POLITICAL GENDER STEREOTYPES

Scholarly research provides extensive documentation of the policy and trait stereotypes people hold when they think about women and men candidates for office. These stereotypes tend to follow traditional feminine and masculine images, with people seeing women candidates as more consensus oriented, compassionate, and honest than men and more focused on relationships with constituents. Men, on the other hand, are seen as more intelligent and competent than women, and they are thought to be stronger leaders and perform better in a crisis (Alexander and Andersen Reference Alexander and Andersen1993; Burrell Reference Burrell2008; Huddy and Terkildsen Reference Huddy and Terkildsen1993b; Kahn Reference Kahn1996; King and Matland Reference King and Matland2003; Lawless Reference Lawless2004; Leeper Reference Leeper1991; Paul and Smith Reference Paul and Smith2008; Sapiro Reference Sapiro1981/82). Stereotypes extend to assumptions about the policy interests and abilities of women and men in predictable ways, with men being seen as better suited for handling the military, economic issues, trade, and agriculture and women thought to be more competent at education, health care, children and family issues, and poverty than men (Alexander and Andersen Reference Alexander and Andersen1993; Brown, Heighberger, and Shocket Reference Brown, Heighberger and Shocket1993; Koch Reference Koch1999; Huddy and Terkildsen Reference Huddy and Terkildsen1993b; Rosenwasser and Dean Reference Rosenwasser and Dean1989). From an ideological perspective, women of both parties are thought to be more liberal than their male party counterparts (Koch Reference Koch2000, 2002; McDermott Reference McDermott1997).

Past literature also suggests that stereotypes can shape vote choice and influence whether voters will choose or reject women candidates (Fox and Smith Reference Fox and Smith1998; Huddy and Terkildsen Reference Huddy and Terkildsen1993b). For example, Lawless (Reference Lawless2004) finds that concerns about women's competence to handle terrorism could dampen support for the idea of a woman president. Sanbonmatsu (Reference Sanbonmatsu2002) finds that many voters have a basic preference for candidates of one sex or the other and that these preferences are determined, in part, by gender stereotypes. Koch (Reference Koch2002) finds that stereotypes about the perceived liberalism of women candidates, particularly Democratic women, can pull them further away from the average voter, which, he suggests, can result in these candidates losing votes. While some work finds that women's supposed strengths, such as greater personal honesty or an outsider status that makes them appear to be more ethical than men, can lead voters toward women candidates (Dolan Reference Dolan1998; McDermott Reference McDermott1997), most studies suggest that stereotypes have more negative consequences for women than benefits (Huddy and Terkildsen Reference Huddy and Terkildsen1993b; Fox and Smith Reference Fox and Smith1998).

STEREOTYPES AND LEVELS/TYPES OF OFFICES

Beyond the extensive literature on the presence of stereotypes is the question of whether stereotyped thinking about women's fitness for elective office is related to the office they seek. Because political gender stereotypes relate to whether voters see women as possessing the “right” skills and abilities to hold public office, it follows that these expectations could differ across offices with different levels and types of responsibility. In fact, previous research on support for women candidates suggests that the levels and types of offices women seek interact with stereotypes to shape women's opportunities. There is a significant literature that focuses on the presidency, finding that voters value male personality traits and policy expertise for this office (Adams Reference Adams1975; Kinder Reference Kinder, Lau and Sears1986; Mueller Reference Mueller, Moore and Spitze1986; Rosenwasser and Seale Reference Rosenwasser and Seale1988).

This same literature finds that public preferences for male characteristics are strongest when considering the presidency and become less limiting as lower-level offices are considered. For example, public desires for strong leaders or those with experience dealing with foreign policy may work against women seeking the presidency, but they may be of much less importance when a woman seeks local office (Adams Reference Adams1975; Burrell Reference Burrell2008; Dolan Reference Dolan1997). Indeed, Huddy and Terkildsen (Reference Huddy and Terkildsen1993a) find clear evidence that voters expect male traits and policy expertise from candidates for national office but that there are diminished concerns about these qualities when people consider local office. Female traits and abilities are only valued in candidates for local-level office, which is consistent with other work that focuses on level of office (Sigelman, Sigelman, and Fowler Reference Sigelman, Sigelman and Fowler1987).

Another dimension on which gender stereotypes can have an impact involves the types of offices women seek. Executive offices such as president, governor, or mayor are more clearly stereotyped with regard to their responsibilities and clearly place officeholders in a position of single authority compared with legislative office or many judicial positions (Fox and Oxley Reference Fox and Oxley2003). As a result, candidates running for an executive office might be advantaged by exhibiting stereotypical male traits such as decisiveness and leadership, whereas female traits such as consensus building might be advantages when running for a legislative office. Findings from past research on the impact of stereotypes on women running for different types of offices supports these expectations. Adams (Reference Adams1975) finds that hypothetical women candidates were penalized for not having the appropriate characteristics when they ran for executive office (president, governor, mayor) more often than when they ran for legislative office (Congress, town council). Huddy and Terkildsen (Reference Huddy and Terkildsen1993a) find that voters expect male characteristics from candidates for executive office more so than legislative candidates and expect female expertise from legislative candidates as opposed to executive office candidates. Sigelman, Sigelman, and Fowler (Reference Sigelman, Sigelman and Fowler1987) find greater support for a female state supreme court candidate who displayed more female than male characteristics.

REMAINING QUESTIONS

The conclusion that voters filter their evaluations of women candidates through the lens of the offices they seek makes intuitive sense. If gender stereotypes focus on whether women and men have the “right” mix of policy competence and personal traits, what is seen as appropriate and valued may change across offices. However, before we accept this conclusion, we need to acknowledge two realities of the existing work on the link between stereotypes and level/type of office. First, this literature is small and fairly dated. There are only a handful of works, and almost none since Huddy and Terkildsen's (Reference Huddy and Terkildsen1993b) seminal article, that address the impact of stereotypes across different offices. Instead, most contemporary work on the impact of gender stereotypes focuses on real or hypothetical women candidates running for one type or level of office, most often the presidency or Congress (Dolan Reference Dolan2014; Fridkin and Kenney Reference Fridkin and Kenney2009; Fridkin, Kenney, and Woodall Reference Fridkin, Kenney and Woodall2009; Hayes Reference Hayes2011; Koch Reference Koch2002; Lawless Reference Lawless2004; McDermott Reference McDermott1998; Sanbonmatsu Reference Sanbonmatsu2002). While these works provide great insight into the influence of gender stereotypes on women candidates, they do not allow us to examine whether the impact of stereotypes differs across the range of offices women seek.

The second limitation of past work on level/type of office is that it has largely relied on hypothetical candidates in experiments or public opinion surveys (Adams Reference Adams1975; Dolan Reference Dolan1997; Huddy and Terkildsen Reference Huddy and Terkildsen1993a, Reference Huddy and Terkildsen1993b; Mueller Reference Mueller, Moore and Spitze1986; Rosenwasser and Seale Reference Rosenwasser and Seale1988; Sigelman, Sigelman, and Fowler Reference Sigelman, Sigelman and Fowler1987). Much of this experimental research presents subjects with the speech or campaign biography of a fictional candidate and asks them to offer their assessments of the candidate. Survey-based work usually asks respondents about their attitudes toward a hypothetical “woman” candidate proposed to be seeking some office. While these approaches have their own value, they lack generalizability and offer us little understanding of how people employ stereotypes and evaluate actual candidates in real-world elections. The isolation of candidate sex results in a host of important variables, such as the impact of level/type of office, political party, and incumbency, remaining unexamined, which could lead to results that appear to give candidate sex outsize importance in voter decision making (Conover and Feldman Reference Conover and Feldman1989; Downs Reference Downs1957; Lau and Redlawsk Reference Lau and Redlawsk2001; Popkin Reference Popkin and Grofman1993; Rahn Reference Rahn1993). This leaves us with a conventional wisdom about the impact of gender stereotypes that is not well suited to predicting what will happen in the more complex world of actual elections.

With these limitations in mind, the challenge for researchers is to move beyond the hypothetical and examine whether and when voters use gender stereotypes when they are choosing among women and men candidates in actual elections. In doing so, the current research extends our earlier work on the impact of gender stereotypes on voting for women candidates, which finds a relatively limited influence for stereotypes, particularly vis-à-vis other important political influences such as partisanship and incumbency (Dolan Reference Dolan2014). However, this earlier work only examined elections for the U.S. House of Representatives, which leaves us with questions about whether gender stereotypes have a different influence when women run for different types of offices. One limitation in conducting this analysis is that so much of the previous research on stereotypes across different offices has been conducted with experiments and hypothetical candidates, which leaves us without clear theoretical expectations or clear support from the data. As a result, we are left to follow the logical assumptions about stereotypes mattering more as the level or type of office increases or becomes more prestigious.

First, we test the hypothesis that the level of office should be related to voter use of gender stereotypes. Here we expect that male policy and trait stereotypes will be positively related to vote choice for candidates as the level of office they seek increases and that female policy and stereotype traits will be more likely to be positively related to voting for lower-level offices. In this project, our definition of level of office is more truncated than is ideal, as we include U.S. House, U.S. Senate, and governor. For our purposes, House races, which focus on local constituencies and are lower-visibility elections than those for Senate or governor, become our “lower-level” races; these are compared with the dynamics in Senate and governor elections. Second, we test the hypothesis that the type of office a woman seeks should drive stereotype use, with male policy and trait stereotypes being positively related to vote choice in races for executive office and female policy and trait stereotypes being positively related to voting for legislative office. This leads to a comparison of House and Senate elections with races for governor.

Having framed these hypotheses, we must keep in mind that all offices for which candidates run have a “level” and a “type” at the same time, which can create situations in which expectations may conflict. For example, we could hypothesize that voters are more likely to employ male policy and trait stereotypes when voting for a national-level office such as the U.S. Senate (Huddy and Terkildsen Reference Huddy and Terkildsen1993a). At the same time, the U.S. Senate is a legislative office, which could lead to the competing hypothesis that voters are more likely to employ female trait and policy stereotypes for this office compared with an executive position, which is thought of as more “male” (Fox and Oxley Reference Fox and Oxley2003). As a result, we must be open to exploring the results of this examination of these three offices for which women regularly run and win—U.S. House, U.S. Senate, and governor—and determining whether they conform to expectations or lead us to rethink our assumptions.

DATA AND METHODS

The data for this article come from a survey project that is one of the first large-scale examinations of public opinion specifically designed to examine gender attitudes, gender stereotypes, and women candidates conducted during an election and asking respondents about actual candidates. To accomplish this, a two-wave panel survey of 3,150 U.S. adults was conducted during the midterm elections of 2010. Funded by the National Science Foundation, the surveys were conducted in an online environment by Knowledge Networks during September and October/November of 2010.Footnote ¹ The nationally representative sample was drawn from 29 states and was stratified to include people who experienced either mixed-sex or single-sex races for the U.S. House, U.S. Senate, and governor. A series of questions in the first wave of the survey were designed to probe respondents' attitudes about the place of women in American politics, their abstract gender stereotypes, and a host of other attitudinal, behavioral, and political measures. The second wave of the survey was designed to gather respondents' reactions and behaviors toward the specific candidates they experienced in their elections. This article focuses on findings for respondents who experienced 91 House races, 11 Senate races, and 5 races for governor in the 29 states in the sample.

As a result of the intentional panel design, we have survey responses from the same respondents at two different points in the campaign cycle, fully two months apart, on a wide range of attitudes and behaviors.Footnote ² This design allows us to measure the abstract gender stereotypes respondents may hold separately from the specific candidate evaluations they make, thereby greatly reducing the possibility that responses to the abstract stereotype measures colored behaviors toward particular candidates. It also allows us to examine a wide range of traditional influences on vote choice, providing the most comprehensive understanding of how voters evaluate women candidates and whether they choose to vote for them.

Measuring gender stereotypes and reactions to women candidates can be a challenging task because of concerns that people will hide their true feelings behind socially acceptable responses of egalitarianism. To minimize this possibility, we draw on current research on surveying sensitive topics that demonstrates that offering more anonymity leads to more truthful responses (Schaeffer and Presser Reference Schaeffer and Presser2003; Tourangeau and Smith Reference Tourangeau and Smith1996). The data for this survey were collected in a Web-TV environment, with respondents answering questions alone in their own homes. This allows respondents the greatest level of anonymity by eliminating interaction with an interviewer. Recent research comparing Internet surveys with other methods of collecting survey data clearly demonstrates lower levels of socially desirable answers from people who were surveyed using this format (Dennis and Li Reference Dennis and Li2007; Heerwegh Reference Heerwegh2009; Kreuter, Presser, and Tourangeau Reference Kreuter, Presser and Tourangeau2008).

The elections of 2010 provide a particularly good opportunity to investigate the impact of gender stereotypes on the fortunes of women candidates. First, there was no presidential election to overwhelm voters and compete for their attention. Second, 2010 offers researchers the first election cycle in which to examine gender stereotypes after the historic candidacies of Hillary Clinton and Sarah Palin. In addition, a record number of women candidates ran for statewide and congressional office that year, and this group of women allowed for more partisan diversity among women candidates than is often the case in U.S. elections, with approximately 40% of the women candidates for Congress and governor running as Republicans (CAWP 2012). Finally, the midterm elections in 2010 were more competitive than many in recent memory, particularly for Congress (Cook 2010).

The survey for this project asked respondents for their vote choice in races for U.S. House, U.S. Senate, and governor. These data give us a way to examine the different influence of stereotypes based on the level and type of office being sought by women candidates. With regard to the level of office, we will examine the impact of stereotypes on vote choice decisions for House races and compare them to races for Senate and governor. This will allow us to compare the influence of gender stereotypes on vote choice decisions in legislative and executive offices.

VARIABLES

ANALYSIS

Stereotypes

The first step in the analysis is to briefly consider the stereotypes respondents possess and examine whether there are patterns in who holds them. Tables 1 and 2 present the frequencies for the stereotype measures. In general, these data are in line with recent research that suggests that stereotyped impressions of women and men are easing (Burrell Reference Burrell2008; Dolan Reference Dolan2010; Eagly and Carli Reference Eagly and Carli2007; Fridkin and Kenney Reference Fridkin and Kenney2009; Lawless Reference Lawless2004). With regard to policy competence (Table 1), there are only two issues on which a majority see women or men as better than the other sex at handling the issue: child care and abortion. Not surprisingly, women are seen as better able to handle both of these issues than men. On each of the other issues, the modal response is “no difference,” with anywhere from 55% of respondents (national security) to 75% of respondents (economy, deficit) saying that they see no difference in women's and men's ability to handle these issues. For the issues on which the modal response is no difference between women and men, the next most likely response on each issue is in the expected stereotyped direction.

Table 1. Abstract policy stereotypes, all respondents

Table 2. Abstract trait stereotypes, all respondents

The same general pattern is true when we examine stereotypes about traits (Table 2). For each of the eight traits examined, large majorities see no difference between women and men. For most of the issues, those who hold stereotypes hold them in the expected direction—women as more compassionate, men as more decisive—with two exceptions. Women and men are essentially seen as equally likely to be able to build consensus, and women are seen as more intelligent than men. While these findings of similar evaluation of women and men may appear to fly in the face of the traditional literature on stereotypes, they are consistent with the most recent examinations of candidate trait evaluations, both from survey and experimental data (Brooks Reference Brooks2013; Pew Research Center 2008).

Having examined the distribution of political gender stereotypes about women and men, it may be helpful to more fully understand the sources of these stereotypes among voters. To that end, Table 3 presents an analysis of the determinants of gender stereotypes with a focus on demographic and political characteristics of the respondents. The first thing to notice is that stereotype beliefs appear to be a function of respondent sex and political beliefs. Women and men are significantly different in their stereotype perspectives, with women being more likely than men to see women as better at female policies and more likely to possess female traits (which is the expected stereotype direction) but less likely to see men as better at male policies and likely to possess male traits. Men hold the opposite positions, which results in women and men seeing candidates of their own sex as “superior” on each of the four dimensions. Beyond respondent sex, the other demographic characteristics do not appear to influence stereotype holding at all. There are no differences in stereotypes among respondents of different racial groups, education levels, or age groups, with the exception that older people are more likely to see women candidates as more likely to possess female traits.

Table 3. Stereotype predictors

Note: Ordinary least squares regression for respondent stereotypes.

* p < .05, two-tailed test of significance. Clustered standard errors in parentheses.

When we look at respondent party identification and ideology, we see some significant influences. Democrats are more likely than Republicans to hold the expected stereotype that women are better at female policy areas, while Republicans are more likely to hold the expected male stereotypes about both male policy and traits. Conservative respondents are also more likely to hold the expected stereotype that men are better suited to handle male policy areas. One final political variable acknowledges that because stereotypes are shaped, in part, by political and ideological influences, we might expect stereotype holding to be related to where a respondent lives. Stereotypes might be more prevalent among people who live in more conservative states or areas of the country. To examine this, we include a variable that measures the two-party national vote in 2004 and 2008 for each state represented by races in the dataset. Interestingly, this variable is not significantly related to any of the stereotype measures, which suggests that the pattern of stereotype use among respondents is not different across states.

U.S. House

To test our hypotheses, we begin by examining whether gender stereotypes are related to voting for women candidates, and then we compare these findings across the different offices to see whether the impact of stereotypes differs across offices. Table 4 presents a series of models that analyze voting for women who run for each of the three offices analyzed. There are two models for each office: one of the determinants of vote choice when a Democratic woman runs against a Republican man and the second in which a Republican woman runs against a Democratic man. Taking House races with Democratic women candidates first (column 1), we see that none of the four measures of policy and trait stereotypes is significantly related to vote choice in these races. This suggests that people are not employing their gender stereotypes in choosing between these women and men candidates, a finding that runs counter to the notion that gender stereotypes matter most in relatively low-visibility elections such as the U.S. House (McDermott Reference McDermott1997, 1998).

Table 4. Vote for woman candidate: Mixed-sex House, Senate, and governor races

Notes: Independents omitted due to perfect prediction in analyzing Republican women candidates in the House, both sets of candidates in the Senate, and Democratic women candidates in governor's races. No Republican women candidates for Senate were incumbents.

* p < .05, two-tailed test of significance. Clustered standard errors are in parentheses for House races. Standard errors in parentheses for Senate and governor races. Spending is excluded in analyzing Democratic women gubernatorial candidates because only two races (Florida and Wyoming) featured a Democratic woman running against a Republican man.

However, we see a different influence for stereotypes in races with Republican women candidates. The model in column 2 suggests the first evidence that female stereotypes can work to hurt women candidates, as we find that voters who hold expected stereotypes about women's superiority on female policy issues are actually less likely to vote for Republican women. Instead, these voters are more likely to vote for the Democratic male opponent. Beyond this, we also see a positive relationship between female trait stereotypes and voting for women candidates, with voters who see women as more likely to be compassionate and consensus oriented being more likely to vote for the Republican woman.

COMPARISONS ACROSS LEVEL AND TYPE OF OFFICE

If the previous research employing experimental and hypothetical designs that finds public evaluations of women candidates being shaped by the office they seek is accurate, we would expect to see different patterns in the power of gender stereotypes to influence voting for women candidates. However, as Table 4 demonstrates, there are very few clear or consistent patterns in the way that voters employ stereotypes in the presence of women candidates and few relationships that conform to expectations. First, while gender stereotypes are not a strong or consistent influence on vote choice in these races with women candidates, we can say that voters in these races are a bit more likely to employ stereotypes in their evaluations of women running for Congress (6 of 16 coefficients reach significance) than they do for governor (2 of 8 coefficients reach significance), which is in line with previous research (Dolan Reference Dolan1997; Huddy and Terkildsen Reference Huddy and Terkildsen1993a). However, we do not see evidence of support for the expectation that people would employ male policy and trait stereotypes to evaluate women candidates for these offices more often than for other offices. While we do see one instance of a male stereotype being related to vote choice—the negative impact of male policy stereotypes on voting for Democratic women Senate candidates—female policy and trait stereotypes are much more likely to be related to voting for women for the House and Senate. This is the same pattern that we see in terms of voting for governor, with female policy and trait stereotypes appearing to be more relevant than male stereotypes. From these comparisons, it is hard to see voters relying on different criteria to evaluate women for offices of different levels of visibility.

The same is true when we think about executive versus legislative positions. Executive elections such as those for governor should be the races where we see people relying on male policy and trait stereotypes (Adams Reference Adams1975; Huddy and Terkildsen Reference Huddy and Terkildsen1993a). Instead, these stereotypes are not related to voting for any women candidate for governor. Perhaps most striking here is the complete lack of relationship between male trait stereotypes and executive office. The early works in the field suggested that voters display clear preferences for male personality traits in the candidates they choose for higher-level and executive office (Adams Reference Adams1975; Huddy and Terkildsen Reference Huddy and Terkildsen1993a; Sigelman, Sigelman, and Fowler Reference Sigelman, Sigelman and Fowler1987), but this analysis from real-world elections suggests a disconnect between what people might say they value in abstract situations and what they actually employ in making real voting decisions.

There is one other interesting observation to note from these findings. For each office examined, the relationship of gender stereotypes to vote choice for women candidates is different for Democratic and Republican women. In House races, none of the stereotypes is significantly related to vote choice for Democratic women candidates. For Republican women, traditional female policy stereotypes pull voters away from them and toward their Democratic male opponents, while female trait stereotypes result in their receiving a boost in support from some voters. For Senate races, the dynamic shifts, with stereotypes being more closely related to vote choice for Democratic women candidates than for Republican women. Voters employ their female policy and trait stereotypes along with their male policy stereotypes when faced with a Democratic woman Senate candidate, but they only rely on their female trait stereotypes when considering Republican women for this office. Finally, voting for Democratic women candidates for governor is higher among voters who hold traditional female policy stereotypes. For Republican women running for governor, female trait stereotypes are the important influence. The finding that stereotypes have a different relationship to vote choice decisions involving women candidates of different parties is in keeping with research that finds that party and candidate sex often interact in shaping the attitudes and behaviors of voters (Koch Reference Koch2000, 2002; Lawless and Pearson Reference Lawless and Pearson2008; Sanbonmatsu and Dolan Reference Sanbonmatsu and Dolan2009).