Baker, Bus Driver, Babysitter, Candidate? Revealing the Gendered Development of Political Ambition Among Ordinary Americans

Men’s candidacy numbers are also higher in 2018 than 2016, but women’s numbers have increased at a higher rate than men’s (Dittmar 2018).

Each of these occupations was listed—often more than once—among the labels city council candidates in California selected as ballot designations between 1995 and 2011 (CEDA 2011).

Specifically, 8% of female/1% of male state legislators were nurses; 3% of female/6% of male state representatives and 1% of female/8% of male state senators were farmers; 4% of female/0.2% of male state representatives and 5% of female state senators were homemakers (Carroll and Sanbonmatsu 2013).

I use the terms “ordinary” and “elite” Americans throughout this paper to differentiate between Americans who have achieved educational and professional status nearer to the level of the average American citizen or below and those Americans who hold higher levels of education and professional achievement typical of traditional elected officials. Among my survey respondents, I delineate “elites” as those with graduate or professional degrees—the top 15% of my sample by educational attainment. When comparing my sample to other studies of political ambition, I consider other samples as focusing largely on elites when they study individuals who have achieved advanced professional positions in, for example, law, business, education, or politics, including those who have held political offices. While I distinguish between individuals based on their socioeconomic characteristics for purposes of studying the development of political ambition, research indicates that individuals with “ordinary” socioeconomic backgrounds are effective officeholders (Carnes and Lupu 2016) and that representation by elected officials from diverse socioeconomic backgrounds helps provide substantive representation for a diverse public (Carnes 2013).

Overcoming these challenges is not a simple task, as the popularity of recent work by elite women can attest (Slaughter 2012; Sandberg 2013). These challenges help explain why women continue to be underrepresented in many elite professions (Catalyst 2018). While this study focuses on how women overcome these challenges to pursue political office, the results here may also identify factors leading women to pursue other elite positions as well.

Replication materials are available in the Political Behavior Dataverse at https://​doi.​org/​10.​7910/​DVN/​6FBXGL.

Further details about the characteristics of my survey sample, a comparison of my sample to an alternative sample of the US population, and descriptive statistics for all the variables used in models in this paper are available in Online Appendix Tables A.1 and A.2.

Respondents are coded as having low political ambition (0) if they responded “No, I’ve never really thought about it,” while other responses “Yes, I’ve sometimes thought about running,” “Yes, I’ve thought a lot about running,” “Yes, I’ve seriously investigated the possibility of running,” “Yes, I’ve run for public office,” and “Yes, I’ve held an elected office” are coded as displaying political ambition (1). Most of the 14% of respondents who expressed political ambition indicated that they have “sometimes thought about running.” Fewer than 5% responded affirmatively to the other statements indicating political ambition combined.

I measure self-perceived qualifications by asking “Overall, how qualified do you feel you are to do the job of an elected official?” and allowing respondents to indicate they are not at all qualified (1) to very qualified (4). In addition to measuring respondents’ self-perceived Qualifications to Hold Office, I also asked respondents to indicate their perceptions of their Qualifications to Run for Office. I do not include both the Run For and Hold Office qualifications variables in my models due to high collinearity. My findings are substantively similar in models using either qualifications variable (Online Appendix Tables A.3 and A.4). The only change from Table 2 is that the interaction of female and qualifications to run for office among my elite respondents becomes statistically significant at p = 0.08, suggesting that gender does interact with self-perceived qualifications to run to affect ambition among elites in my sample as it does among elites in previous ambition studies.

Race and ethnicity may intersect with gender to shape how political ambition develops. In interviews with potential and actual candidates in Texas, Frederick (2014) finds that white women were more likely to indicate their decisions to run were largely due to others’ encouragement, while black women were more likely to explain their decisions based on their political ambition, confidence in their qualifications, and personal initiative. These differences could yield differences by race in how self-perceived qualifications and recruitment influence ambition. Additionally, Shames’ research on potential candidates pursuing law or policy degrees reveals that women of color demonstrate lower political ambition because of their concerns about racial and gender bias—concerns higher among women of color than white women or men of color—and their lower likelihood, relative to white women, white men, or men of color, of believing problems important to them could be solved through politics (Shames 2015). Investigating these intersections is unfortunately beyond the scope of this paper. The analyses in this paper do reveal that, once one controls for the other variables in my models, few effects of racial and ethnic identity on political ambition emerge. While Table 3 indicates Black men have lower ambition than white men, this difference is not echoed in my results among women nor among other racial and ethnic groups. I look forward to explaining this finding and its connection to racial inequality in political representation and to exploring how race and gender intersect to affect ambition in future work.

Political Participation measures what proportion of four political activities each respondent has engaged in during the last few years: regularly voting, regularly contacting elected officials to express their views, volunteering for a campaign organization (e.g., political party or candidate’s campaign), and donating money to political causes or campaigns. Community Participation measures what proportion of two community activities each respondent engaged in during the last few years: volunteering for a civic group (e.g., the Chamber of Commerce, a union, or the Sierra Club), and volunteering for other community organizations like churches or community service organizations. Political Knowledge identifies the proportion of correct answers respondents gave to five factual questions about American government (e.g., which party has the most members in the U.S. House, which of the two major parties is usually referred to as the more conservative). And, Political Efficacy is a scale measuring respondents’ agreement with four statements about the government (e.g., people like me don’t have any say about what the government does, sometimes politics and government seem so complicated that a person like me can’t really understand what’s going on).

Given the smaller number of elites in my sample (181), standard errors are large in my elite model. Thus, it is useful to compare mass ambition to elite ambition not only by examining distinctions between the ordinary and elite respondents to my survey, but also by comparing ordinary respondents to my survey to previous studies’ larger samples of elites as in the final column of Table 2.

Running my analyses on my full sample of respondents – adding back in the 15% of my sample with the highest educational attainment—produces results (Online Appendix Table A.5) that are substantively similar to those displayed in Table 3.

As an alternative way to evaluate differences in men’s and women’s ambition development, I also ran a single logistic regression model including interactions between respondent gender and each independent variable. The results of this analysis (Online Appendix Table A.6) confirm those displayed in Table 3. As an additional robustness check, I also conducted each of my analyses using Rare Events Logistic Regression (ReLogit) (Tomz et al. 1999; King and Zeng 2001), to address the fact that having political ambition is a fairly rare event in my mass data—14% of all respondents (13% of those with ordinary education and 19% of those with elite education levels) indicated having at least thought about running for office. These results (Online Appendix Tables A.7 and A.8) are also substantively similar to those presented here.

Specifically, in Lawless and Fox’s (2010) landmark study of business, legal, education, and political professionals, they find very few instances in which sex interacts with predictors of political ambition in significant ways. In models predicting political ambition they report no significant interactions between sex and: education, income, political knowledge, political interest, and political participation (54, note 6); marital status or the presence of children in one’s household (78); party identification (95, note 12); or recruitment from a political source (109). The only interaction they describe as significant in their models of political ambition among elites is between self-perceived qualifications and sex; women’s self-perceived qualifications affect ambition more than men’s (120).

In these figures, the 90% confidence intervals for men’s and women’s estimates often overlap. However, testing for differences in the factors shaping ambition for men and women using models interacting respondent gender with each of my independent variables (Table A.6 in the Online Appendix) and using post hoc Wald tests for equality between the effects of a given independent variable on men’s and women’s ambition following those models, demonstrates that there are statistically significant differences between men and women in the effects on political ambition of: self-perceived qualifications to hold office (Wald test p < 0.06) and run for office (p < 0.1), encouragement from political sources (p < 0.05) and personal sources (p < 0.05), and being married (p < 0.01). The statistically significant gender differences in these tests are not reflected in the confidence intervals displayed in my figures because non-overlapping 90% or 95% confidence intervals is not the equivalent test to comparing values across groups at the p < 0.05 significance level; rather, these are much more conservative tests (Payton et al. 2003; Schenker and Gentleman 2001). In sum, my analyses indicate men’s and women’s ambition is affected differently, at standard levels of statistical significance, by the variables in Figs. 1, 3, and 4. Additionally, Table 3 demonstrates that the effects of political participation and education on men’s ambition are statistically different from zero and this is not the case for women’s ambition, however the interaction models in Table A.6 and Wald tests comparing the effects of these variables on men’s and women’s ambition cannot confirm a statistical difference between the effects of the variables in Fig. 2 on men’s and women’s ambition.

I do not include both qualification variables in the same model due to their collinearity. Full results from models with the qualification to run for office variable are in Online Appendix Tables A.3 and A.4.

Note that the scale in Fig. 3 is distinct from that in the other figures, to display the particularly large change in women’s probability of political ambition resulting from changes in recruitment.

Some support for this interpretation can be found in Online Appendix Table A.9 where I display the results of models like those in Table 3, adding interactions between my encouragement variables and a measure of whether or not a respondent would be concerned about “losing the election” (coded 1 if concerned, 0 if not) if they were “thinking of running for public office in the next few years.” For men, this variable is not significantly related to ambition, nor are interactions between concern about losing and either encouragement variable. For women, on the other hand, there is a positive and marginally significant relationship between concern about losing and personal (though not political) encouragement; being encouraged by personal sources appears to increase ambition even more among women who are worried about losing if they were to run.

Traditional Gender Roles are measured by either agreement (coded 0) or disagreement (1) with the statement “women should have an equal role with men in running business, industry, and government.” I separate the models interacting traditional roles with Married and with Has Child due to collinearity.

My results, however, suggest a somewhat different interpretation than that typically made about male and female officeholders’ marital status. Rather than marriage decreasing women’s likelihood of seeking office, it may instead be that marriage is so beneficial to men’s determination of whether they can run successfully that married men dominate their single counterparts in developing political ambition.

I also ran the models in Table 3 using a variable indicating a respondent has a child under the age of 6. My results are substantively unchanged in these models (see Table A.11 in the Online Appendix).

Table 4 displays these figures for my entire sample of the American public—not just the sample without elite education that I include in my analyses in Tables 2 and 3. In this way, Table 4 provides a clearer picture of the proportion of men and women the entire American public equipped with the characteristics that lead to ambition.