Hillary Clinton's 2008 presidential run brought with it an unwelcome reminder of the added obstacles women candidates often encounter. From “Iron my Shirt” banners to debates regarding whether Clinton was overly emotional or an ice princess, we were faced with gender stereotypes that have often plagued women in leadership roles. Her campaign strategy attempted to counteract preconceived notions of female candidates by accentuating Clinton's toughness and strength, even to the point of trying to “outmale” her opponents (e.g., see Lawrence and Rose Reference Lawrence and Rose2010). The result was a prime example of the double bind in which women who appear strong are then seen as unlikeable (Jamieson Reference Jamieson1995). However, Clinton reached new levels of popularity from 2008 to 2012 as secretary of state and a presumed presidential contender in 2016. This leads us to question whether she was able to successfully navigate the double bind—perhaps through additional offsetting information or her ability to competently handle the masculine position of secretary of state—or if public attitudes have evolved to a point where gender no longer matters in the assessment of political candidates.
There is reason to believe that stereotypes have shifted and that women running for lower offices do not face a penalty in the way they did previously (Brooks Reference Brooks2013). However, research continues to show that the gender of the candidate impacts mass political behavior—from the assessment of the candidate's character traits and policy competence to the amount of time it takes voters to make a decision about the candidate they will support (Dolan Reference Dolan2010; Fulton and Ondercin Reference Fulton and Ondercin2013; Huddy and Terkildsen Reference Huddy and Terkildsen1993a; Lawless Reference Lawless2004; Sanbonmatsu Reference Sanbonmatsu2002). Additionally, in the same way women are making gains as middle managers but not as CEOs (Toegel Reference Toegel2011), there is reason to believe that the chief executive position of the country may prompt more gendered assessments of candidate traits than other political positions. Therefore, while Clinton is unique for a variety of reasons, assessing the role of gender attitudes in mass opinions about her specifically is particularly important as she prepares another potential run for president at the time of this writing.
Hillary Clinton has been described as one of the most polarizing figures in American politics. Since the early days of the 1992 presidential campaign, much of that polarization has been ascribed to her feminist persona (Dowd Reference Dowd1992; Jamieson Reference Jamieson1995; Troy Reference Troy2006). Indeed, Hillary Clinton became a lightning rod for antifeminist backlash when she famously proclaimed in March of that election year, “I suppose I could have stayed home and baked cookies and had teas, but what I decided to do was to fulfill my profession which I entered before my husband was in public life.”Footnote 1 Her statement clearly struck a nerve in a nation still divided by proponents of traditional marital roles and supporters of gender equality in the workplace. Or, as Jamieson (Reference Jamieson1995, 223) aptly put it, “Hillary Clinton became a surrogate on whom we projected our attitudes about attributes once thought to be incompatible, that women either exercised their minds or had children but not both.”
It is not too surprising, then, that attitudes about gender roles factored into the public divide over Hillary Clinton before she ran for president. Winter (Reference Winter2000; Reference Winter2008) found her to be a particularly gendered figure in American National Election Studies (ANES) surveys conducted between 1992 and1996 as attitudes about the woman's movement and traditional gender roles significantly influenced assessments of the then first lady. Sulfaro (Reference Sulfaro2007) also shows that opinions about women's roles were significantly linked to Hillary Clinton's thermometer ratings in the 1996 ANES. And Tesler and Sears's (Reference Tesler and Sears2010) exploratory analyses of 2004 ANES data found that modern sexism (Swim et al. Reference Swim, Aikin, Hall and Hunter1995)—a measure discussed in greater detail below—was a strong and independent predictor of Senator Clinton's (un)favorability ratings. Simply put, the sharp polarization in public opinion about Clinton was driven in part by attitudes about gender.
But Hillary Rodham Clinton's tenure as secretary of state seemed to be much less polarizing than her prior decades in the national spotlight. Her popularity steadily increased after conceding the Democratic presidential nomination in June 2008 and continued to rise after being confirmed as secretary of state by a nearly unanimous Senate vote in January 2009.Footnote 2 Secretary Clinton's favorability climbed to even greater heights during her time at the state department; so much so that she finished her term with a favorability rating approaching 70%—an increase of nearly 20 percentage points from the days of her divisive presidential primary contest against Barack Obama and an almost 25-point increase from her record lows in 1996 (Silver 2012). Those public opinion dynamics were perhaps most fittingly captured by the title of a 2012 CBS News story: “Hillary Clinton: From Divisive to (Mostly) Beloved.”Footnote 3
One might assume that Secretary Clinton's unprecedented popularity was driven by deactivating the gender-based opposition she had faced for decades. After all, the last time Hillary Clinton had a favorability rating approaching 60 percent—during the 1998 Monica Lewinsky scandal—gender attitudes did not influence assessments of the then first lady (Sulfaro Reference Sulfaro2007; Winter Reference Winter2000). Yet, gender traditionalists presumably became more supportive of Hillary Clinton amid her husband's sex scandal because she now visibly stood for things they believed in, such as the importance of marriage and family over career ambition (Winter Reference Winter2000). There was not such a compelling reason for gender conservatives to become more favorable to Hillary Clinton after she became secretary of state. Rather, we identify reasons below why Hillary Clinton's unprecedented popularity as secretary of state may have been disproportionately driven by gender egalitarians. In other words, gendered attitudes may have actually become more important as Secretary Clinton grew more popular.
The growing influence of gender egalitarianism on public support for Hillary Clinton was precisely what we found. Drawing on an original panel survey, which reinterviewed a nationally representative sample of 3,000 registered voters in 2011 who had previously been surveyed during the 2008 campaign, we found that both men and women with liberal views about gender roles rated Hillary Clinton much more favorably in 2011 than they had back in 2008. Meanwhile, gender traditionalists evaluated Clinton at similarly low levels in both the 2008 and 2011 panel waves. We also analyzed data from the 2012 ANES and the 2012 Cooperative Congressional Election Study (CCES) and found that mass assessments of Hillary Clinton were shaped more by modern sexism than evaluations of any other well-known political figure (male or female). At the same time, though, those outsized effects of gendered attitudes were driven by Secretary Clinton's unusually strong support from gender egalitarians that scored low on the modern sexism scale. As a result, Hillary Clinton was both one of the most gendered and one of the most popular politicians in American politics when she left the state department.
EMPIRICAL EXPECTATIONS
The American public often interprets new information to maintain consistency with prior predispositions. Building on a long line of psychological research on cognitive consistence, Taber and Lodge's work (Reference Taber and Lodge2006; Lodge and Taber Reference Lodge and Taber2013) on “motivated skepticism” shows individuals tend to selectively process information that is consistent with their prior opinions (i.e., confirmation bias). As a result, partisans may have completely different perceptions of political reality. Democrats (Republicans), for instance, generally think the unemployment rate and federal budget deficit are lower than Republicans (Democrats) do when a member of their own party is in the White House (Bartels Reference Bartels2002; Bullock et al. Reference Bullock, Gerber, Hill and Huber2013; Druckman Reference Druckman2013; Lavine, Johnson, and Steenbergen Reference Lavine, Johnston and Steenbergen2012).
Along with partisan attachments, racial predispositions can also bias information processing. Moskowitz and Stoh's (Reference Moskowitz and Stoh1994) experimental findings, in fact, suggest that voters often “alter reality” in order to render black candidates’ messages consistent with prior expectations and racial beliefs. In keeping with that finding, racial predispositions have become increasingly important in perceptions of political reality during Obama's presidency (Tesler Reference Tesler2016; Tesler and Sears Reference Tesler and Sears2010). Perhaps most importantly for our present purposes, Barack Obama's increased popularity during the 2008 presidential primary was concentrated among racial liberals who were predisposed to interpret the positive press following his victory in the Iowa caucuses more favorably than racial conservatives (Tesler and Sears Reference Tesler and Sears2010).
We suspect that gender attitudes may have similarly influenced how the public reacted to the mostly positive press Hillary Clinton received as secretary of state. While overt aversion to women holding political office is rare, more subtle forms of sexism and the reliance on gendered stereotypes remain a factor in how the public evaluates female political figures. Huddy and Terkildsen (Reference Huddy and Terkildsen1993b), for instance, found that individuals “rate typical ‘male’ political strengths as more important for national office” (520). Lawrence and Rose (2012, 40) similarly conclude, “Just as a man who seems unmanly may not be able to attain the presidency, a female candidate who seems unwomanly may be profoundly disconcerting to a public still influenced by deeply rooted notions of the qualities each gender ‘should’ display.” Moreover, Swim et al. (Reference Swim, Aikin, Hall and Hunter1995) show that “respondents who were high in Modern Sexism were more likely to prefer a male political candidate portrayed by his opponent as insensitive to women's issues” (209). And most importantly for our present purposes, multiple studies similarly showed that attitudes such as gender traditionalism and modern sexism have long factored into public opposition to Hillary Clinton (Sulfaro Reference Sulfaro2007; Tesler and Sears Reference Tesler and Sears2010; Winter Reference Winter2000, Reference Winter2008).
We believe that these attitudes became even more important in public opinion about Hillary Clinton as she became more popular. Gender traditionalists, who think women should return to their traditional roles in society, would naturally have trouble with a woman in such a powerful position as secretary of state, as it violates traditional gender roles. Modern sexists, who believe women are receiving undeserved special favors, would also be more opposed to Secretary Clinton as her presence in American politics may be seen as a result of riding her husband's coattails. As such, gender conservatives could have discounted Secretary Clinton's well-received job performance as secretary of state to maintain consistency with their antifeminist predispositions (disconfirmation bias). Meanwhile gender egalitarians, who were predisposed to accept positive portrayals of Secretary Clinton, would seem most likely to update their support for her in light of this new information (confirmation bias). Such increased support from gender egalitarians from before to after Clinton became secretary of state was all the more likely following the 2008 presidential primary as well. Clinton's strategy to run a masculinized campaign in order to neutralize gender stereotypes (Carroll Reference Carroll2009; Lawrence and Rose Reference Lawrence and Rose2010), partnered with her rivalry with Obama, who was particularly popular among social liberals, was thought to decrease her support among gender egalitarians to unusually low levels in 2008 presidential primary voting (Kinder and Dale-Riddle Reference Kinder and Dale-Riddle2012; Tesler and Sears Reference Tesler and Sears2010).
Thus, we expect that individuals' attitudes about gender affected their evaluation of Hillary Clinton's job performance as secretary of state and that Americans who more readily subscribed to modern sexism and traditional gender roles were more resistant to viewing her in a positive light regardless of her overall increase in public support. If, as suspected, gender traditionalists and gender egalitarians interpreted Hillary Clinton's job performance as secretary of state in line with their gender predispositions, then we should see the overall relationship between mass assessments of Hillary Clinton and gendered attitudes grow stronger over time. Our central hypothesis, then, is that gender egalitarianism became a more important source of support for Hillary Clinton as she became more popular during her tenure as secretary of state.
METHOD
Data
We test our expectations about how gender attitudes influenced public support for Secretary Clinton with a national panel survey that we commissioned in the spring of 2011 and two 2012 national surveys. Our original panel data reinterviewed a nationally representative subsample of 3,000 registered voters who completed the March 2008 wave of the Cooperative Campaign Analysis Project (Jackman and Vavreck Reference Jackman and Vavreck2009). The original 2007–2008 CCAP was conducted by the polling firm YouGov, which used a matching algorithm to produce an internet sample of 20,000 respondents that closely approximated the demographic makeup of the U.S. Census Bureau's American Community Study (Rivers Reference Rivers2006; Vavreck and Rivers Reference Vavreck and Rivers2008). YouGov then invited a subsample of these 20,000 respondents to take a 12-minute reinterview survey in spring 2011 and constructed our final sample of 3,000 panelists to ensure that the 2011 data were representative of the U.S. registered voter population. Because both the 2007–2008 CCAP and the 2011 reinterviews were constructed to be representative of the registered voter population, there were virtually no differences between those two samples in March 2008 measures of such key variables as party identification, ideology, education, and, most importantly, Hillary Clinton ratings.Footnote 4
We augment our original panel data with two 2012 national election surveys, both of which contained measures of modern sexism. That cross-sectional data comes from the 2012 American National Election Study and an original team module of 1,000 respondents that we fielded as part of the 2012 Cooperative Congressional Election Study (Ansolabehere Reference Ansolabehere2012). Unfortunately, the modern sexism items in the 2012 ANES differ from their prior measures in 2004 and 2008, and the CCES's five-category favorability ratings differ from the ANES's 101-category thermometer ratings. As a result, we cannot directly compare modern sexism's influence on Hillary Clinton's thermometer ratings over time.Footnote 5 We can, however, compare the effects of modern sexism on Hillary Clinton in both the 2012 ANES and the 2012 CCES to their effects on mass assessments of more than 20 different political figures to determine if, as expected, gender attitudes had a significantly stronger impact on public opinion about Secretary Clinton than they had on evaluations of other well-known politicians.
Gender Attitudes
Our 2011 CCAP reinterviews included two gender attitude items that tap into the continuum ranging from egalitarian to traditionalist beliefs about gender roles. These two questions are quite similar in content to a two-item ANES gender traditionalism scale used and validated in previous studies, which assessed the impact of gender traditionalism on various political evaluations including mass assessments of Hillary Clinton (|Winter Reference Winter2000, Reference Winter2005, Reference Winter2008). Like that prior research, the first gender item asked explicitly about women's roles in society, with respondents indicating how strongly they agreed or disagreed with the statement, “Women should return to their traditional roles in society.”Footnote 6 The second item asked respondents to rate feminists on a five category measure, which ranged from very favorable to very unfavorable. Feminists, as Winter (Reference Winter2008, 127) noted, are “closely associated with efforts to make gender arrangements more egalitarian, and so people's positive or negative evaluations of them should relate closely to their own beliefs about proper gender arrangements (Huddy, Neely, and LaFay Reference Huddy, Neely and Lafay2000).” We combined these two items into an additive index (Chronbach's alpha = .59) that was recoded to range from 0 (strong gender egalitarian) to 1 (strong gender traditionalist). The average score on this 0–1 scale was .41, with a standard deviation of .25. Gender traditionalism, as measured with questions about women's roles and feminists in the 1994–1996 ANES panel study, was quite stable over time at the individual level.Footnote 7 Moreover, stable predispositions, such as gender attitudes, rarely change in accordance with mass assessments of well-known political figures (Tesler Reference Tesler2015). It is unlikely, then, that our results were not artificially enhanced by using 2011 measures of gender traditionalism.
The 2012 ANES and our 2012 CCES module contained various items that measure modern sexism. Modern sexism, which is modeled after contemporary conceptualizations of racial prejudice such as symbolic racism and racial resentment (Kinder and Sears Reference Kinder and Sears1981; Kinder and Sanders Reference Kinder and Sanders1996; Sears Reference Sears1988), taps into subtle biases against women that often go unnoticed (Swim et al. Reference Swim, Aikin, Hall and Hunter1995; Swim and Cohen Reference Swim and Cohen1997). Like modern racism, the beliefs that underlie modern sexism are thought to include denial of discrimination, antagonism toward demands for equality, and resentment of undeserved special favors (Swim et al. Reference Swim, Aikin, Hall and Hunter1995).
The 2012 ANES included six items explicitly designed to measure these underlying themes of the modern sexist belief system.Footnote 8 Four of the ANES questions asked about gender discrimination. Two of the ANES' modern sexism items measured how common respondents think discrimination against women is; another one asked how much attention the media should pay to discrimination against women; and the forth gauged how often “women who complain about discrimination cause more problems than they solve.” The remaining two items asked if men or women have more opportunity for achievement in the United States and how often women demanding equality are really seeking special favors. Together these six items form a reliable additive index (Chronbach's alpha = .66), which was recoded to range from 0 (most egalitarian) to 1 (most sexist). The average score on this 0–1 modern sexism scale was .44, with a standard deviation of .14.
Our CCES module measured modern sexism by asking respondents how strongly they agreed or disagreed with the following two statements: (1) “Women who complain about harassment cause more problems than they solve,” and (2) “When women demand equality these days, they are actually seeking special favors.” Together these two items form a highly reliable additive index (Chronbach's alpha = .77), with an average score of .33 on this 0–1 modern sexism scale and a standard deviation of .28.
Control Variables
Attitudes about gender, like those captured by our modern sexism and gender traditionalism scales, are significantly correlated with several other important political and social attitudes. In fact, the correlations between these various measures of gender conservatism and party identification, ideology, and racial resentment ranged from r = .30 to r = .50 in our three surveys. Gender traditionalism is not as strongly linked with demographic variables, but was still significantly correlated with male gender (r = .12), low education levels (r = .19), and age (r = .09) in the CCAP. It is important, then, to control for such factors to help parse out the independent effects of gender attitudes on changing support for Hillary Clinton from other relevant considerations. As such, all our analyses include controls for party identification, ideological self-placement, Kinder and Sanders's (Reference Kinder and Sanders1996) four-item racial resentment scale, and standard demographic variables.
GENDER TRADITIONALISM AND HILLARY CLINTON EVALUATIONS (2008–2011)
Our first tests of how gender attitudes influenced public support for Secretary Clinton during her time at the state department analyzes data from the 2008–2011 CCAP panel study. That survey, as described above, reinterviewed a nationally representative sample of 3,000 registered voters in the spring of 2011 who had previously been surveyed in March 2008. As we would expect, Hillary Clinton's popularity increased substantially from 2008 to 2011 among those reinterviewed panelists. Her average rating on the CCAP's five-category favorability question increased by 13% of the favorability scale's range, from a mean score of 40 (on a 0–100 scale) in March 2008 to an average rating of 53 in April/May 2011. Likewise, the percentage of panelists who rated Hillary Clinton either very favorably or somewhat favorably rose from 36% in 2008 to 48% in 2011.Footnote 9 Furthermore, the percentage of respondents who rated her either very or somewhat unfavorably dropped by 16 percentage points from 53% in 2008 to 37% in 2011.
The question then becomes, who was driving this increased public support for Secretary Clinton? Or more specifically for our purposes, how did attitudes about gender impact Hillary Clinton's growing popularity? Figure 1 helps answer that question by displaying Hillary Clinton's 2008 and 2011 favorability ratings as a function of our previously discussed gender traditionalism scale. The first analysis in that figure breaks the nine-category gender traditionalism scale into five categories to ensure that there are enough panelists in each cell to make valid comparisons across the 2008 and 2011 panel waves.Footnote 10 As can be seen, that analysis produces some distinct findings. The first display of Figure 1, for example, shows that panelists who scored lowest on the gender traditionalism scale were especially likely to increase their support for Hillary Clinton from 2008 to 2011. Indeed, these panelists, who most strongly rejected traditional gender roles, rated Hillary Clinton 21 percentage points more favorably in 2011 than they had back in March 2008.
Figure 1. Hillary Clinton favorability ratings, as a function of gender traditionalism, 2008–2011. Predicted probabilities in the upper-right, lower-left, and lower-right panels based on logistic regression coefficients in Table 1. Probabilities in the upper-right panel calculated by setting partisanship, ideology, racial resentment, education, and indicator variables for female, African-American, and Latino to their sample means. Probabilities in two bottom panels calculated by setting partisanship, racial resentment, education, and indicator variables for female, African-American, and Latino to the respective subgroup mean in the analysis. Source: 2008−2011 CCAP reinterviews.
Table 1. (Logistic regression) predictors of rating Hillary Clinton favorably in 2008 and 2011
Note: Dependent variable is coded as a dummy, taking on a value of 1 (rate Clinton very favorably or somewhat favorably) or 0 (all other answers). All explanatory variables (except actual birth year) are coded 0–1, with 1 being the highest or most conservative value. Regression analyses utilize poststratification weights with robust standard errors presented in parentheses. Significance codes: *P < .10; **P < .05; ***P < .01. Source: 2008–2011 CCAP.
The first display of Figure 1 also shows that this surge in public support over time for Hillary Clinton from 2008 to 2011 was much less pronounced as we move along the gender traditionalism scale: CCAP panelists who scored in the second fifth of gender traditionalism rated Clinton 17 percentage points more favorably in 2011 than they did in 2008; panelists who scored at the midpoint of the gender traditionalism scale rated her nine points more favorably in 2011 than in 2008; and perhaps most importantly, panelists who were on the conservative side of the gender traditionalism spectrum rated Hillary Clinton almost exactly the same before and after she became secretary of state. In fact, Figure 1 shows that only 12% and 11% of the strongest gender traditionalists had favorable opinions of Hillary Clinton in the respective 2008 and 2011 CCAP panel waves.
It appears, then, that gender traditionalists did not update their opinions of Hillary Clinton despite her well-received job performance as secretary of state. Those results are to be expected given gender traditionalists' general opposition to women in positions of power. Meanwhile, gender egalitarians were especially likely to rate Clinton more favorably as secretary of state. The upshot was that strong gender egalitarians and strong gender traditionalists were 22 percentage points more divided in their 2011 assessments of Clinton than they had been back in March of 2008. Or more simply put, attitudes about gender became more important in public opinion about Hillary Clinton as she became more popular.
However, this growing importance of gender attitudes in mass assessments of Hillary Clinton may have been spuriously produced by their aforementioned correlations with other omitted variables. That is, attitudes correlated with gender traditionalism like Republican partisanship, conservative ideology, racial resentment, and Fox News consumption may have been the primary reasons why the strongest gender traditionalists remained steadfastly opposed to Hillary Clinton in both 2008 and 2011. Moreover, racial liberals, who also tend to be gender egalitarians, could have grown more supportive of Hillary Clinton from 2008 to 2011 because she went from an adversarial relationship with Barack Obama to his trusted advisor (e.g., Tesler and Sears Reference Tesler and Sears2010). Table 1 tests those alternatives by including partisanship, ideological self-placement, racial resentment, and standard demographic variables along with the gender traditionalism scale in multiple regression models.Footnote 11
The upper-right panel of Figure 1 graphically displays the gender traditionalism coefficients from the first two columns as Table 1. As can be seen, our conclusion that gender egalitarianism became a more important factor in Hillary Clinton's favorability ratings from 2008 to 2011 remains intact after controlling for several other factors. All else being equal, moving from lowest to highest on the gender traditionalism scale decreased Hillary Clinton's favorability rating by 25 percentage points in March of 2008. That statistically significant relationship more than doubled in size to 54 percentage points in 2011—a highly significant difference in overtime effects (p = .004).Footnote 12 Moreover, the upper-right panel of Figure 1 shows that Hillary Clinton's predicted favorability rating increased by 30 percentage points from 2008 to 2011 among CCAP panelists who most strongly rejected traditional gender roles (after controlling for the variables in Table 1). Panelists who scored highest on gender traditionalism, however, rated Hillary Clinton consistently negative from 2008 to 2011 in both the raw/bivariate (Figure 1, top-left panel) and the modeled (Figure 1, top-right panel) data.
To be sure, these increased effects of gender attitudes from 2008 to 2011 did not displace the longstanding impact of partisan and ideological attitudes on mass assessments of Hillary Clinton. The coefficients in Table 1 indicate that the most liberal Democrats were roughly 80 percentage points more likely to rate Hillary Clinton favorably in 2011 than the most conservative Republicans, even after controlling for gender traditionalism and demographic factors. At the same time, though, those large effects of partisanship and ideology on our panelists' evaluations of Hillary Clinton were roughly stable from 2008 to 2011. So, while gender attitudes strongly predicted changes in support for Hillary Clinton from 2008 to 2011, party and ideology did not.
Moving on to columns for women and men in Table 1 indicate that these dynamic effects of gender attitudes on Hillary Clinton's favorability ratings from 2008 to 2011 were nearly identical for both sexes—much the way that prior research found modern sexism to predict important outcomes similarly for women and men (Swim et al. Reference Swim, Aikin, Hall and Hunter1995). Our female panelists consistently rated Hillary Clinton more favorably than their male counterparts.Footnote 13 Yet, both men and women who scored low on gender traditionalism were increasingly likely to rate Hillary Clinton more favorably in 2011 than they had been back in 2008. In contrast, men and women who scored high on the gender traditionalism scale were unaffected by Hillary Clinton's growing popularity as secretary of state. Strong gender traditionalists of both sexes remained consistently opposed to Hillary Clinton in the 2008 and 2011 CCAP panel waves.
While sex did not condition the enhanced effects of gender attitudes on our respondents' evaluations of Hillary Clinton from 2008 to 2011, the bottom panels of Figure 1 show that ideology did. More specifically, the increased impact of gender attitudes on our respondents' evaluations of Hillary Clinton from 2008 to 2011 was most pronounced among political moderates. The bottom-left panel of Figure 1, in fact, shows that the negative impact of gender predispositions on rating Clinton favorably nearly quadrupled in magnitude among political moderates from 2008 to 2011. All else being equal, moving from lowest to highest on the gender traditionalism scale decreased Hillary Clinton's favorability rating among moderates by 16 percentage points in March 2008 and 62 points in 2011—a highly significant difference in overtime effects (p = .001).Footnote 14 Meanwhile, the bottom-right panel of Figure 1 shows that the effects of gender attitudes on liberals' and conservatives' evaluations of Hillary Clinton were much more stable from 2008 to 2011. These enhanced effects among political moderates suggest that Americans who were unburdened by their ideological aversion or affinity toward Hillary Clinton were freer to interpret her job performance as secretary of state in line with their gender predispositions.
All told, the results in Figure 1 indicate that gender attitudes were an important predictor of public support for Hillary Clinton in 2008 and that these attitudes became significantly more important determinants of Secretary Clinton's favorability ratings in the spring of 2011. They also show that gender egalitarians were especially likely to update their opinions about Hillary Clinton from 2008 to 2011, while gender traditionalists were unresponsive to Secretary Clinton's generally well-received job performance as secretary of state. These results, which show that gender egalitarians were the primary drivers of Hillary Clinton's unprecedented popularity, were found in the raw data and persisted after controlling for several relevant factors. This pattern of gendered support also extended to both men and women's evaluations of Hillary Clinton; and it was strongest among political moderates presumably because they were freed from interpreting her job performance through the usual ideological lenses. We can confidently conclude, then, that gendered attitudes became more important as Secretary Clinton grew more popular.
MODERN SEXISM AND HILLARY CLINTON EVALUATIONS IN 2012 ELECTION SURVEYS
In addition to determining how gender traditionalism impacted changing support for Hillary Clinton from 2008 to 2011 in panel data, we can also look at the effects of modern sexism on her thermometer ratings in the 2012 ANES and her favorability ratings in our team module of the 2012 CCES. As mentioned earlier, the wording and response choices of modern sexism and Clinton ratings in those two surveys differ from how they were asked in prior ANES surveys, so we cannot directly compare modern sexism's influence on Hillary Clinton's thermometer ratings over time. We can, however, compare the effects of modern sexism on mass assessments of Hillary Clinton in the ANES and CCES to their effects on public support for several different political figures to determine if gender attitudes had a stronger impact on public opinion about Secretary Clinton than it had on evaluations of other well-known politicians.
Figure 2 presents the results from those analyses. The two panels graphically display the independent effects of modern sexism on the thermometer and favorability ratings of every public figure assessed in the 2012 ANES and our team's 2012 CCES survey. Each dot in the display represents the change in thermometer (ANES) or favorability (CCES) rating associated with moving from lowest to highest on the modern sexism scale with partisanship, ideology, racial resentment, and standard demographic variables held constant. The coefficients presented on the top lines of the two displays, therefore, show that moving from lowest to highest on modern sexism was associated with decreases of about 30% of the rating scales' ranges in both the ANES and CCES surveys—effects that are both substantively large and highly significant.
Figure 2. Impact of modern dexism on evaluations of political figures. Points are OLS coefficients with party identification, ideology, racial resentment, education, black, female, and Latino included in the model. Each point denotes the change in 0–100 ANES thermometer ratings (recoded on a 0–1 scale in the left panel) of five-category CCES favorability ratings (recoded on a 0–1 scale in the right panel) associated with moving from lowest to highest on the modern sexism scale, with party identification, ideology, racial resentment, education, black, female, and Latino held constant. Solid lines represent the 95% confidence intervals around those point estimates. Source: 2012 ANES; 2012 CCES, team module.
A number of other interesting patterns emerge in the displays. First and foremost, modern sexism had a larger negative effect on mass assessments of Hillary Clinton in the ANES and CCES than it had on any other public figure. Moreover, those large negative effects of modern sexism on Hillary Clinton ratings were significantly stronger than their impact on Barack Obama, Democratic House candidates, and the Democratic Party's thermometer ratings in the ANES; they were also significantly stronger than the impact of modern sexism on any other Democrats' favorability ratings in the CCES.Footnote 15 The display further shows that modern sexism had a significant and independent negative effect on every single Democratic figure's ratings, presumably because Democrats have established themselves as the party that is more supportive of women's rights. Modern sexism, as can be seen, generally had nonsignificant effects on evaluations of Republicans. But in a testament to the scale's validity, modern sexism had its strongest positive impact on ratings of Clarence Thomas, who faced very public charges of sexual harassment during his 1991 Supreme Court confirmation hearings.
While the results in Figure 2 suggest that gendered attitudes had a stronger effect on mass assessment of Hillary Clinton than they had on any other public figure in 2012, Hillary Clinton was also more popular than almost any other politician in both the ANES and CCES surveys. Aside from Bill Clinton, who was slightly more popular, Hillary Clinton was rated at least eight points higher than any other Democrat in the CCES. And aside from Michelle Obama, who was rated slightly higher than her Democratic predecessor in the White House, Hillary Clinton was significantly more popular than any other Democrat in the 2012 ANES.
Hillary Clinton was more popular than these other Democrats in large part because of her stronger support from gender egalitarians. Figure 3, which graphically displays the differences between respondents' evaluations of Hillary Clinton and Barack Obama in the 2012 ANES and CCES surveys, helps illustrate this point. Hillary Clinton was about five points more popular than Obama in the ANES and nine points more popular than the president in the 2012 CCES. As can be seen, that enhanced popularity was concentrated among respondents who scored lowest on modern sexism. Figure 3 shows—after controlling for partisanship, ideology, racial resentment, and demographics—that Hillary Clinton was much more popular than Barack Obama among gender egalitarians and slightly less popular than the president among modern sexists in both surveys. This same pattern was found with just about every other Democrat evaluated, with the exception of Michelle Obama who was both very popular and a highly gendered public figure in the 2012 ANES. The result of Hillary Clinton's unusually strong support from gender egalitarians, as shown in Figure 3, was that she was both the most gendered and one of the most popular public figures in our two 2012 surveys. Those results are consistent with our earlier findings showing that gender egalitarians were the primary drivers of Clinton's growing popularity as secretary of state.
Figure 3. Hillary Clinton–Barack Obama ratings, as a function of modern sexism. Predicted values are based on OLS coefficients with party identification, ideology, racial resentment, education, black, female, and Latino included in the model. Gray lines represent the 95% confidence bands around the predicted point estimates. Source: 2012 ANES; 2012 CCES, team module.
DISCUSSION
Hillary Clinton's historic 2008 run for president was a reminder of the difficulties women face in electoral politics. Sexist comments were prevalent in both the media and public sphere, as were biases based on gendered stereotypes (e.g., see Lawrence and Rose Reference Lawrence and Rose2010; Uscinski and Goren Reference Uscinski and Goren2011). While gendered stereotypes persist in many professions, a woman running for president faces additional obstacles as well. For example, the combination of post-9/11 concerns with gender stereotypes regarding men's ability to better handle national security and military crises led Lawless (Reference Lawless2004) to conclude that “levels of willingness to support a qualified woman presidential candidate are lower than they have been for decades” (479). So while gender stereotypes may not be as damaging to certain political offices or campaigns, there is reason to believe that they need to be heeded in a presidential run.
Similarly, Huddy and Terkildsen (Reference Huddy and Terkildsen1993b) found that gendered stereotypes play a larger role in the assessment of candidates for nonlocal political offices in that individuals “rate typical ‘male’ political strengths as more important for national office” (520). The authors caution that their “findings do not suggest that women cannot gain national office. Rather, they imply the existence of a bias against candidates who lack masculine traits. From this perspective, female candidates can win national office if they convince voters that they possess masculine traits and are competent on ‘male’ policy issues” (520). Unfortunately, this has proven difficult as attempts to “act masculine” have often backfired for female candidates. Hillary Clinton's 2008 run provides a prime example (e.g., see Lawrence and Rose Reference Lawrence and Rose2010). When women act counter to stereotypes, instead of being seen as qualified, they can be seen as unlikable. Jamieson (Reference Jamieson1995) termed this catch-22 the femininity/competence double bind. “The evaluated woman has deviated from the female norm of femininity while exceeding or falling short of the masculine norm of competence. She is too strident and abrasive or not aggressive or tough enough. Or, alternatively, she has succumbed to the disabling effects of the feminine stereotype of emotionalism (121).”
Our findings suggest that Hillary Clinton successfully navigated the double bind as secretary of state by tapping into a wealth of support from gender egalitarians. This could mean that future female candidates will garner support by embracing a feminist persona in order to capture the support of gender liberals. While such a strategy would run counter to the conventional wisdom, there is some evidence suggesting that the political landscape has changed. For example, political mobilization has occurred in reaction to the perceived war on women. Groups such as EMILY's list that focus on electing women have experienced record donations and growth in membership which they attribute to sexist comments made by male politicians or against female politicians or candidates (Chozick Reference Chozick2014).Footnote 16 This speaks to an increasing commitment to address gender discrimination and to do so by electing women, not just candidates that are liberal on women's issues.
In addition, there is reason to believe that a greater focus on women's issues and gender equality will increase—not decrease—support for women candidates, at least among gender egalitarians. It should be noted that this is a sobering caveat as it presumes an ongoing inability to garner the support of gender traditionalists who, as we saw above, did not increase their support for Hillary Clinton even as her popularity soared to new levels as secretary of state. Huddy and Carey (Reference Huddy and Carey2009) found that “a concern about gender discrimination boosted Clinton evaluations and accounted for almost a half (45%) of women's higher ratings of Clinton” (91). This was the case even in a campaign that tried to avoid the specter of gender and largely deactivated the influence of gender attitudes on primary voting (Kinder and Dale-Riddle Reference Kinder and Dale-Riddle2012; Tesler and Sears Reference Tesler and Sears2010). If more explicit claims are made regarding issues that gender egalitarians care about, a female candidate may be seen as both a substantive and symbolic representative of these concerns and harness the growing support of gender egalitarians.
However, it is entirely possible that our findings are limited to Hillary Clinton's unique position in American public life during her time in the state department. Indeed, it remains to be seen whether Hillary Clinton's very large drop in favorability from 2012 to 2015 as she reentered the partisan arena as a presidential candidate was accompanied by disproportionate defections among gender egalitarians, or if those with more conservative gender attitudes drove this decline.Footnote 17 We therefore hope our results will lead to more research on how appealing to gender egalitarianism may or may not increase support for female politicians beyond our potentially unrepresentative example of gender egalitarians driving Hillary Clinton's unprecedented popularity as secretary of state. Given that Clinton's unique position included the likelihood that she will run for president in 2016, though, our findings are suggestive of where she might—and might not—find electoral support in another potential bid to become the first female president.
