Do voters hold gendered expectations about politicians that affect female representation in Brazil? Theories about the contents of gendered stereotypes and their application to candidate evaluations suggest that context matters. In a highly personalized political system in which women remain outsiders, what are voters’ expectations about the profile of a female politician, and to what extent do these stereotypical predictions create gendered bias in voters’ preferences? This article uses two survey experiments in Brazil to investigate how context influences the contents of gendered stereotypes of politicians and whether complying with the stereotypical profile affects voters’ evaluations of hypothetical candidates.
Brazil every day seems further from being a political leader in the world, but when it comes to female underrepresentation, it remains a front-runner. Somehow, a country where women account for no more than 15% of elected representatives managed to elect its first (and so far, only) female president, Dilma Rousseff, in 2010. Her gender was a prominent aspect of the framing of Rousseff’s campaign, which aimed, at first, to overcome the harsh personal profile attached to the candidate and to normalize her as a woman. That strategy worked so well that Brazilians called her “Dilmom,” during a honeymoon period of popular approval (Dos Santos and Jalalzai Reference Santos, Pedro and Jalalzai2021).
It did not last, though. The country’s first presidenta was impeached after reelection and six years in office, in the face of a severe economic crisis. Although the popular perception of Rousseff was related to her honesty and strong character at first, she was later seen as incompetent to handle the country’s economy, and the media described her as unreasonable and hysterical. A context in which the vote relies more on politicians’ attributes than on partisanship can explain, at least in part, why voters still reject female candidates, who are expected to be less fit for office than their male counterparts (Valdini Reference Valdini2013).
It is common knowledge that people assume that an individual will display a set of traits or behave in a given way because of their membership in certain social groups. But the contents of these stereotypes are not always the same, nor is their use when judging others automatic. Role congruity theory predicts that people rely on traditional gendered socialization norms and often assume that women do not hold the same attributes they would expect from political leaders.
This perceived incongruence between women and politics might vary, however, across societies and over time, as women play different social roles (Schneider and Bos Reference Schneider and Bos2019). Stereotype activation and application theories, in turn, suggest that voters’ judgments rely on group membership cues when they cannot count on other relevant information, such as partisanship or ideology (Bauer Reference Bauer2015). Both theoretical frameworks indicate that gender stereotypes are contextually bounded.
Although robust work has developed an understanding of the role played by gendered stereotypes in voters’ preferences, the empirical evidence provided so far has been limited to the U.S. context. Stereotypes arise from shared beliefs that people use when making judgments about others, relying on characteristics that are salient for that society and culture. That means the contents of the stereotypes are expected to vary from one context to another—hence it is reckless to assume that gendered stereotypes shape voters’ behavior in other countries the same way they do in the United States.
Voters do not necessarily rely on stereotypes to evaluate a candidate, but they are more inclined to do so when information is scarce. This article investigates the impacts and contents of gendered stereotypes of politicians in Brazil, a highly personalized political system in which female representation has become stuck, as there are specific contexts where women are still mainly considered outsiders in the political realm. Additionally, low levels of partisanship among both the electorate and politicians in Brazil’s electoral system suggest that this might be a context in which voters apply stereotypes when making a judgment (Bohn Reference Bohn2008; LAPOP Lab Reference Lab2018).
This article aims to advance our understanding of the effects of gendered stereotypes on voters’ behavior in different contexts. Brazil emerges as an interesting political setting to explore since it combines highly personalized electoral institutions that create incentives for voters to rely on personal attributes when evaluating candidates with an environment in which women are still outsiders from political power, expected to display attributes that are incongruent with the role of political leader. I employ two survey experiments that take place in Brazil. In the first, I analyze whether there is a difference in the traits associated with politicians according to their sex. In the second study, I estimate the extent to which complying or not complying with gendered stereotypes affects voters’ behavior toward politicians.
The findings from Study 1 suggest that female politicians are associated not only with the expected feminine traits but also with qualities that are often assumed to be masculine, whereas stereotypes of male politicians are not strongly defined. I discuss how such outcomes may be a consequence of the perception of women as outsiders in a highly personalized political context. Such results reinforce the motivation for the current research, considering how it deviates from findings regarding the contents of female stereotypes that came from previous studies based mainly on United States politics.
In Study 2, I manipulate the traits associated with both male and female candidates and compare voters’ perceptions of those who do and do not comply with the gendered stereotypes derived from Study 1. Experimental evidence indicates that voters are biased in favor of women politicians, regardless of their traits, and in favor of candidates described by stereotypically female attributes.
This research contributes to the literature on gender stereotypes in politics both substantively and methodologically. It advances gender and political science scholarship empirically by bringing new evidence of the importance of context for identifying the contents of gendered stereotypes. As for the methodological contribution, it draws on previous approaches to stereotype content measurement, adopting more rigorous criteria for considering certain traits as stereotypical.
Empirical and Theoretical Background
Scholars have found empirical evidence that people hold stereotypes about politicians that do not match what is expected of women, and they have investigated how different contexts may influence the way this perceived incongruence regarding stereotypes impacts female representation. Findings suggest that in political systems with low levels of partisanship and ideological congruence, voters are poorly informed about their options, and thus they are expected to apply stereotypical expectations when judging potential candidates.
Such personalized political contexts, as well as other electoral institutions, may enhance the relevance of the personal attributes of candidates. The current research draws on the combination of these references and the literature explaining the limits to female representation in Brazil. It explores the influence of gender stereotypes on voters’ preferences toward female politicians in the country.
Gendered Stereotypes and Vote Choice
Scholars have been studying the role played by gendered stereotypes in politics for the past 40 years. Stereotypes are a set of culturally shared beliefs about individuals that create biases toward them based on membership in a social group (Bos, Madonia, and Schneider Reference Bos, Madonia and Schneider2018). Stereotypes are often unconscious cognitive shortcuts used in people’s evaluation processes, contributing to expectations of personal attributes and behaviors based solely on the perceived characteristics of the groups that individuals belong to.
Most findings support the important role of stereotypes in affecting evaluations of female politicians, their perceived suitability to handle different policy issues, and voters’ attitudes toward them (Barnes and Beaulieu Reference Barnes and Beaulieu2014; Dolan Reference Dolan2009; Dolan and Lynch Reference Dolan and Lynch2013; Huddy and Terkildsen Reference Huddy and Terkildsen1993; Sanbonmatsu Reference Sanbonmatsu2002; Sanbonmatsu and Dolan Reference Dolan2009; Sapiro Reference Sapiro1981). However, the direction and intensity of the effects of gendered stereotypes on vote choice are still under dispute.
The extensive research on this topic is mainly restricted to the U.S. political context. Considering the limited cultural and institutional settings of empirical studies, one should not assume that the patterns observed there can easily travel to other contexts. In a series of cross-country comparisons in Latin America, Le Foulon and Reyes-Householder (Reference Le Foulon and Reyes-Housholder2021) show different outcomes in voters’ gendered political preferences across countries according to the local level of corruption. While voters punish female politicians less severely for corruption in comparison with men in Uruguay, similar evidence is not found in other countries such as Argentina and Chile, where corruption is more widespread. They also fail to identify the expected bias in favor of men that has been historically supported by evidence collected in U.S.-based scholarship.
In Brazil, Aguilar, Cunow, and Desposato (Reference Aguilar, Cunow and Desposato2015) find an unexpected bias in favor of female candidates in a series of experiments regarding vote choice for federal deputy. The authors investigate whether there is a relationship between the preference for female candidates and exposure to women as mayors. Although the studies may lack adequate external validity for more robust and general arguments about the perception of female politicians, they do not find support for that explanation. Therefore, the mechanism that could explain a gender bias toward women in politics remains unknown. One possible explanation, which this article explores, may be related to the stereotypes about women and corruption, considering both the salience of this issue and the widely disseminated view of female politicians as less corrupt than males (LAPOP Lab Reference Lab2018). Although the aforementioned findings indicate the absence of voter bias against female candidates, as far as this research has acknowledged, the contents of the expected stereotypes of female politicians have not been scrutinized in the Brazilian context. What positive attributes might predispose voters toward female politicians, instead of against them?
From the extensive U.S.-centered literature on gender stereotypes, we learn that stereotypes reflect diverging expectations regarding the traits and beliefs of men and women, derived from their traditional social roles (Bos, Madonia, and Schneider Reference Bos, Madonia and Schneider2018; Dolan and Lynch Reference Dolan and Lynch2016). While women are perceived as more compassionate, honest, and consensus oriented, men are seen as strong leaders, assertive, and self-confident (Huddy and Terkildsen Reference Huddy and Terkildsen1993; Sanbonmatsu Reference Sanbonmatsu2002; Sapiro Reference Sapiro1981). Since masculine traits are more strongly associated with the attributes required for the exercise of power, it is expected that voters will display a biased preference, considering male candidates, in general, better suited for office than their female counterparts (Dolan and Lynch Reference Dolan and Lynch2016; Schneider and Bos Reference Schneider and Bos2014).
Role congruity theory predicts that from consistently observing men and women playing specific social roles, people create associations between each sex and the traits required to adequately perform such roles (Schneider and Bos Reference Schneider and Bos2019). In the case of political leadership, the attributes a woman is expected to display—communal—are, in theory, incompatible with the ones required for office—agentic. In that sense, the advancement of female economic and educational enfranchisement could lead to more interactions between men and women over time, in contexts in which women are more socially equal with men, updating the contents of historically constructed gendered stereotypes.
Even if women are still perceived to be incongruent with political leadership, stereotypes are not necessarily detrimental to female politicians. If the traits traditionally associated with female politicians are incongruent with voters’ views of politicians in general, women should not display the same characteristics as traditional political leaders—qualities and defects included. In that sense, while it may be expected that female politicians do not display the agentic attributes expected from strong traditional leaders, such as determination and objectivity, they are also more likely to show other qualities that are not associated with the average male politician, such as honesty, ethics, and morality (Barnes and Beaulieu Reference Barnes and Beaulieu2014; Funk, Hinojosa, and Piscopo Reference Funk, Hinojosa and Piscopo2021).
On the other hand, some studies suggest that female politicians are not associated with stereotypes of women in general, such as honesty and morality. In that sense, women in politics would not benefit from the somewhat positive effects of gendered perceptions, since they are seen as a group that differs both from traditional male political leaders and from women (Schneider and Bos Reference Schneider and Bos2014). Barnes and Beaulieu’s (Reference Barnes and Beaulieu2019) latest findings suggest that women may not be perceived as more honest, but as risk averse and lacking necessary political links. Recent findings suggest that the same female stereotypes that are usually considered to make candidates unfit for office become an advantage for women when certain issues are salient (Johnson and Williams Reference Johnson and Williams2020).
Stereotypes can harm women candidates because they affect voters’ behavior. Sanbonmatsu (Reference Sanbonmatsu2002) finds empirical evidence of the consequences of gender stereotype reliance on vote choice. According to her proposed “gender schema theory,” individuals’ predispositions to vote for female or male candidates arise from their stereotypical expectations regarding male and female politicians’ traits, beliefs, and competencies. Similar results have been found even when controlling for candidates’ and voters’ partisanship (Dolan Reference Dolan2009; Sanbonmatsu and Dolan Reference Dolan2009). Nonetheless, studies that have tried to expand these experimental findings to actual electoral races have found that the role played by stereotypes is not a determinant of vote choice for women as much as other traditional factors, such as the office’s level, type of electoral race, and incumbency (Dolan and Lynch Reference Dolan and Lynch2013, Reference Dolan and Lynch2016).
From this overview, there is no resolute conclusion regarding the positive or negative effects of gender stereotypes on voters’ attitudes toward women in politics. There are at least two possible explanations for this variability in results. The first is related to measurement, since scholars have adopted different measures for masculine and feminine stereotypes, considering different traits and issues to be the more salient contents in each study (Bos, Madonia, and Schneider Reference Bos, Madonia and Schneider2018; Schneider and Bos Reference Schneider and Bos2014). The second is related to the types of contexts and how they influence both the contents of gender stereotypes and their cognitive use by voters when evaluating candidates. One cannot assume that the contents of such stereotypes can travel from the contexts where experimental evidence has been collected to elsewhere, since stereotypes are so culturally rooted. Although it is expected that stereotypes are widespread across one society, their use in making judgments is not a norm. On the contrary, certain types of contexts enhance the likelihood that individuals rely on stereotypical norms to evaluate others.
The Importance of Context
Bauer (Reference Bauer2015) points out that much of the scholarship on gender stereotypes in politics starts from the misleading assumption that voters’ cognitive processes always rely on such heuristics as cues to make judgments. Context matters to understand what to expect from voters’ behavior in various ways. First, specific cultural and social backgrounds lead to different stereotypes. Second, the context may vary according to the salience of different issues that influence voters’ preferences. Finally, in contexts in which voters cannot rely on policy preferences to inform their vote choices, or those in which they are ill-informed about their ideological or partisan affinity with the potential candidates, voters need to rely on beliefs about the contenders’ attributes, including gender.
Certain types of contexts are expected to influence how voters make judgments because they enhance the reliance on personal attributes. Two theories suggest the importance of context in determining whether voters are more willing to rely on stereotypes to make political choices. A broader take is proposed by the personal vote theory, according to which certain institutional arrangements create the proper incentives for voters to use the personal attributes of candidates as shortcuts. When electoral competition gets more personal, voters rely more on candidates’ time-invariant characteristics, meaning that voters rely less on partisanship and ideological affinity to make a choice (Shugart, Valdini, and Suominen Reference Shugart, Valdini and Suominen2005; Valdini Reference Valdini2013).
Additionally, Bauer’s (Reference Bauer2015) empirical evidence for the activation and application theory suggests that there are conditions under which voters can be expected to rely on gender stereotypes. Bauer gives further empirical evidence that well-informed voters do not rely on simple heuristics, such as gender stereotypes. However, the less voters are aware of other candidate preferences and behavior, the more willing they are to make judgments by relying on cues given by a candidate’s sex.
Context can also influence voters’ gendered preferences depending on the salience of issues that make certain gendered attributes more welcoming in comparison to others. When the main political debate centers on “masculine” issues, such as war, national security, and the economy, the incongruity between women and office is more salient to people’s perception, for example. On the other hand, the type and the level of office can also predict if voters will rely on gender stereotypes to evaluate candidates. When competing for both executive and national-level office, women are often perceived as less congruent with the position, in comparison to those who run for legislative positions and to the local level politics (Schneider and Bos Reference Schneider and Bos2019).
Besides creating or limiting incentives for voters to rely on personal attributes, context is also determinative when it comes to the contents of gendered stereotypes. Drawing on role incongruity theory, one can assume that gendered stereotypes arise from the observed social roles played by men and women, which can widely vary across societies, cultures, and time.
Gender and Politics in Brazil
Considering the relevance of contextual conditions for both the contents of and reliance on gendered stereotypes, there is a need to expand the boundaries of scholarship. In that sense, Brazil appears to be a suitable setting in which to advance investigation of the influence of gendered stereotypes on voters’ preferences, since its electoral institutions favor the conditions for the personal vote. Additionally, the underrepresentation of women in a country that managed to elect a woman to the highest political office creates a cultural and social setting in which the contents of gendered stereotypes might diverge from the ones where scholars usually test their theories.
Historically, public opinion research has shown that although there is a group that consistently rejects gender equality in politics, most Brazilians are not against the presence of women in politics in a straightforward way (Bohn Reference Bohn2008; LAPOP Lab Reference Lab2018). However, observation of electoral outcomes in races for subnational executive offices or even for the National Congress shows that voters still choose male candidates over females, in general. In 2020, women accounted for 16% of mayors in Brazilian municipalities and only one out of 27 state governors. Two years before, the highest share of women in Brazilian history was elected to the lower house of the national legislature. Despite a 50% increase, women occupy only 15% of the seats.
With an electoral system tainted by personalism, large voting districts with an elevated number of candidates, and voters’ lack of attachment to political parties, candidates’ characteristics can play a determinant role on voters’ choices (Shugart, Valdini, and Suominen Reference Shugart, Valdini and Suominen2005; Valdini Reference Valdini2013). In such contexts, this type of heuristic that creates cues for the evaluation of politicians becomes a more relevant factor (Boas Reference Boas2014; Schwindt-Bayer and Reyes-Housholder Reference Schwindt-Bayer and Reyes-Housholder2017).
Although the inconsistent measurement of gendered stereotypes may partly explain the inconclusive findings of voters’ gendered bias, it is also important to acknowledge how context can influence the salience of certain stereotypes, leading to different results in different scenarios. Voters rely on stereotypes more often in settings in which they are poorly informed about the candidates. In a political setting such as the Brazil, where one can hardly rely on partisanship as a cue for vote choice, personal attributes tend to be more influential (Bauer Reference Bauer2015; Bos and Schneider Reference Schneider and Bos2019; Shugart, Valdini, and Suominen Reference Shugart, Valdini and Suominen2005; Valdini Reference Valdini2013).
This research helps fill the remaining gap in the scholarship regarding the contents of gender stereotypes in a political context other than the United States, where a wide range of studies on this realm has been published for over four decades (Sapiro Reference Sapiro1981). Although the U.S. case is the most explored in the women and politics literature, its findings can hardly be treated as universal. Presuming that such findings should be valid elsewhere would be misleading, since both the political institutions regarding female participation and the historical socialization process that resulted in stereotypes of women in politics are deeply context dependent.
Stereotypes of female politicians in the United States are shaped by political factors that are salient to U.S. citizens but irrelevant to Brazilians, such as war, defense, and terrorism. Partisanship also plays a significant role in perceptions of female politicians in the United States, but not in an electoral system with low levels of voter-party attachment, as in the case of Brazil. While both countries are still far from achieving gender parity in politics, one of them managed to elect a woman for president. And Brazilians did so relying on different gendered expectations regarding Dilma Rousseff than those that prevented U.S. voters from endorsing Hillary Clinton.
This article’s main goal is to investigate the contents of stereotypes of female and male politicians in Brazil and to determine whether complying or not complying with these gendered expectations impacts a candidate’s chance of getting elected. Because of the expected effect of role incongruity between women and political leadership, it is expected that voters consider male politicians more suitable for office in comparison to their female counterparts (Ryan et al. Reference Ryan, Haslam, Morgenroth, Rink, Stoker and Peters2016). Hypothesis 1 describes the expectations related to the contents of gendered stereotypes.
H 1 : The attributes that comprise the stereotype of politicians will be the same for male politicians, but different from the ones that are associated with female politicians.
Besides measuring the contents of gendered stereotypes, it is also the purpose of this research to test the influence of gender bias in a context of low information, where respondents are likely to rely on and apply stereotypes to evaluate candidates. Furthermore, it is important to disentangle whether the source of such bias is related to role incongruity between women and politics, or to sex-related prejudice.
If it is plain prejudice that explains voters’ bias against female politicians, voters’ preferences will not be impacted by their awareness of the actual attributes and traits a candidate display. Voters will manifest their rejection against women as candidates in a more straightforward way, which is summarized in Hypothesis 2.
H 2 : Voters will prefer male candidates over female candidates, regardless of the attributes they are associated with.
On the other hand, if role incongruence theory explains voters’ preferences against women, this means voters assume that female candidates do not hold the proper attributes required of political leaders. Hypothesis 3 describes the expected preference for candidates that display the traits that are expected from a male political leader.
H 3 : Voters will prefer candidates who display attributes that are consistent with the stereotypes of male politicians in comparison to those with the stereotypes of female politicians.
Finally, drawing on role congruity theory, an additional hypothesis is tested regarding compliance with the expected gendered attributes of the candidates, based on their sex. The traits associated with political leadership are, in theory, perceived as masculine. In that sense, they are expected to be displayed by the men candidates. Women, in turn, would be subjected to two different types of punishments, either for complying with the female stereotype that is incongruent with the exercise of power or for showing a counter-stereotypical male profile that is inconsistent with what is expected from women. If gender congruence with the typical attributes of political leadership is required by voters, they will prefer candidates who are men and who are consistent with the male politician stereotype, as enunciated in Hypothesis 4.
H 4 : Voters will prefer male candidates who comply with the male politician stereotype over all other candidates.
In the following sections, I present two survey experiments that took place in Brazil. As a first step, I identify the content of politicians’ stereotypes and whether they differ across genders. Study 1 aims to single out which attributes are associated with female and male politicians. In Study 2, I test voters’ reliance on gender stereotypes in evaluating hypothetical candidates. Overall, I test whether the content of gender stereotypes of politicians rely on role incongruity theory in a context that combines the persistence of women as outsiders with a highly personalized electoral system. Additionally, I further investigate voters’ gendered bias, trying to understand whether voters’ preferences are influenced by candidates’ sex and/or by their compliance with their prescribed gendered stereotypes.
Study 1
Scholarship has acknowledged the influence of gender stereotypes on evaluations of politicians’ competence, suitability for office, and performance (Dolan Reference Dolan2009; Huddy and Terkildsen Reference Huddy and Terkildsen1993; Sanbonmatsu Reference Sanbonmatsu2002). However, previous studies do not consistently converge in the stereotype measures adopted. The contents of the gendered expectations regarding personality traits and competence in certain political issues vary from one study to another, which may lead to conflicting findings regarding stereotypes’ impacts on political behavior (Bos, Madonia, and Schneider Reference Bos, Madonia and Schneider2018; Schneider and Bos Reference Schneider and Bos2014). Moreover, stereotypes’ contents are expected to be contextually and culturally bounded, varying from society to society and over time (Fiske et al. Reference Fiske, Amy, Glick and Xu2002).
Studies regarding active stereotypes related to politics are much scarcer in Brazil. Scholars have found that Brazilian voters display an implicit bias toward supporting women candidates, instead of against them, as anticipated (Aguilar, Cunow, and Desposato Reference Aguilar, Cunow and Desposato2015). The contents of the gendered expectations that could explain voters’ positive attitudes toward female politicians, however, have not been identified.
Over the past 10 years, Brazil has elected, reelected, and impeached its first female chief executive and has promoted institutional improvements to the enforcement of the gender quota law. The presence of women remains stuck at a 15% ceiling both at the national legislative and subnational executive levels, despite the growing number of female candidates (TSE 2018). Understanding how women who run for office are perceived by voters in comparison to their male counterparts is a necessary step to advance explanations for the gender gap in Brazil.
The current study was designed to fill that gap, investigating the salient attributes that are attached to voters’ perceptions of female politicians in Brazil. I employ a survey experiment in which participants indicate traits that describe individuals of different categories, one of which is female politicians. The idea is to capture perceived stereotypes latent in the society, rather than the individual perceptions of the respondents.
Method
The current study builds on the research conducted by Schneider and Bos (Reference Schneider and Bos2014), which was based on the experimental paradigm originally proposed by Katz and Braly (Reference Katz and Braly1933) to measure group stereotype content. Participants were asked to report whether each of the adjectives presented to them was descriptive of one given social group.
Sample
Participants were recruited using Facebook Ads between July 22 and August 2, 2020. As an incentive to engagement, it was advertised that all participants who completed the questionnaire could participate in a lottery for a R$100 gift card. Recruiting participants from Facebook is an adequate solution for survey experimental research, even though the convenience sample is not representative of the broader population (Samuels and Zucco Reference Samuels and Zucco2013).
A total of 756 people initiated the survey. I dismissed duplicate responses from those who provided their email addresses to participate in the lottery, as well as the ones from participants who did not agree with the consent form and whose IP addresses were not from Brazil. The final sample consisted of 623 respondents, 51.2% female, 57.1% white, 48% from the southeast region of Brazil, 33.4% with a college degree, and 40% between 46 and 60 years old. The balance of respondents across groups is detailed in Appendix A in the supplementary materials online. The sample is not representative of Brazil’s population, but it is sufficiently well balanced for the survey experiment.
Procedure
After consenting to participate, respondents were asked to provide some demographic information (gender, geographic region, education, income, and religion). Next, they read a brief text describing how stereotypes are a common heuristic used by society to make associations between individuals and the social groups in which they participate. On the same page of the questionnaire, they were asked to evaluate whether people in general in Brazil would consider each of several given attributes as a descriptor of that one social group.
Although it was expected that respondents would not be embarrassed to express their perceptions regarding gender stereotypes (in comparison to racial or ethnic stereotypes, for instance), the survey instructed participants to answer based on how “people in general” would describe each group to minimize the risk of desirability bias. Following Schneider and Bos (Reference Schneider and Bos2014), the text also informed them about how stereotyping and categorizing individuals is a natural cognitive process.
Each participant was randomly assigned to one out of four treatment conditions. The sole difference between each treatment condition was that individuals were asked to consider how “people in general” would describe individuals belonging to one of these relevant social groups: politicians in general (N = 154), men in politics (N = 163), women in politics (N = 158), and women in general (N = 148). It is worthy of notice that the choice for both the names of the traits presented and the descriptions of the treatment categories were impacted by the lack of gender neutrality in the Portuguese language.
The word “politician” is theoretically the gender-neutral form to refer to those involved in politics. However, since the general stereotype of politicians is mostly masculine, a different word choice was made, to emphasize the neutrality of the category regarding gender. Treatment conditions that referred to politicians in general and of each gender were described as “People/Women/Men that run for elections and occupy public offices.” As for the translation of the attributes, the traits were adjusted to the gender inflection of the treatment category.
Participants were then presented with several traits, and for each, they were asked to indicate whether it was associated with the group of reference. Each respondent was presented with 10 items at a time, randomly selected from a long list of possible attributes. The complete list of attributes included 106 items, most of them drawn from Schneider and Bos’s (Reference Schneider and Bos2014) list of 111 traits, translated into Portuguese by the author and validated by two colleague graduate students. Ten of the adjectives translated were dismissed for being considered redundant in Portuguese. To this initial pool of attributes were added another five traits that are considered strong descriptors of agentic and communal leadership stereotypes thought to be enhanced in contexts of crisis (Kulich, Iacoviello, and Lorenzi-Cioldi Reference Kulich, Iacoviello and Lorenzi-Cioldi2018).
After completing their responses for 30 traits, participants were asked whether they wanted to continue or end their participation. As an incentive for continuing the survey, an additional chance to enter the lottery for the gift card was offered. Those who chose to continue evaluated another randomly selected set of 30 attributes, answering whether each of them was associated with the same social group. The choice to present a random selection of the attributes instead of the complete list was made in the expectation that the list extension would compromise respondents’ attention. All traits were presented to at least 50 participants in each condition. On average, each attribute was exhibited to 46.1% of the participants, and each participant rated an average of 48.9 items.
Analysis
The current study was developed as a strategy to identify the contents of gendered stereotypes in politics. The main outcome of interest is the proportion of respondents who associate each attribute with each social group. More specifically, it is important to understand whether there are attributes that represent an exclusive stereotype of one social group and to differentiate them from traits that are shared stereotypes among two or more groups and from the ones that are nonstereotypical traits.
The first step in the analysis is to calculate the proportion of respondents in each treatment condition who associate each trait with the social group. Following the pre-analysis plan, I considered an attribute to be stereotypical of one group if it met two criteria: the first was a 65% cut point of a positive association of the trait with the considered group, a threshold that follows the one adopted in previous studies (Bos and Schneider Reference Schneider and Bos2014). Traits that were not associated with one group by at least 65% of respondents were classified as nonstereotypical of that particular group.
The second applied only to attributes that passed the first, to determine the extent to which a stereotype was exclusive of one group or shared among two groups. An attribute was classified as an exclusive stereotype of one group in comparison to another when it presented a minimum of 0.125,Footnote 1 the difference in the proportions that are statistically significant at the 0.1 level on a two-sided, two-sample test for equality of proportions.
This combination of criteria for exclusive stereotypes is a novel proposition introduced by the present study. Previous research on gendered evaluations of politicians has considered stereotype content to be sufficiently different by relying simply on significant statistical differences, especially the minimum coefficient value for the difference in proportions. Since the results of the current study will be used to describe the stereotypical characteristics of politicians by gender in the following survey experiment, it is particularly important to identify the stereotype content that is less likely to be shared among groups. Traits that are perceived to be descriptive of both men and women in politics at similar levels are not suitable for describing a hypothetical counter-stereotypical candidate of any gender, for instance.
The comparisons of interest here are the ones that help identify how women in politics are perceived by society. From the specialized literature, it is expected that the contents of stereotypes of female politicians are different from those of politicians in general and male politicians, whereas these last two are expected to be convergent.
-
- Women in politics * Men in politics
-
- Women in politics * People in politics
-
- People in politics * Men in politics
The analysis will also explore potential heterogeneous effects on groups stereotype content considering the gender of the respondent.
Results
The main outcome of interest in the current study is the identification of both gendered and gender-neutral stereotypes of politicians. Table 1 presents the attributes that fit the established criteria for exclusive stereotype contents for “women in politics,” “men in politics,” and “politicians in general”; the proportions of positive associations with each group; and the results for the test of difference in proportions comparing two groups at a time.
Prevalence of exclusive stereotypes across three different groups
Note: *p < 0.1, **p < 0.05, ***p < 0.01, two-tailed proportion test of significance. Values in the “prevalence” columns report the share of respondents who answered that the society, in general, would consider each trait as descriptive of men (MP), women (MP), or people (Politicians - PP) that run for elections and occupy public offices. Values in the “differences” columns report the differences in shares between pairs of groups.
Results obtained from Study 1 deviate widely from those described by Schneider and Bos (Reference Schneider and Bos2014). In their research, all observed differences between female and male politicians or generic politicians were for traits that were positively associated with the latter groups. Female politicians were not attached to any stereotypical attribute. While the authors did not find traits descriptive of women in politics that presented significant differences from the other groups, here we observe just the opposite, identifying a total of 25 attributes as the contents of the stereotype of women in politics, and hardly a handful of traits significantly descriptive of both men and people in general involved in politics.
Several traits emerged as exclusive contents of the stereotype of female politicians, presenting statistically significant differences in proportions compared at the same time to both male politicians and politicians in general. Five of them were widely used as stereotypes of both women in general and female leadership in the scholarship (“compassionate,” “gentle,” “sensitive,” “caring,” and “feminine”). Others, such as “inspiring” and “decent,” are indicators for the dimension of integrity on the scale proposed by Funk (Reference Funk1999) to measure the stereotype of politicians. As supported by previous scholars, I also find that adjectives related to women’s superior morality are held in the case of Brazilian female politicians, with women in politics being described as “decent” and “ethical” in comparison to both men and people in politics, and as “moral” in comparison to men. The most unforeseen results of this joint comparison were the exclusive stereotypes of women in politics as “powerful,” “determined,” and “objective,” traits that were previously associated with male politicians and politicians in general (Schneider and Bos Reference Schneider and Bos2014).
Additionally, other traits emerged as exclusive stereotypes of female politicians in the separate pairwise comparison with each of the other groups. When compared to male politicians, women were described as more intuitive, moral, critical, strong, and better at dealing well with different opinions. While women are usually described as moral and intuitive, the description of women as stronger than men is rather unusual.
When compared to politicians in general, women in politics are described as more “charismatic,” “hard-nosed,” “good at solving problems,” “hardworking,” “fussy,” “complaining,” and better at “doing what it takes” and “caring about others.” Although the stereotype content of female politicians combines traits traditionally associated with both men and women, its valence is different among each gender. The typically masculine traits associated with women in politics refer to mostly positive characteristics of a leader (i.e., “good at solving problems”), whereas the feminine is most likely employed in a negative perspective when applied to leadership contexts (i.e., “complaining”).
Such findings are in line with Wylie’s (Reference Wylie2018) ideal types of female politicians who manage to overcome electoral barriers in Brazil. The author describes three different profiles of women candidates: the supermoms, the technocrats, and the warriors or fighters. Each of these profiles has better prospects of achieving electoral success according to the combinations of the degrees of adversity that women face at the district and party levels.
All profiles portrayed by Wylie (Reference Wylie2018) seem to be compatible with the results of the current study. The first group, the supermoms, is characterized by attributes such as “feminine,” meaning that they conform with traditional gender norms. The female fighters are described by Wylie as empowered women who have party support to confront norms and advance on hard issues. Such a profile could be related to the traits that came from the Study 1 findings, such as “powerful,” “determined,” and “strong.” Finally, attributes such as “objective,” “hardworking,” and “practical” can be applied to describe the technocrat ideal type of female politician.
To further investigate these unexpected outcomes, I examined whether stereotyping varied across demographic categories. The statistically significant association of most of the presented traits with female politicians remains even when controlling for respondents’ gender and other relevant demographic characteristics, as depicted in Appendix A. The exceptions were “strong” and “handles well different opinions,” which remained as contents of the exclusive stereotype of female politicians only for male respondents.
Regarding exclusive stereotypes for the other groups of interest, for both men and politicians in general, the results of Study 1 are mostly inconclusive. In comparison to female politicians, the only trait identified as an exclusive stereotype for male politicians was “masculine.” Even though this result is somewhat redundant, it is essential to point out that the adjective was not as firmly attached to politicians in general. The fact that politicians are not stereotypically described as “masculine” might suggest a broader recognition of women’s presence in politics. Challenging that optimistic perception, compared to female politicians, the only trait considered exclusively stereotypical of politicians was “good with numbers.” The trait is consistently described by the literature as a cognitive stereotype of men but was considered as a stereotype here for a “gender-neutral” group of people in general involved in politics.
Study 1 also provides information about traits that are stereotypically associated with more than one group at once, considered to be shared or gender-neutral stereotypes. This category refers to adjectives considered descriptive of more than one group by 65% of respondents that failed to present significant differences in the proportion tests. Table 2 shows the attributes that attended the described criteria, the proportions, and differences in the three groups’ pairwise comparisons.
Prevalence of shared stereotypes across three different groups
Note: All differences presented above are non-significant at 0.1 level. Values in the “prevalence” columns report the share of respondents who answered that the society, in general, would consider each trait as descriptive of men (MP), women (MP), or people (Politicians) that run for elections and occupy public offices. Values in the “differences” columns report the differences in shares between pairs of groups.
In all, 19 traits are shared among the three groups. Only two of these are associated with feminine stereotypes in the scholarship (“sympathetic” and “creative”), whereas four of them are used as measures of male stereotypes (“rugged,” “rational,” “active,” and “competitive”), and the other four are part of the scale proposed by Funk (Reference Funk1999) to measure the stereotype of politicians in general (“in touch with the people,” “knowledgeable,” “commands respect,” and “intelligent”). The remaining attributes considered nongendered stereotypes were not previously associated with any of the groups. Remarkably, almost all contents of the gender-neutral stereotypes of politicians are qualities, except for “self-interested,” that present negative valence.
In general, Study 1 largely deviates from Schneider and Bos’s (Reference Schneider and Bos2014) findings, suggesting that stereotypes of female politicians are more positive and well defined in Brazil than they are in the United States. The common ground for both cases is that women in politics are not associated with traits usually associated with femininity. In Brazil, Study 1’s outcomes confirm that their stereotypes are composed of communal attributes theoretically associated with women’s leadership style (Dolan and Lynch Reference Dolan and Lynch2016; Huddy and Terkildsen Reference Huddy and Terkildsen1993; Kulich et al. Reference Kulich and Ryan2017; Sanbonmatsu and Dolan Reference Dolan2009). However, female politicians’ stereotypes also include qualities that derive from the idea of a good politician (Funk Reference Funk1999), which are often assumed by the literature to be masculine traits, components of male leadership stereotypes.
Therefore, the pattern predicted by H 1 is partially supported by this first survey experiment. The profile of female politicians is composed of several traits that are exclusively associated with that particular group in opposition to others, while there is no consistent stereotype that differentiates male politicians from politicians in general. At the same time, all three groups share most of the agentic traits that were expected to be exclusively associated with male political leaders. Hence, the content of the gendered stereotypes identified does not corroborate the idea of women being incongruent with political leadership. On the contrary, it suggests that voters expect women to display traits that are both agentic and communal.
Such a difference in outcomes of similar studies highlights the importance of being cautious before assuming U.S.-based research findings as universal. The sample used here is much larger, although still not representative of the population. Randomization tests do not suggest any particular bias for either one of the groups. Public opinion data indicates that Brazilians hold a positive perception of female politicians. More than 78% of Brazilians disagree with the idea that men are better politicians, while they perceive men to be more corrupt than women (LAPOP Lab Reference Lab2018).
In line with that, the positive stereotype of female politicians found in Study 1 may be a consequence of the enduring underrepresentation of women in Brazilian politics, which supports the perception that women are outsiders whose image is not as damaged in the public perception as traditional male politicians. In Study 2, I use the contents of gendered stereotypes of politicians to test how the profile depicted by hypothetical candidates in a context of low information affects voters’ gendered preferences.
Study 2
In Study 2, I draw on the previous findings of which traits compose stereotypes of female and male politicians and test how voters’ preferences vary when considering different types of candidates, who comply or do not with their gendered profile, relying on the outcomes from Study 1. The survey experiment manipulates the traits used to describe four hypothetical aspirants running for mayor in a fictitious municipal election in Brazil. Participants were asked to read a news report about a municipality and then to evaluate different prospective mayoral candidates.
The choice of a hypothetical race for an executive position was made based on the findings that female attributes are perceived as unfit for that type of office, in comparison to legislative positions, which do not require as many agentic traits (Schneider and Bos Reference Schneider and Bos2019). Although it would also be expected that women would be more incongruent with higher-level offices, the experiment takes place in a fictitious municipality so as to create a low-information environment in which judgments rely more on stereotype application (Bauer Reference Bauer2015). If higher levels of office were considered, such as a race for president, it would be reasonable to expect that voters had certain real-life politicians in mind.
If voters are biased against women as political leaders because they would not display the attributes required for the job, as role congruity theory predicts, respondents would be expected to prefer candidates associated with the male politicians’ stereotype in comparison to the female one. On the other hand, if it is plain prejudice that prevents voters from choosing women, it is the candidate’s sex that should influence voters’ behavior.
Method
The study consisted of a survey experiment using Qualtrics. Respondents were recruited using Facebook Ads, with an incentive of participating in a lottery for a R$100 gift card. I employed a 2 x 2 factorial design, in which I manipulate within subjects the candidate’s gender (man or woman) and the candidate’s gendered stereotypes (female stereotypes x male stereotypes).
Procedure
Participants who agreed to participate in the anonymous and voluntary research were asked to provide their demographic information (gender, region, education, income, religion, etc.). Next, they were presented with a brief news report regarding a fictitious municipality in Brazil where elections for mayor would be held in the following months. They read a vignette about the political context of the municipality that described different issues faced by the current mayor.
After that, subjects were asked to evaluate four hypothetical candidates for mayor regarding their fitness for office and to cast a vote for one of them. Participants read brief descriptions of the contenders that included their names, their age, and their profession. The prompt also presented four traits associated with each candidate. Each candidate combined two sets of conditions: sex and gendered stereotypes. There were two female and two male candidates—one of each pair described using attributes associated with women, the other with a masculine/gender-neutral profile, drawing on the results from Study 1. The candidates’ descriptions also included additional information on their professions and marital status, following the pattern adopted by the established scholarship (Aguilar, Cunow, and Desposato Reference Aguilar, Cunow and Desposato2015; Batista Pereira Reference Batista Pereira2020). The candidates’ presentation is depicted in Appendix C.
One candidate of each gender was either described using traits descriptive of either female or male politicians’ stereotypes. Relying on the outcomes of Study 1, it was not possible to identify a set of traits that are exclusively associated with male politicians, with the exception of “masculine.” Hence, the attributes used to describe the compliance with the male politicians’ stereotype come from the ones that are shared across the three groups in Study 1. As for the female stereotype, all traits used were exclusively associated with women in politics.
Respondents graded each of the candidates from a 0–10 scale regarding how suitable their profile is for office in the context presented. After evaluating all candidates, participants were also asked whom they would vote for. Lastly, respondents answered two attention checks and were asked whether they wanted to provide their email address in order to participate in the lottery for the gift card. There was also a debriefing warning explaining that the candidates, the municipality, and the newspaper report presented to them were fictitious.
Sample
Recruitment was again done using Facebook Ads, over the period between October 6 and November 6, 2020. The final sample was composed of 929 participants, after dismissing responses that were either nonconsented, had duplicate email addresses, came from foreign IP addresses, or were randomized but not treated.Footnote 2 The final sample consisted of 54.7% female, 63.3% white, 44.8% with a college degree, and 25.1% of them between 51 and 60 years old. Despite being nonrepresentative of the country’s population, the convenience sample is considered adequate to estimate causal effects (Samuels and Zucco Reference Samuels and Zucco2013).
Analysis
Study 2 aimed to estimate voters’ preferences toward male and female candidates who comply with or defy gendered stereotypes in politics. For that purpose, the analysis relies on two dependent variables: respondents’ perceptions of the suitability of each candidate for office and their vote choice. Suitability for office is measured as the grade attributed to each candidate, a continuous variable that ranges from 0 to 10, whereas vote choice is the probability for voting for each candidate. Respondents were required to cast votes for one of the four described candidates.
The independent variables considered are the candidate’s sex, the candidate’s gender stereotypes, and each candidate specifically, which is described as a combination of the candidate’s sex with the profile that complied or did not company with the expected gender stereotype. I include a control variable for the different contextual settings described in the vignette, that shed light on different main political issues faced by the municipality’s current administration.
Results
In general, voters are not biased against women or candidates with a female gender profile. On the contrary, respondents prefer women over men and candidates who display female stereotypes over those with male attributes. Considering both measures analyzed, the evaluation of suitability for office, as well as the probability to vote for one of the candidates presented, the top-ranked contender was the woman who complied with the female stereotype.
I test the statistical differences across groups using an ordinary least squares regression model that includes a control for the context presented to the participants in the vignette. Three different types of settings were used to describe the hypothetical municipality, that shed light on different issues, such as the economy and corruption. Although it might be worth exploring the differences across such contextual framings, the focus here is to test the overall behavior of respondents considering the manipulated profile of candidates, and not its relationship with the most salient issue in the voters’ minds.
The outcomes described in Table 3 have the scenario of political stability as the baseline. Findings suggest that respondents prefer female candidates as well as women displaying feminine traits contradicting the bias that was expected when considering role congruity theory. Similar patterns are observed for the overall sample of respondents when analyzing the differences across the candidates’ sex and gender separately.
Suitability for office and voting probability across candidates
Note: *p<0.1; **p<0.05; ***p<0.01
As for the overall results across all scenarios, respondents considered the female candidates better fit for office than the male contenders. Women were rated with an average 6.59 grade, while men received 6.07. The difference across them were 0.52 (p < .01). As Figure 1 shows, the probability to vote for female candidates was 0.38 higher than to vote for men (p < .01).
Average vote probability and suitability for office between candidates’ sex.
Comparing the gender of the candidates’ attributes, candidates who complied with the female stereotype were also preferred over the ones with the male stereotype. Figure 2 shows that the average grade regarding the suitability for office of candidates with female attributes was 6.54, while the grade for male stereotypical candidates was 6.12. The difference across groups was 0.42, smaller than the difference between men and women, but still statistically significant (p < .01). More than 40% of respondents were also more willing to vote for candidates with female stereotypes.
Average vote probability and suitability for office between candidates’ gendered stereotype
Considering the profile of each candidate regarding their sex and gendered attributes, the overall preference of voters was to vote for the woman who complied with the female stereotype. The stereotypical woman candidate was also the one that received the higher average grade. However, as for the fitness for office, respondents did not perceive statistically significant differences among that contender and both the candidates described with the counter-stereotypical profile. As displayed in Figure 3, the least preferred candidate was the traditional politician, a man who complied with the masculine stereotype in politics.
Average vote probability and suitability for office between candidates’ sex and gendered stereotype interacted.
Discussion
Study 2 shows a survey experiment regarding the mayoral elections in a fictitious municipality in Brazil, where the traits displayed by four hypothetical candidates are manipulated so that respondents have to compare men and women that comply or reject their gender stereotypes. The political contestants were presented in a low-information context that, in line with stereotype activation theory (Bauer Reference Bauer2015), would encourage respondents to rely on the personal attributes of candidates when making judgments about them.
Findings suggest that respondents’ preferences are not biased against women candidates, nor against candidates that show female attributes. There is no evidence to support H 2 , according to which voters would prefer men over women as candidates simply relying on plain sex prejudice. On the contrary, the preference for female candidates is robust.
Also contradicting the expectations based on role incongruity between gendered stereotypes and political leadership, summarized in H 3 and H 4 , respondents’ least preferred candidates are the more traditional and role-congruent men with male stereotypes. Results show an overall preference in favor of candidates with a female stereotype, even though the traits displayed are not shared among politicians of both genders, according to the outcomes of Study 1. If there is a reward for compliance with stereotypical profile, it is in favor of female candidates.
Conclusions
Although results from survey experiments such as the ones presented here should be carefully interpreted because of their lack of external validity, the outcomes corroborate the importance of considering the role of cultural, social, and political contexts when proposing theories to explain the relationship between gender stereotypes and women’s underrepresentation in politics. The case of Brazil is especially interesting since the country managed to elect a female president while it remained a world leader in female exclusion from the political realm. It is a country where voters hold low expectations about their traditional male politicians and are not openly inclined to exclude women from politics (LAPOP Lab Reference Lab2018), but, when it comes down to it, they still do.
In that vein, the first significant contribution of the current research is related to exploring the contextual boundaries of politicians’ gendered trait stereotypes. Most previous studies about reliance on gendered expectations in politics assume its contents as given, despite acknowledging that stereotypes may vary over time and cultural contexts. Results corroborate that the traits that comprise female politicians’ stereotypes in the Brazilian context are mostly different from traditional male politicians, corroborating what scholarship has acknowledged so far.
Nonetheless, the expectations regarding female politicians in the context of the presented experiments differ from previous literature in that they are strongly defined and composed of mainly positive attributes, including both traditionally feminine characteristics related to the moral dimension and qualities consistently associated with the masculine political leadership style (Schneider and Bos Reference Schneider and Bos2014). Meanwhile, male candidates, as well as the generic notion of politicians, are not precisely stereotyped.
Different stories can explain the unforeseen contents of female politicians’ stereotypes that emerged from Study 1 worthy of further investigation. Scholars acknowledge that stereotypes are bound to contexts—both culturally and temporally. Thus, these results should reflect how Brazilians see women in politics or perhaps suggest an update of people’s past views about them. Given the persistent underrepresentation of women in Brazil, that is hardly the case.
Voters may still see women as outsiders and, thus, they do not carry the burden of the general dissatisfaction with traditional politicians. Alternatively, these results can express female politicians’ tokenization, meaning that the few women who get there are so remarkably overqualified that it creates an idealized view of the whole group. Both mechanisms can also be the explanation behind the broad pro-female bias that emerged from Study 2. Nonetheless, it should be acknowledged as a limitation of the current study that it dismissed the relevant role played by intersectionality. Previous studies have shown that stereotypical expectations are bounded by the intersection between gender and other sociodemographic identities—such as race and class (Bos, Madonia, and Schneider Reference Bos, Madonia and Schneider2018). In Brazil, female underrepresentation is even greater for black and indigenous women, for instance.
Study 2 identified a generalized and unexpected pro-female bias that is inconsistent with the observed behavior of voters. Several factors can explain the incongruence in voters’ behavior in the survey experiment and real-life elections. The design did not incorporate relevant political determinants, such as partisanship and incumbency of the candidates. Also, the survey experiment creates an environment of equity between female and male politicians that is rather unlikely to happen in electoral races. There was gender parity in the survey ballot voters were equally exposed to all candidates and informed about them—emulating a resource equivalence that does not occur between male and female contenders. The gender imbalance in campaign funding and party support is determinant for poor female electoral results in Brazil (Wylie Reference Wylie2018). Such puzzling interaction between experimental findings and voters’ actual behavior remains a challenge for future research.
This article’s experimental findings, although limited, add more support for the idea that female politicians are perceived as outsiders from politics and that can be an asset—in a system in which the male-dominated image of politicians is so damaged, voters seem to have a baseline preference for candidates who are incongruent with traditional leaders.
Supplementary Materials
To view supplementary material for this article, please visit http://doi.org/10.1017/S1743923X22000113.
