians empower women to vote or run for office? A regression discontinuity approach. Electoral Studies, 34, 190–204.

    Article  Google Scholar 

  • Burrell, B. (1994). A woman’s place is in the house. Ann Arbor: University of Michigan Press.

    Google Scholar 

  • Burrell, B. (2014). Political Parties and women’s organizations: Bringing women into the electoral arena. In S. J. Carroll & R. L. Fox (Eds.), Gender and elections: Shaping the future of American Politics (Vol. 3). Cambridge: Cambridge University Press.

    Google Scholar 

  • Calonico, S., Cattaneo, M. D., Farrell, M. H., Titiunik, R. (2016). Regression discontinuity designs using covariates. Working Paper.

  • Cattaneo, M. D., Keele, L., Titiunik, R., & Vazquez-Bare, G. (2016). Interpreting regression discontinuity designs with multiple cutoffs. Journal of Politics, 78(3), 1229–1248.

    Article  Google Scholar 

  • Caughey, D., & Sekhon, J. S. (2011). Elections and the regression discontinuity design: Lessons from Close U.S. House Races, 1942–2008. Political Analysis, 19, 385–408.

    Article  Google Scholar 

  • Center for American Women and Politics (CAWP). (2017). Women In The U.S. Congress 2017.

  • Cox, G. W., & Katz, J. N. (1996). Why did the incumbency advantage in U.S. House elections grow? American Journal of Political Science, 40(2), 478–497.

    Article  Google Scholar 

  • Crowder-Meyer, M., & Lauderdale, B. E. (2014). A partisan gap in the supply of female potential candidates in the United States. Research & Politics, 1(1), 1–7.

    Article  Google Scholar 

  • Darcy, R., & Choike, J. R. (1986). A formal analysis of legislative turnover: Women candidate and legislative representation. American Journal of Political Science, 30(1), 237–255.

    Article  Google Scholar 

  • Dolan, K. (2004). Voting for women: How the public evaluates women candidates. Boulder: Westview Press.

    Google Scholar 

  • Dolan, K. (2010). The impact of gender stereotyped evaluations on support for women candidates. Political Behavior, 32, 69–88.

    Article  Google Scholar 

  • Dolan, K. (2014). When Does Gender matter: Women candidates & gender stereotypes in American elections. Oxford: Oxford University Press.

    Book  Google Scholar 

  • Eggers, A., Fowler, A., Hainmueller, J., Hall, A. B., & Jr. Snyder, J. M. (2014). On the validity of the regression discontinuity design for estimating electoral effects: Evidence from over 40,000 close races. American Journal of Political Science, 59(1), 259–274.

    Article  Google Scholar 

  • Elder, L. (2008). Whither Republican women: The growing partisan gap among women in congress. The Forum, 6(1), 1–21.

    Article  Google Scholar 

  • Elder, L. (2012). The partisan gap among women state legislators. Journal of Women, Politics & Policy, 33(1), 65–85.

    Article  Google Scholar 

  • Fox, R. L., & Lawless, J. L. (2004). Entering the arena? Gender and the decision to run for office. American Journal of Political Science, 48(2), 264–280.

    Article  Google Scholar 

  • Fox, R. L., & Lawless, J. L. (2010). If only they’d ask: Gender, recruitment, and political ambition. Journal of Politics, 72(2), 310–326.

    Article  Google Scholar 

  • Fox, R. L., & Lawless, J. L. (2011). Gendered perceptions and political candidacies: A central barrier to womens equality in electoral politics. American Journal of Political Science, 55(1), 59–73.

    Article  Google Scholar 

  • Fulton, S. A. (2012). Running backwards and in high heels: The gendered quality gap and incumbent electoral success. Political Research Quarterly, 65(2), 303–314.

    Article  Google Scholar 

  • Gelman, A., Imbens, G. (2014). Why higher-order polynomials should not be used in regression discontinuity designs. NBER Working Paper Series Working Paper #20405.

  • Gerber, E. R., & Hopkins, D. J. (2011). When mayors matter: Estimating the impact of mayoral partisanship on city policy. American Journal of Political Science, 55(2), 326–339.

    Article  Google Scholar 

  • Gimpel, J. G., Lee, F. E., & Pearson-Merkowitz, S. (2008). The check is in the mail: Interdistrict funding flowsin congressional elections. American Journal of Political Science, 52(2), 373–394.

    Article  Google Scholar 

  • Hall, A. B. (2015). What happens when extremists win primaries? American Political Science Review, 109(1), 18–42.

    Article  Google Scholar 

  • Hayes, D., & Lawless, J. L. (2015). A non-gendered lens? Media, voters, and female candidates in contemporary congressional elections. Perspectives on Politics, 13(1), 95–118.

    Article  Google Scholar 

  • Hill, S. J., & Huber, G. A. (2017). Representativeness and motivations of the contemporary donorate: Results from merged survey and administrative records. Political Behavior, 39, 3–29.

    Article  Google Scholar 

  • Imbens, G., & Lemieux, T. (2008). Regression discontinuity designs: A guide to practice. Journal of Econometrics, 142(2), 615–635.

    Article  Google Scholar 

  • Jacobson, G. C., & Kernell, S. (1983). Strategy and choice in congressional elections. New Haven: Yale University Press.

    Google Scholar 

  • Jenkins, S. (2007). A womans work is never done? Fund-raising perception and effort among female state legislative candidates. Political Research Quarterly, 60(2), 230–239.

    Article  Google Scholar 

  • Kahn, K. F. (1994). Does gender make a difference? An experimental examination of sex stereotypes and press patterns in statewide campaigns. American Journal of Political Science, 38(1), 162–195.

    Article  Google Scholar 

  • Kanthak, K., & Woon, J. (2014). Women don’t run? Election aversion and candidate entry. American Journal of Political Science, 59(3), 595–612.

    Article  Google Scholar 

  • King, D. C., & Matland, R. E. (2003). Sex and the grand old party: An experimental investigation of the effect of candidate sex on support for a republican candidate. American Politics Research, 31(6), 595–612.

    Article  Google Scholar 

  • Koch, J. W. (2000). Do citizens apply gender stereotypes to infer candidates’ ideological orientations? The Journal of Politics, 62(2), 414–429.

    Article  Google Scholar 

  • Lawless, J. L., & Pearson, K. (2008). The primary reason for womens underrepresentation? Reevaluating the conventional wisdom. The Journal of Politics, 70(1), 67–82.

    Article  Google Scholar 

  • Lee, D. S. (2008). Randomized experiments from non-random selection in U.S. House elections. Journal of Econometrics, 142(2), 675–697.

    Article  Google Scholar 

  • Matland, R., King, D. (2002). Women as candidates in congressional elections. In Women Transforming Congress. Norman, OK: University of Oklahoma Press.

  • McDermott, M. L. (1997). Voting cues in low-information elections: Candidate gender as a social information variable in contemporary united states elections. American Journal of Political Science, 41(1), 270–283.

    Article  Google Scholar 

  • McDermott, M. L. (1998). Race and gender cues in low-information elections. Political Research Quarterly, 51(4), 895–918.

    Article  Google Scholar 

  • Mo, C. H. (2014). The consequences of explicit and implicit gender attitudes and candidate quality in the calculations of voter. Political Behavior, 37(2), 357–395.

    Article  Google Scholar 

  • Palmer, B., & Simon, D. (2008). Breaking the political glass ceiling: Women and congressional elections (Vol. 2). New York: Routledge.

    Google Scholar 

  • Palmer, B., & Simon, D. (2012). Women & congressional elections: A century of change. Boulder: Lynne Rienner Publishers.

    Google Scholar 

  • Pearson, K., & McGhee, E. (2013). What it takes to win: Questioning gender neutral outcomes in U.S. House elections. Politics & Gender, 9, 439–462.

    Article  Google Scholar 

  • Pettigrew, S., Owen, K., Wanless, E. (2014). U.S. House primary election results (1956–2010). doi:10.7910/DVN/26448. Harvard Dataverse, Version 3.

  • Pilpot, T. S., & Jr. Walton, H. (2007). One of our own: Black female candidates and the voters who support them. American Journal of Political Science, 51(1), 49–62.

    Article  Google Scholar 

  • Sanbonmatsu, K. (2002a). Gender stereotypes and vote choice. American Journal of Political Science, 46(1), 20–34.

    Article  Google Scholar 

  • Sanbonmatsu, K. (2002b). Political parties and the recruitment of women to state legislatures. The Journal of Politics, 64(3), 791–809.

    Article  Google Scholar 

  • Sanbonmatsu, K. (2006). Do parties know that “women win”? Party leaders beliefs about women’s electoral chances. Politics & Gender, 2, 431–450.

    Google Scholar 

  • Sanbonmatsu, K., & Dolan, K. (2009). Do gender stereotypes transcend party? Political Research Quarterly, 62(3), 485–494.

    Article  Google Scholar 

  • Seltzer, R., Newman, J., & Leighton, M. V. (1997). Sex as a political variable: Women as candidates and voters in U.S. elections. Chicago: The University of Chicago Press.

    Google Scholar 

  • Sen, M., & Wasow, O. (2016). Race as a ‘Bundle of Sticks’: Designs that estimate effects of seemingly immutable characteristics. Annual Review of Political Science, 19, 499–522.

    Article  Google Scholar 

  • Smith, E. R. A. N., & Fox, R. L. (2001). The electoral fortunes of women candidates for congress. Political Research Quarterly, 54(10), 205–221.

    Article  Google Scholar 

  • Swers, M., Thomsen, D. M. (2014). Who donates to women candidates? Gender, partisanship, and candidate donor networks. American Political Science Associtation Annual Conference.

  • Thomsen, D. M. (2015). Why so few (Republican) women? Explaining the partisan imbalance of women in the U.S. Congress. Legislative Studies Quarterly, 40(2), 295–322.

    Article  Google Scholar 

  • Thomsen, D. M., & Swers, M. L. (2017). Which women can run? Gender, partisanship, and candidate donor networks. Political Research Quarterly, 70(2), 449–463.

    Article  Google Scholar 

Download references


I am grateful to Chris Berry, Ryan Enos, Adam Glynn, Andrew Hall, Horacio Larreguy, Audrey Latura, Shauna Shames, and Jim Snyder for helpful suggestions and advice. I also benefited significantly from feedback by participants at Harvard’s Graduate Political Economy Workshop and Inequality and Social Policy Seminar.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Peter Bucchianeri.

Additional information

Data and code to reproduce all of the findings in this paper can be found at

Electronic supplementary material

Below is the link to the electronic supplementary material.

(pdf 570 kb)

Appendix 1: Regression models:

Appendix 1: Regression models:

See Tables 3, 4, 5 and 6 and Figs. 9 and 10.

Table 3 Quadratic polynomial regressions of general election vote share on female candidate primary win
Table 4 Local linear regressions of general election vote share on female candidate primary win
Table 5 Quadratic polynomial regressions of general election win on female candidate primary win
Table 6 Local linear regressions of general election win on female candidate primary win
Fig. 9
figure 9

Variation in effect of female candidate primary win on general election win probability across election cycles, republicans only

Fig. 10
figure 10

Local linear RDD estimates of female candidate primary win on general election contribution share, by party

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Bucchianeri, P. Is Running Enough? Reconsidering the Conventional Wisdom about Women Candidates. Polit Behav 40, 435–466 (2018).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Elections
  • Gender
  • Women candidates
  • Regression discontinuity design