This work was supported by a Grants-in-Aid Award from the Society for the Psychological Study of Social Issues (SPSSI) awarded to the first author.
Sarah J. Gervais, Department of Psychology, University of Nebraska-Lincoln, Lincoln, NE 68588–0308 [e-mail: email@example.com].
This research compares prejudice toward female politicians Hillary Clinton and Sarah Palin through the lens of role congruity theory. We measured participants’ evaluations of stereotypicality, competence, warmth, and voting likelihood. Consistent with hypotheses, Clinton was evaluated as less stereotypically feminine and less warm than Palin, whereas Palin was evaluated as less competent than Clinton. Furthermore, participant gender, benevolent sexism, hostile sexism, and political orientation predicted differential voting likelihood for Clinton and Palin. Implications for role congruity, ambivalent sexism, and female politicians are discussed.
Women have come a long way from the time when wearing a pair of pants was considered “borrowing from the boys.”So it would be highly regressive to suggest that the candidate is using trousers to heighten the perception that she can be as tough as a man. And yet…Givhan (2007)
There are few U.S. female politicians who have garnered as much attention and controversy as Hillary Clinton and Sarah Palin. Hillary Clinton is a Democrat and the 67th U.S. Secretary of State. Clinton was the First Lady during the 1990s and a senator from New York in the 2000s. Clinton was initially a front-runner for the presidency in the 2008 U.S. presidential election, even though no woman has been nominated for the presidency from a major political party in the United States. Sarah Palin is a Republican and was the first female governor of Alaska. Palin was also John McCain's vice presidential running mate in the 2008 U.S. presidential election, and she was only the second female vice presidential candidate on a major U.S. ticket. Many (including Palin) speculate that she will make a run for the presidency in the 2012 election.
Both Clinton and Palin have been remarkably successful in politics compared with other female politicians (Ryan, Haslam, & Kulich, 2010), but one feature that differentiates Clinton and Palin is the degree to which they violate traditional gender norms. Although they both violate traditional gender roles because they are female leaders in a stereotypically masculine domain, they differ in stereotypicality relative to one another, at least as they appear through media depictions. Clinton appears to possess fewer stereotypically feminine characteristics (e.g., appearance, mannerisms, interests) than Palin. For example, commentators frequently characterize Clinton's behavior and appearance as not appropriately feminine (Conners, 2010). For example, in the above quote, Givhan (2007) links Clinton's penchant for sporting pantsuits to attempting to appear masculine. In addition, Clinton is often depicted as a radical feminist (Jamieson, 1995) and a perpetrator of violence (Zurbriggen & Sherman, 2010) in political cartoons, which is counter-stereotypical for women. On the other hand, Palin's appearance and mannerisms are often regarded as appropriately feminine. During her speech at the Republican National Convention, for example, she famously likened herself to an average hockey mom (Parker, 2008). Additionally, she was Miss Wasilla 1984 and runner up in the Miss Alaska beauty pageant (Harnden, 2008). Furthermore, although both Clinton and Palin are mothers, Palin's role as a “super mom” has been especially highlighted in the press.
The purpose of the present work was two-fold. First, we extend and elaborate role congruity theory and research (Eagly & Karau, 2002) by examining whether stereotypicality moderates prejudice toward female leaders. Second, we directly compare prejudice against Clinton and Palin. Although social commentators have extensively contrasted Clinton and Palin, few psychological studies have examined perceptions of Clinton or Palin (e.g., Dwyer, Stevens, Sullivan, & Allen, 2009; Hall, 2009; Heflick & Goldenberg, 2009), and none have directly compared the two. This consideration also is particularly timely. Although women have reached more powerful positions than ever before, the gender balance in politics remains remarkably unequal. For example, 51% of workers in management and professional occupations are women; however, only 18% of representatives in Congress are women. Identifying factors that exacerbate (or temper) prejudice toward female leaders may reduce these inequities.
We argue that prejudice may be directed toward both Clinton and Palin, but it may manifest differently because Clinton and Palin vary in terms of stereotypicality. Specifically, three questions motivated the present work: Will prejudice be directed toward Clinton and Palin? Will prejudice directed toward Clinton and Palin be different? And what might prejudice directed toward Clinton and Palin look like? Toward that end and to derive testable hypotheses, we review theory and research on role congruity (Eagly & Karau, 2002), stereotype content (Fiske, Cuddy, Glick & Xu, 2002), and ambivalent sexism (Glick & Fiske, 1996). We then examine whether stereotypicality moderates prejudice toward female leaders. Specifically, we measured U.S. undergraduates’ evaluations of stereotypicality, competence, warmth, and voting likelihood for Clinton and Palin. We also examined gender and sexism as moderators.
Will Prejudice Be Directed Toward Clinton and Palin?
According to role congruity theory, people are evaluated positively when their characteristics are consistent with their social roles. Prejudice results from an inconsistency between one's attributes or behaviors and one's role. Role congruity theory posits that female leaders may experience prejudice because the female gender role is inconsistent with the leader role. Consequently, female leaders are in a double bind: They cannot simultaneously act in ways that are consistent with both the female gender role and the leader role because the roles are contradictory (Eagly & Karau, 2002). The female gender role involves being sympathetic, nurturing, dependent, weak, and emotional (Eagly & Karau, 2002). However, the leader role is regarded as requiring attributes that are stereotypically associated with men but not with women (e.g., agency, logic, reason, and strength; Eagly, Karau, & Makhijani, 1995). In fact, the leader role associations with masculinity are especially evident in U.S. politics. For example, all American presidents have been men with stereotypically masculine traits, and the mythology surrounding the president is often masculine in nature (Chernow, 2010). Thus, the male gender role is congruent with the leader role, whereas the female gender role is incongruent with the leader role (Heilman, 2001). Although Clinton and Palin have been relatively successful in politics, role congruity theory suggests that they may be evaluated unfavorably because the female role is incongruent with the leader role. Consistent with this notion, female politicians are often depicted in the media much more negatively than their male counterparts (Zurbriggen & Sherman, 2010).
Will Prejudice Directed Toward Clinton and Palin Be Different?
According to role congruity theory, all female leaders should experience prejudice because they cannot simultaneously be good women and good leaders; however, the type of prejudice that female leaders experience may depend on whether they violate descriptive or prescriptive norms. Descriptive norms are beliefs about actual differences between men and women (i.e., what men and women are like). For example, women may be excluded as potential leaders because women are assumed to be dependent, weak, and emotional, which are incongruent with the leader role. Prescriptive norms, on the other hand, are beliefs about desirable differences between men and women (i.e., what men and women should be like; Eagly, 1987). For example, women may be relatively included as leaders when they act independently, strongly, and logically (i.e., congruent with the leader role), but they may still experience prejudice because people believe that they ought to be feminine (i.e., incongruent with the leader role). Applied to this study, Palin violates descriptive norms of the leader role, and Clinton violates prescriptive norms of the gender role (Eagly & Karau, 2002). As a result, Palin and Clinton may experience prejudice differentially.
Although Eagly and Karau (2002) argued that several factors moderate prejudice toward female leaders, most research has focused on domains that attenuate or exacerbate role congruity effects. For example, prejudice is more pronounced for female leaders in stereotypically masculine (vs. feminine) domains (Eagly & Carli, 2007). In one study, Spanish participants evaluated a female leader more negatively when she had been working in the automobile (vs. clothing) manufacturing industry (Garcia-Retamero & López-Zafra, 2006). Similarly, U.S. women were less likely to emerge as leaders in mixed gender dyads when the domain was football (vs. wedding planning; Ritter & Yoder, 2004).
Another factor that may moderate role congruity effects is stereotypicality of the female leader. Stereotypicality refers to physical and psychological characteristics (e.g., appearance, mannerisms, interests) of a woman that are more or less typical for her gender. Although Eagly and Karau (2002) did not explicitly link stereotypicality with role congruity, consistent with this notion, they suggested that feminine features (e.g., feminine dress or mannerisms) increase the accessibility of the female gender role. This increased accessibility may cause more stereotypically feminine women to appear particularly deviant from descriptive norms of the leader role and unqualified for leadership. Similarly, features that increase the salience of the leader role (e.g., masculine dress or mannerisms) may cause less stereotypically feminine women to appear particularly deviant from prescriptive norms of the female gender role.
Clinton may experience prejudice because she is less stereotypically feminine; her masculine characteristics are congruent with the leader role but violate prescriptive norms of the female role (e.g., Gutgold, 2007). For example, Clinton has a more masculine leadership and communication style (Carlin & Winfrey, 2009; Rifkind, 2000). Her appearance also is less stereotypically feminine. Palin, on the other hand, may experience prejudice because she is more stereotypically feminine. Her feminine characteristics are congruent with the female gender role but violate descriptive norms of the leader role. For example, the media harped on Palin's more stereotypically feminine appearance, including her beauty queen background and wardrobe expenses. Additionally, Palin's dress has been characterized as more “pretty” than “powerful” (Givhan, 2007). Thus, anecdotal evidence indicates that Palin and Clinton differ relatively in stereotypicality. We tested our assumption that Clinton and Palin indeed vary in stereotypicality, because no empirical findings have supported this suggestion. Specifically, we hypothesized that Clinton would be regarded as less feminine than Palin (Hypothesis 1).
What Might Prejudice Directed Toward Clinton and Palin Look Like?
Role congruity theory suggests that both Clinton and Palin may experience prejudice, or negative evaluations (Allport, 1954), because they violate prescriptive norms regarding gender or descriptive norms, respectively, regarding leadership. Integrating our considerations of role congruity theory and stereotypicality, Clinton and Palin both may be evaluated negatively but for different reasons. To understand the specific ways in which prejudice might be directed toward Clinton and Palin, we examined evaluations of stereotypicality, competence, warmth, and voting likelihood, as well as the moderating role of participant gender, sexism, and political orientation.
Competence and Warmth
The Mixed Model of Stereotype Content (Fiske et al., 2002) suggests that people tend to evaluate others on warmth (i.e., the degree to which they harm or benefit ingroup goals) and competence (i.e., the degree to which they effectively pursue those goals). Research shows that women tend to be regarded as either warm or competent, but not both (Cuddy et al., 2009). Specifically, women who are less stereotypically feminine and violate traditional gender roles (e.g., feminists and career women) tend to be evaluated as competent but cold, whereas women who are more stereotypically feminine and obey traditional gender roles (e.g., homemakers) tend to be evaluated as warm but incompetent. Although both Palin and Clinton violate traditional gender roles to some degree because they are both career women, relative to Clinton, Palin may be more stereotypically feminine. Both politicians were (and are) affected by sexist media coverage (Carlin & Winfrey, 2009). Dominant media frames provide evidence for the assumption that Palin is likely perceived as more stereotypically feminine, which affects perceptions of competence and warmth. Specifically, the media frames for Clinton are less feminine and apply the “iron maiden” stereotype (i.e., competent but cold); the media frames for Palin are more feminine and apply the sex object and mother stereotypes (i.e., warm but incompetent; Carlin & Winfrey, 2009). As a result, we hypothesized that Palin would be regarded as less competent than Clinton (Hypothesis 2a), whereas Clinton would be regarded as less warm than Palin (Hypothesis 2b).
We also examined the likelihood of voting for Clinton and Palin. Voting likelihood is a particularly significant indicator of prejudice because it assesses people's evaluations of women as actual and potential leaders. Although voting likelihood for Clinton and Palin may generally be low, we examined whether participant gender and ambivalent sexism moderated these effects.
Gender stereotypes are most prevalent among people with traditional gender attitudes (Huddy, 1994), and men hold more traditional attitudes than women (Eagly, Diekman, Johannesen-Schmidt, & Koenig, 2004). In addition, men are more concerned about gender role conformity and violations than women (e.g., Hort, Fagot, & Leinback, 1990). For example, men experience more threat than women after gender atypical behavior and are more likely to engage in action to restore masculinity (e.g., precarious manhood; Vandello, Bosson, Cohen, Burnaford, & Weaver, 2008). Combined with the research suggesting that agentic women experience backlash (Rudman, 1998), men may evaluate more stereotypically masculine (vs. feminine) women more negatively because they violate traditional gender roles. Consistently, men were less likely than women to support a female candidate for president, and men reported less support for Clinton when primed with their male identity (Simon & Hoyt, 2008). Thus, we hypothesized that men would be less likely to vote for Clinton than would women (Hypothesis 3). However, no research has examined gender differences in support of Palin, and Palin likely violates traditional gender roles to a lesser degree than Clinton does. As a result, we did not expect voting likelihood for Palin to differ for men and women.
We also examined the role of ambivalent sexism (Glick & Fiske, 1996), which has received strong empirical support across people and cultures (e.g., Glick et al., 2000). Ambivalent sexism theory suggests that sexist beliefs about women involve seemingly conflicting (i.e., positive and negative) but related attitudes toward women. Benevolent sexism is a chivalrous attitude toward women, who are seen as good but weak and in need of protection. Benevolent sexists positively evaluate women—particularly women who obey gender roles (e.g., homemakers). Hostile sexism is the traditional form of sexism, or a negative attitude toward women. Hostile sexists negatively evaluate women—particularly women who violate traditional gender roles (e.g., feminists). Although benevolent sexism is subjectively positive and hostile sexism is subjectively negative, they represent two sides of the same coin because underlying both are traditional attitudes toward women. Additionally, previous research shows that benevolent and hostile sexism are positively and highly correlated (e.g., Glick et al., 2000). Because benevolent and hostile sexism are associated both theoretically and empirically, we made parallel hypotheses for the effects of benevolent sexism and hostile sexism on voting likelihood for Palin and Clinton.
First, we hypothesized that benevolent sexism would be associated positively with voting likelihood for Palin and negatively with voting likelihood for Clinton (Hypothesis 4a). Although women are penalized for succeeding in stereotypically masculine domains, these effects are sometimes mitigated when women are depicted as consistent with gender roles (e.g., nurturing or identified with the parenting role; Dwyer et al., 2009). Negative evaluations of Palin may be mitigated for benevolent sexists because Palin was often depicted as a nurturing mother, which is consistent with the female gender role. As a result, people higher in benevolent sexism should be more likely to vote for Palin, but less likely to vote for Clinton, because Clinton is depicted as inconsistent with the female gender role.
Second, like benevolent sexism, we hypothesized that hostile sexism would be associated positively with voting likelihood for Palin and negatively with voting likelihood for Clinton (Hypothesis 4b). Benevolent sexism is a reward for women who “stay in their place,” but hostile sexism is a punishment for women who venture outside of their prescribed role. Because both Clinton and Palin are pursuing positions in a masculine domain, we would expect them both to be more negatively evaluated than a homemaker, for example. However, we still expected relative differences in the degree to which Clinton and Palin are seen as stereotypically feminine (as noted by Hypothesis 1). Although we would expect both Clinton and Palin to be negatively evaluated by hostile sexists compared with other gender-consistent women, people high in hostile sexism should be less likely to vote for Clinton because she is less stereotypically feminine (if Hypothesis 1 is supported) than Palin. On the other hand, hostile sexists might more positively evaluate Palin relative to Clinton because Palin is more stereotypically feminine.
Variability of voting likelihood
Finally, we explored the variability in voting likelihood explained by participant gender, participant political orientation, competence, warmth, benevolent sexism, and hostile sexism. As previously stated and consistent with past research showing a gender gap in support for Clinton (Simon & Hoyt, 2008), we expected that men (vs. women) would be less likely to vote for Clinton; we made no prediction about how participant gender would influence voting for Palin. We also expected that those who were more politically liberal would be more likely to vote for Clinton whereas those who were more politically conservative would be more likely to vote for Palin; we thus included political orientation in models predicting voting likelihood but also other evaluations of candidates. Given that we expected Clinton to be evaluated as competent but not warm, evaluating Clinton as warm may be associated positively with voting likelihood. Given that we expected Palin to be evaluated as warm but not competent, evaluating Palin as competent may be associated positively with voting likelihood. Finally, we expected that hostile sexism would be associated negatively with voting for Clinton, whereas benevolent sexism would be associated positively with voting for Palin. That is, benevolent sexists should reward Palin for her gender role congruity, whereas hostile sexists should punish Clinton for her role incongruity.
Undergraduates (N = 244; 90 men, 146 women, 8 unspecified) from psychology courses at a U.S. Midwestern university participated for course credit. The sample included primarily European American students (87.1%) with a mean age of 19.48 years (SD = 2.04). Participants completed the study in fall 2008 when Clinton was a senator and Palin was a governor.
Design and Procedure
The study had a 2 (Candidate: Hillary Clinton, Sarah Palin) × 2 (Participant Gender: men, women) between-participants design. After providing informed consent, participants completed a measure of ambivalent sexism, ostensibly as part of another study. Next, participants reported perceptions of stereotypicality, evaluations of competence and warmth, and likelihood of voting for the candidate. Finally, participants reported demographic information.
Participants completed the Ambivalent Sexism Inventory (ASI; Glick & Fiske, 1996), which has subscales for benevolent sexism (e.g., “Women should be cherished and protected by men”) and hostile sexism (e.g., “Feminists are making entirely reasonable demands of men”—reverse coded). Participants rated the degree to which they agreed with each statement on a 5-point scale (1 = disagree strongly, 5 = agree strongly). Mean hostile sexism (α= .84) and benevolent sexism (α= .77) scores were calculated.
Participants were first asked to judge the candidate's femininity and masculinity on a 7-point scale (−3 = very feminine, 3 = very masculine).
Competence and warmth
Participants evaluated candidate competence by rating six traits (competent, confident, capable, efficient, intelligent, skillful) on a 5-point scale (1 = not at all, 5 = extremely). Participants evaluated candidate warmth by rating seven traits (friendly, well-intentioned, trustworthy, warm, good natured, sincere) on a 5-point scale (1 = not at all, 5 = extremely). Mean competence (α= .87) and warmth (α= .87) scores were calculated.
Participants also reported their voting likelihood. Assuming it was possible to vote for their candidate in each position, participants indicated the likelihood of voting for the candidate as president, vice president, and in her current position (e.g., Clinton as Senator; Palin as Governor) on a 5-point scale (1 = very unlikely, 5 = very likely). A mean voting likelihood score (α= .92) was calculated.
Participants next completed two measures of political orientation. First, they reported political orientation on a 7-point scale (1 = extremely liberal, 7 = extremely conservative). They also reported political orientation as a forced-choice item (liberal or conservative).
Participants then reported their gender, race, and age. Finally, participants were thoroughly debriefed.
As Table 1 shows, candidate evaluations of competence and warmth were above the midpoint and positively correlated for both candidates. Hostile sexism was below the midpoint and benevolent sexism was at the midpoint; both were positively correlated for both candidates. Political orientation was at the mid-point. Some of the patterns of correlations differed across candidates. Benevolent sexism was correlated with competence for Palin but not for Clinton. Voting likelihood also was positively correlated with competence and warmth. Political orientation (higher numbers representing more conservative views) was associated positively with competence, warmth, benevolent sexism, hostile sexism, and voting likelihood for Palin and associated negatively with warmth and voting likelihood for Clinton.
Table 1. Mean, Standard Deviations, and Correlations for All Variables
Note. *p < .05. Correlations for Clinton are above and Palin below the diagonal, p < .05.
3. Benevolent sexism
4. Hostile sexism
5. Voting likelihood
6. Political orientation
Candidate Stereotypicality (Hypothesis 1)
Hypothesis 1 predicted that Clinton would be regarded as less stereotypically feminine than Palin. Consistently, a t-test confirmed that Clinton (M =−0.59, SD = 1.35) was perceived as less feminine than Palin (M =−1.17, SD = 1.15), t(1,241) = 3.61, p < .001, d = .46.
Competence and Warmth (Hypothesis 2a, 2b)
Hypothesis 2a predicted that Palin would be regarded as less competent than Clinton, and Hypothesis 2b predicted that Clinton would be regarded as less warm than Palin. To test hypotheses, competence and warmth were submitted to separate 2 (Candidate: Clinton or Palin) × 2 (Participant Gender: men or women) × 2 (Participant Political Orientation: liberal or conservative) between participants Analyses of Variance (ANOVAs).
Consistent with Hypothesis 2a, a significant main effect of candidate on competence emerged, F(1, 227) = 15.41, p < .001, η2p= .06, indicating that Palin (M = 3.19, SD = 0.97) was evaluated as less competent than Clinton (M = 3.69, SD = 0.81). There was also a significant main effect of political orientation, F(1, 227) = 4.29, p < .05, η2p= .02, indicating that conservatives (M = 3.54, SD = 0.91) evaluated the candidates as more competent than liberals evaluated them (M = 3.32, SD = 0.93). No other significant effects emerged for competence, Fs < 1.
Consistent with Hypothesis 2b, a significant main effect of candidate on warmth emerged, F(1, 226) = 8.02, p < .01, η2p= .03, indicating that Clinton (M = 3.06, SD = 0.85) was evaluated as less warm than Palin (M = 3.47, SD = 0.81). There was also an unexpected interaction between candidate and political orientation, F(1, 226) = 7.76, p < .01, indicating that liberals evaluated Palin (M = 3.24, SD = 0.61) and Clinton (M = 3.17, SD = 0.75) as similarly warm, whereas conservatives evaluated Clinton (M = 2.96, SD = 0.91) as less warm than Palin (M = 3.63, SD = 0.88). In other words, the differences between Clinton and Palin on evaluations of warmth were greater for conservatives than for liberals. No other significant effects emerged for warmth, Fs < 1.
Voting Likelihood (Hypothesis 3, 4a, and 4b).
Hypothesis 3 predicted that men would be less likely to vote for Clinton than women. We did not predict gender differences in the likelihood to vote for Palin. To test Hypothesis 3, voting likelihood was submitted to a 2 (Candidate: Clinton or Palin) × 2 (Participant Gender: men or women) × 2 (Participant Political Orientation: liberal or conservative) between participants ANOVA. Consistent with Hypothesis 3, there was a significant interaction between candidate and participant gender on voting likelihood, F(1, 226) = 5.28, p < .05, η2p= .02, indicating that men were less likely to vote for Clinton (M = 2.28, SD = 1.27) than women (M = 3.01, SD = 1.39), F(1, 226) = 8.75, p < .01, whereas men (M = 2.66, SD = 1.37) and women (M = 2.41, SD = 1.27) were equally likely to vote for Palin, F < 1. Two additional significant effects emerged. An interaction between candidate and participant political orientation emerged, F(1, 226) = 57.02, p < .001, η2p= .20, indicating that conservatives were less likely to vote for Clinton (M = 2.09, SD = 1.19) than for Palin (M = 3.07, SD = 1.23), F(1,226) = 24.15, p < .05, whereas liberals were less likely to vote for Palin (M = 1.69, SD = 0.93) than for Clinton (M = 3.49, SD = 1.22), F(1,226) = 33.01, p < .001. Additionally, an unpredicted 3-way interaction between candidate, participant gender, and participant political orientation further qualified these effects, F(1, 226) = 8.77, p < .0, η2p= .04. As Table 2 shows and consistent with Hypothesis 3, both conservative and liberal men were less likely to vote for Clinton than conservative and liberal women, respectively. Also, consistent with the 2-way interaction between candidate and participant political orientation, both men and women were less likely to vote for Clinton if they were conservative (vs. liberal).
Table 2. Voting Likelihood by Candidate, Participant Gender, and Participant Political Orientation
Note. Different superscripts in adjacent cells within candidate are significantly different. Mean (SD).
Hypothesis 4a predicted that benevolent sexism would be associated negatively with voting likelihood for Clinton and associated positively for voting likelihood with Palin. Similarly, Hypothesis 4b predicted that hostile sexism would be associated negatively for Clinton and positively for Palin. To test Hypothesis 4a and 4b, we conducted regression analyses in which participant gender (dummy coded: women = 0, men = 1), candidate (dummy coded: Palin = 0, Clinton = 1), and benevolent or hostile sexism (mean centered) were entered on Step 1, the two-way interactions were entered on Step 2, and the three-way interaction was entered on Step 3 (following Aiken & West, 1991). We controlled for political orientation by including the continuous measure of political orientation first on Step 1 in all of the models. Significant benevolent and hostile sexism effects only emerged on voting likelihood (main effects of and interactions with benevolent sexism on competence and warmth, ps > .07, and main effects of and interactions with hostile sexism on competence and warmth, ps > .16).
Consistent with Hypothesis 4a, the second step of the model was significant for the regression equation including benevolent sexism, F(3, 202) = 2.70, p < .03, ΔR2 = .04, revealing a significant interaction between benevolent sexism and candidate, B =−.78, SE= .27, b=−.24, t(202) =−2.92, p < .004 (see Figure 1 for slopes comprising this interaction). To explore the candidate and benevolent sexism interaction, the simple slopes for benevolent sexism within candidate were probed. For Palin, benevolent sexism was associated positively with voting likelihood, B = .48, SE= .18, b = .20, t(1,117) = 2.65, p = .009. For Clinton, benevolent sexism did not predict voting likelihood, B =−.30, SE= .19, b =−.15, t(1,117) =−1.55, p = .12.
Consistent with Hypothesis 4b for the regression equation including hostile sexism, the second step of the model was significant, F(3, 202) = 2.64, p < .05, ΔR2 = .08, revealing a significant interaction between hostile sexism and candidate, B =−.67, SE= .31, b=−.22, t(202) = 2.17, p < .04. The third step was also significant, F(1, 201) = 3.17, p < .01, ΔR2 = .11, revealing an interaction between hostile sexism, candidate, and participant gender, B = 1.61, SE= .64, b= .36, t(201) = 2.53, p < .02. To explore the interaction, the simple slopes for hostile sexism within candidate were probed for women and men separately. There were no significant effects of candidate and hostile sexism for men. As Figure 2 shows, however, there was a significant Hostile Sexism × Candidate interaction, B =−1.08, SE= .34, b =−.34, t(1,142) =−3.19, p = .002, for women. For Palin, hostile sexism was associated positively with voting likelihood, B = .58, SE= .22, b = .28, t(1,142) = 2.65, p = .009, whereas for Clinton, hostile sexism was associated negatively with voting likelihood, B =−.50, SE= .26, b =−.24, t(1,142) =−1.94, p = .05.
Finally, multiple regression was used to explore the relative variance explained by participant political orientation, participant gender, competence, warmth, benevolent sexism, and hostile sexism for voting likelihood separately for candidates. Using the enter method, a significant model emerged for both Hillary Clinton, F(6, 97) = 17.90, p < .001, R2 = .53, and Sarah Palin, F(6, 99) = 29.83, p < .001, R2 = .64. Table 3 shows the predictors for each model. Political orientation explained a significant amount of variance for Clinton and Palin: Participants who were more liberal were more likely to vote for Clinton, and participants who were more conservative were more likely to vote for Palin. For Clinton, participant gender and warmth were the other significant predictors: Women and participants who rated Clinton as warmer were more likely to vote for Clinton. For Palin, competence, warmth, and benevolent sexism were the other significant predictors: The participants who rated Palin as more competent and warmer, as well as the participants who were higher on benevolent sexism, were more likely to vote for Palin. Also, when political orientation was included in the model, hostile sexism was associated marginally and negatively with voting likelihood for Palin.
Table 3. Predictors of Participants’ Likelihood to Vote for Candidates
*p < .05, **p < .001, + p < .06.
The present study is the first to directly compare prejudice against Clinton and Palin and to consider the role of stereotypicality in prejudice toward female leaders. Consistent with Hypothesis 1, Clinton was perceived as less stereotypically feminine than Palin. Extending role congruity theory (Eagly & Karau, 2002) to evaluations of women who vary in stereotypicality, Palin was evaluated as less competent than Clinton, and Clinton was less warm than Palin, which was consistent with Hypothesis 2. Also, consistent with Hypothesis 3 (i.e., men are more prejudiced toward less stereotypically feminine women than women), men were less likely to vote for Clinton than women, but men and women were equally likely to vote for Palin. This finding is also consistent with system justification theory (Jost & Banaji, 1994), which implies that men would have more positive evaluations of women who support (vs. challenge) the existing power structure.
Beyond the gender effects, benevolent and hostile sexism predicted voting likelihood for Palin and Clinton. Specifically, consistent with Hypothesis 4a, benevolent sexism was associated positively with voting likelihood for Palin. This finding is compatible with ambivalent sexism theory because benevolent sexists have positive evaluations of women—particularly women who obey traditional gender roles (Glick & Fiske, 1996). Also, consistent with Hypothesis 4b, hostile sexism was associated positively with voting for Palin and associated negatively with voting for Clinton for women participants. This finding is consistent with the notion that hostile sexists have negative evaluations of women—particularly women who violate traditional gender roles. Importantly, the pattern of relationships between candidate, sexism, and participant gender held when political orientation was included in the analyses, which suggests that sexism and gender contribute to voting likelihood beyond participant political orientation.
Finally, we found that different variables predicted voting for Palin and Clinton. Those who evaluated Clinton as warm were more likely to vote for her, although the same was true for Palin. Those who evaluated Palin as competent were more likely to vote for her, but the same was not the case for Clinton. Additionally, benevolent sexists were more likely to vote for Palin but not for Clinton. Hostile sexists were marginally less likely to vote for Palin when participant gender was controlled for in the model. However, as noted above, the effects of hostile sexism on voting likelihood for Palin depended on participant gender. Furthermore, although political orientation accounted for the most variance in voting likelihood for both Palin and Clinton, attributions of competence and warmth, ambivalent sexism, and participant gender also accounted for substantial amounts of variance. Taken together, political orientation, participant gender, evaluations of competence and warmth, and sexism explained over half of the variability in voting likelihood for these real-world candidates.
A limitation of this study is that real-world candidates can be seen as problematic given that they vary on a number of variables that cannot be controlled. However, political candidates are idiosyncratic in the real world. Furthermore, the theories we examined using real-world candidates have been tested repeatedly in controlled laboratory and/or vignette studies (e.g., Garcia-Retamero & López-Zafra, 2006; Ritter & Yoder, 2004) but have been less examined in real-world contexts such as elections. In the real world, for example, people have different amounts of information about real candidates. Controlled studies, however, ensure that participants have the same amount of information about mock candidates. Therefore, the use of controlled experiments limits generalizability about real-world behaviors, and there is a trade-off between internal validity through experimental control and external validity. Given the many previous studies examining these effects with experimental control, our study adds to the literature by examining the effects of theoretically important constructs in a more real—if less controlled—context. Realistic, specific scenarios are a better measure of sexism in elections (Falk & Kenski, 2006), and our study was able to account for variance in voting that has both statistical and practical significance.
Another limitation is that we used a college sample, so the average participant in our sample had only been eligible to vote for 1.48 years. A public sample may have yielded different results by including participants who were older and with more voting experience. However, college students have more egalitarian gender attitudes than public samples (Beere, King, Beere, & King, 1984). Using a college sample may actually have produced a stricter test of our hypotheses, and the effects of prejudice on voting may be greater in a public sample.
One final potential limitation is that Clinton and Palin were actually in office and seeking different roles when this data was collected. Clinton was a senator and a potential candidate for president, and Palin was a governor and an actual candidate for vice president. We do not believe that our results were driven by this distinction. It is no longer the case that the vice president serves a primarily supportive role to the president (Dwyer et al., 2009). Both presidents and vice presidents must be perceived as competent because they exert considerable influence over domestic and foreign policy. Further, sexist attitudes influence all levels of political office. For example, people assume that masculine qualities are required for both senatorial and presidential positions (Huddy & Terkildsen, 1993). Finally, our measure of voting included the likelihood of electing the candidate to be president or vice president, or to her current position. The items were highly correlated, revealing that if participants were likely to vote for the candidate for president, they were also likely to support the candidate as vice president or in her current position.
This study suggests several fruitful avenues for future research. First, this research suggests that women's stereotypicality influences role congruity. Specifically, women with more feminine characteristics may violate prescriptive beliefs about the leader role, whereas women with less stereotypically feminine characteristics may violate prescriptive beliefs about the gender role. Stereotypically feminine characteristics include physical appearance features as well as personality and behavioral attributes. For example, Palin has a more feminine appearance and more feminine behaviors than Clinton. Future research could pit appearance features with personality features. It is possible, for example, that feminine appearance features are associated with attractiveness, which may explain the relationship between femininity and positive evaluations.
Second, future research examining the dynamic intersection between stereotypicality and political orientation may also be informative. For example, in the current work, Clinton, a Democrat, was regarded as less stereotypically feminine than Palin, a Republican. Although the effects reported in this article emerged above and beyond the effects of participant political orientation, it remains unclear whether perceptions of female candidates may sometimes be moderated by candidate political orientation. For example, it is possible that conservative (e.g., Republican) female politicians are expected to behave in line with traditional gender roles (e.g., being more stereotypically feminine) because conservative political orientations are associated with more traditional beliefs (e.g., traditional family values). If this is the case, then female conservatives may present themselves as more stereotypically feminine during primary elections to garner the support of their party; however, this presentation style may not garner support from liberals (e.g., Democrats) and other groups in the general election.
Third, men, as well as people high in benevolent and hostile sexism, were particularly prejudiced toward Clinton, who was less stereotypically feminine. In other words, although both Palin and Clinton are women in the stereotypically masculine domain of politics, it appears that Palin may have been somewhat protected from prejudice from those high on sexism because she is more stereotypically feminine. One might speculate that displays of feminine stereotypicality may somewhat buffer female politicians from the double bind of role congruity prejudice. For example, our data are consistent with the notion that the leadership role and the female gender role do not stand in direct opposition to one another. Although the leader role may be generally linked with masculinity and the female role may be linked to femininity, some movement toward the leader role or the female role may not directly correspond to lower perceptions of femininity or leadership, respectively. However, if women display too many feminine or leadership characteristics, perceptions of leadership or femininity may be undermined, respectively. Furthermore, women may be perceived as insincere when strategically presenting feminine or leadership qualities, which may result in prejudice because they are regarded as inauthentic manipulators. For example, Hillary Clinton's tears in New Hampshire during the 2008 presidential election primaries were scrutinized as being disingenuous (e.g., crocodile tears; Elliott, 2008). On the other side of the political spectrum, Michele Bachman, a congresswoman from Minnesota and a 2012 presidential hopeful, was criticized for claiming that works from Ludwig von Mises and Milton Friedman (known for their complex, hard-right economic theories) were her preferred reading at the beach (Leonard, 2011). One is hard pressed to generate parallel examples in which male politicians engage in such strategic behaviors to demonstrate femininity or competence and are scrutinized similarly to female politicians.
Fourth, role congruity theory has primarily focused on the prejudice that female leaders may experience (Eagly & Karau, 2002). It is possible that male leaders may also experience prejudice when they act in ways that are incongruent with the high-status male role. For example, the status incongruity hypothesis suggests that men may experience prejudice when they behave in ways that are incongruent with high status. Demonstrating this effect, modest men were disliked because they violated prescriptive norms suggesting that men should not be weak or dependent and that men should be ambitious and agentic (Moss-Racusin, Phelan, & Rudman, 2010). Even though the male gender role and the leader role are congruent, it is possible that male politicians will experience prejudice when they are inconsistent with the male gender role (i.e., are less stereotypically masculine). Of course, men are not in the same double bind as women, who may find it difficult to strike the balance between being good women and good leaders, but stereotypicality may moderate prejudice toward male politicians in some situations.
Finally, the findings for benevolent sexism, hostile sexism, and voting likelihood also have implications for ambivalent sexism theory. Although benevolent and hostile sexists may generally view women in leadership roles in masculine domains negatively, it appears that this negativity can be eliminated and even reversed for female leaders with stereotypically feminine characteristics. One implication of this finding is that female politicians and leaders in masculine domains more generally may present themselves in feminine ways (e.g., feminine dress and mannerisms) in order to increase favor with both benevolent and hostile sexists. Furthermore, our results suggest that the failure to present oneself in a stereotypically feminine way may actually reduce voting likelihood, particularly for hostile sexists. Future research should further examine the relationship between ambivalent sexism and stereotypically feminine characteristics, particularly for women in leadership roles in stereotypically masculine domains.
This is the first work to compare prejudice directed against Hillary Clinton and Sarah Palin, and it links stereotypicality with differential prejudice against female leaders. If Palin runs for president in the upcoming 2012 elections, as she and others have speculated, this research suggests that she may experience less prejudice from men and hostile sexists as a more stereotypically feminine candidate than a female candidate who has less stereotypically feminine characteristics, such as Clinton. Furthermore, this research suggests that stereotypicality, participant gender, and ambivalent sexism may further explain prejudice toward women in politics, but also toward women in other leadership roles in masculine domains.