Introduction

The gender gaps in politics are numerous. From contacting a politician to working for a political party and running for office, women are less likely to become politically involved (Coffé & Bolzendahl, 2010; Fraile & Gómez, 2017a, 2017b; Lawless & Fox, 2010). The root cause is the persistent gap in internal political efficacy: women are less likely to feel that they have the competence and skills needed to participate fully in the political sphere. This foundational gender gap in internal political efficacy carries important implications for political participation (Reichert, 2016) and political ambition (Crowder-Meyer, 2020; Lawless & Fox, 2010). The size of the gender gap in internal political efficacy is substantial, accounting for ten percent of the total variation in efficacy (Fraile & de Miguel, 2022). Multiple explanations have been offered, including differences between men and women in employment, education, the distribution of family care work, and levels of descriptive representation (Gidengil et al., 2008; Wolak, 2018). However, despite significant progress in women’s socioeconomic status and political representation, there is no evidence of a narrowing of the gender gap in internal political efficacy (Carreras, 2018).

Here, we explore socialization and gender roles as possible explanations for this puzzle. Unlike socioeconomic situation, gender roles are constrained by the process of socialization and change slowly. Prior research shows that ideal gender types persist despite the changing roles of women and men in society (Ellemers, 2018; McDermott, 2016). We argue that women’s depressed sense of political self-efficacy can be explained by the incongruence between the feminine traits and the traits evoked by the political sphere.

We draw on original surveys from Spain that enable us to offer one of the first empirical applications of the social role model outside the United States. Analyses from the first survey show a clear incongruence between (a) the characteristics socially attributed to women and that women attribute to themselves and (b) the traits associated with politics and politicians. While the latter are generally associated with competition, the pursuit of power, and assertiveness, women identify with these characteristics only to a limited extent. Due to this discrepancy, and after controlling for variation in levels of education, women are less likely to believe that they possess the qualities necessary to participate in politics.

We provide a second test of the incongruence argument by experimentally manipulating the nature of politicians’ motivations to work in politics. In two fictitious interviews, politicians describe what drives them to work in politics: in one case, self-promotion and power; in a second case, public service and solving people’s problems. Our analyses show, first, that the gender gap in self-efficacy is similar in our control group (where no interview was shown) and the frame in line with masculine traits of competitiveness and assertiveness. This confirms the result from the first study, namely, that politics is generally perceived in masculine terms (see also Schneider et al., 2016). Second, framing political motivation in line with feminine attribute of serving others appears to narrow the gap in internal political efficacy. It does so by simultaneously boosting women’s self-efficacy and dampening men’s self-efficacy. This result is consistent with prior research from the United States on a diverse but related dependent variable (political ambition): when running for office is framed as a public service to promote the common good, it becomes more appealing to women and less so to men (Pate & Fox, 2018).

The study confirms one of the persistent psychological mechanisms by which women systematically exclude themselves from politics. Because the political realm is less compatible with feminine characteristics, many women do not feel capable of taking an active role in politics. More optimistically, results also indicate that women’s relationships with the political realm are malleable to both how women see themselves and how the political realm is portrayed. The article contributes to an incipient line of research on the application of social role theory to the study of gender and politics (e.g., Schneider & Bos, 2019).

Gender Roles and Political Socialization

Through social interaction, children internalize expectations about the abilities, traits, and roles traditionally associated with each gender (Leaper & Farkas, 2015; Lytton & Romney, 1991). Parents help reproduce gendered roles and behaviors through occupational differences and the household’s division of labor. Children also form beliefs about “appropriate” roles for men and women as they observe these gendered patterns inside and outside the home (Parks-Stamm et al., 2021). For example, the prevalence of female teachers in kindergarten and elementary school signals that caregiving is a woman’s role. Evidence from social psychology suggests that children use gender as a heuristic to evaluate themselves and others, select groups of friends, and choose interests (Bian et al., 2017; Liberman et al., 2017).

Gender socialization has clear implications for political socialization. A recent study from the United States illustrates how early gendered political socialization begins (Bos et al., 2022). When asked to draw a politician, forty-seven percent of girls and seventy-five percent of boys draw male political figures by age six. For girls, this tendency becomes more pronounced with age as gendered political socialization becomes entrenched. By age twelve, seventy-five percent of girls draw male political figures. Another study from the United States offers some clues about the sources of these early stereotypes (Lay et al., 2022). Content analysis of a children’s magazine in elementary school classrooms reveals gendered messages: female and male politicians are portrayed with stereotypes consistent with their gender roles. As a result, from early childhood, boys and girls have the often-unconscious impression that politicians and the political sphere are defined in masculine terms (Bos et al., 2022).

In addition to social resources in the classroom, school curricula, electoral campaign messages, and the media all contribute to the impression that politics is a man’s world. School curricula emphasize the contribution of male leaders to social and political achievements and focus on wars and elections; in so doing, they convey that politics is a man’s world characterized by conflict and competition (Lay et al., 2021; López Navajas, 2014). Academic studies in a variety of disciplines portray good leadership as masculine and with stereotypically masculine characteristics, such as assertiveness and strong leadership (Koenig et al., 2011). Gender stereotypes are also communicated during election campaigns; female candidates are particularly likely to be attacked when they challenge stereotypically feminine traits (Cassese & Holman, 2018). Mass media help reinforce the notion that politics is conflictual and competitive by using analogies from war and sports (Carroll & Fox, 2018) and portraying politics as the exercise of power-seeking, conflict, and dominance (Eagly & Karau, 2002). Such gendered political socialization carries implications for political engagement in adulthood (McDermott, 2016), a point we turn to next.

The Gender Gap in Internal Political Efficacy

At the root of gender inequalities in political engagement is the gender gap in internal political efficacy, defined as the perceived ability to understand and participate in political processes (Almond & Verba, 1965). Self-efficacy is a subjective assessment of one’s abilities and competencies to participate in politics and is an important antecedent to voter turnout and participation in a range of political activities (e.g., Conroy & Green, 2020; Gallego & Oberski, 2012). Prior studies show systematic differences between men and women in self-efficacy not only in the United States (Wolak, 2018 and 2020) but also in Canada (Thomas, 2012), Latin America (Carreras, 2018), and Europe (Fraile & de Miguel, 2022). Here, we develop the argument that the association of politics with conflict, competition, and power-seeking does not align with feminine traits, thus making politics less attractive to women and reducing their perceived efficacy to participate.

An important consequence of the process of gender socialization is the attribution of different personality traits to women and men. Women are typically attributed the communal traits of compassion, affection, patience, caring, and cooperation, whereas men are typically attributed the agentic traits of autonomy, competitiveness, self-confidence, and assertiveness (Eagly & Wood, 2012; Gebauer et al., 2013). The differentiated social characteristics attributed to men and women help perpetuate stereotypical gendered behaviors and traits (Eagly & Wood, 2012; Schneider & Bos, 2019). These traits are relevant to how women and men perceive themselves in relation to different domains (Ellemers, 2018). Women are significantly less likely than men to be motivated to engage in fields perceived as masculine, including science and sports (Hoffman, 2014). Stereotypical traits persist across the lifespan because men and women who engage in activities consistent with traditional feminine and masculine traits are socially rewarded, while those who do not are penalized (Heilman & Okimoto, 2007; Schneider & Bos, 2019).

The realm of politics conveys the agentic traits of competition, conflict, and the pursuit of power (Lay et al., 2021). Lawless and Fox (2010) refer to the alignment of politics with agentic traits as the “masculine ethos” of politics. From the perspective of the social role model, feminine stereotypical abilities and aspirations do not match those evoked by the political arena. One pertinent example that illustrates this incongruence is female political leadership. When office-holding women exhibit communal traits, they often face prejudice because they do not conform to standard notions of leadership. However, when they demonstrate agency, female leaders face prejudice because they do not conform to feminine stereotypes (Conroy & Green, 2020; Eagly & Karau, 2002; Okimoto & Brescoll, 2010).

As women develop traits consistent with their communal qualities of caring for others, they show less interest in pursuing activities that evoke agentic attributes, such as sports, science, and politics (Bian et al., 2017). One implication is that women miss opportunities throughout their lives to engage in sociopolitical learning processes that could enhance their sense of self-efficacy (cf. Beaumont, 2011). In our view, this discrepancy and the resulting missed opportunities lead to a vicious circle in which women are less likely to feel competent to understand and participate in political affairs. If the theoretical mechanism described is true, we should be able to provide empirical evidence for the following expectation:

Women are more likely to identify with traits on the communal dimension of the social role model than men, all else being equal (H1).

Building on these expectations, we consider within-group variation in adherence to typical feminine and masculine traits. While women on average are more likely to identify with communal traits and men are more likely to identify with agentic ones, men and women differ in how strongly they identify with each group of traits (Ellemers, 2018; Spence & Buckner, 2000). Within-group variation is consequential for how men and women behave in the political realm. Conroy and Green (2020) show that women who defy gender norms and express masculine traits are more likely to believe they are qualified for office. From this, we derive the expectation that men and women who identify more strongly with agency likely will show higher levels of self-efficacy.

Individuals who identify more strongly with agentic traits are likely to exhibit higher levels of self-efficacy (H2).

Portraying Politics in Masculine Terms

Evidence of the relationship between politics and the characteristics of competition, conflict, and power-seeking is limited to the U.S. context. Our expectations for this relationship in Spain are in line with previous research, but we are able to test them for the first time outside the United States. Following prior research on the United States, we derive our third expectation:

Politics in general and the motivations of politicians, in particular, are more likely to elicit traits consistent with the agentic dimension of the social role model (H3).

We then examine how differences in the traits elicited by the political realm shape internal political efficacy among men and women. Testing for the effects of such variation is granted given the differential portrayal of male and female political candidates in the mass media in line with feminine and masculine stereotyped traits (Lay et al., 2021). Namely, we consider whether portraying politics in a manner consistent with communal traits closes the gender gap by boosting self-efficacy among women and depressing efficacy among men.

Prior research from the United States helps inform our expectations. In an experimental manipulation of recruitment appeals for political office, a community service message was found to exert differentiated effects on political ambition among men and women (Pate & Fox, 2018). When running for office was framed as a public service to promote the common good, women expressed higher levels of political ambition than women in the control group, where no message was shown. In contrast, men expressed significantly lower interest in running for office than men in the control group. In other words, when politics is framed in line with communal traits and in opposition to masculine notions of leadership and the exercise of power, running for office becomes more appealing to women and less so to men.

Consistent with Pate and Fox (2018), we expect that political motivation in line with communal traits will exert opposite effects on the self-efficacy levels of men and women. Because feminine traits are congruent with communal characteristics, framing political motivation as a public service will likely boost women’s self-efficacy. In contrast, because communal traits are incongruent with masculine characteristics, we expect this framing to dampen men’s self-efficacy.

Women are likely to show higher levels of self-efficacy when politicians evoke communal traits (H4a).

Men are likely to show lower levels of self-efficacy when politicians evoke communal traits (H4b).

In short, we argue that the gender gap in internal political efficacy is a product of the distance between the agentic ethos of politics and the feminine and masculine traits along the agentic-communal dimension.

Research Design and Case Study

To test how perceptions of the political realm affect the gender gap in self-efficacy, we conduct two online surveys of a representative sample of the Spanish population. Spain constitutes a postindustrial society that has undergone profound changes related to the mass entry of women into higher education, the labor market, and politics. The Spanish political system has moved steadily toward gender equality in politics since the transition to democracy in the late 1970s. The percentage of women in the national parliament has steadily increased from 5% in 1979 to 44% in the 2019–2023 legislative term (at the time of writing). Increasing gender parity in politics is often attributed to “the changing rules of the game” (e.g., gender quotas in political parties; see Verge, 2020). Recent comparative surveys show that Spaniards have more progressive views on gender equality and are less likely to engage in gender stereotyping than other European citizens (European Commission, 2017). Given its comparatively higher gender parity in politics and more progressive views on gender equality, Spain is a somewhat harder case for experimental manipulation from a comparative perspective.

We rely on an opt-in access panel from the commercial firm Netquest, which compensates all participants economically.Footnote 1 Study 1 was fielded between the 1st and the 10th of June 2020, and Study 2 between the 15th and the 22nd of December 2020, A total of 1,506 individuals (for Study 1) and 1,504 individuals (for Study 2) were recruited from Netquest’s representative Web panel, with quota selection by gender, education, age, and region (50.7% female, aged 18 to 91 years). These quotas ensured that the final sample matched the characteristics of the Spanish population.

Study 2 contains observational data to test the bulk of our hypotheses. We present evidence of women’s and men’s alignment with agentic and communal traits (H1), respondents’ views about the traits evoked by politics (H3), and their implications for self-efficacy (H2). Study 1 embeds an experiment to test H4a-b. Both surveys were fielded during the pandemic year of 2020, and due to this special timing, we conducted several robustness tests for possible bias in our results. We do not find evidence that the timing of the surveys had an impact on the main variables of interest (see note on survey data collection in the Appendix). For example, the mean value of our dependent variable, self-efficacy, and its standard deviation are almost identical in June and December of 2020 (see Tables 2 and 3 in the Appendix for descriptive statistics of all variables).

Gender Traits and Internal Political Efficacy

To measure the level of adherence to feminine and masculine personality traits, we ask respondents to evaluate their own personalities with a list of agentic and communal characteristics using a reduced version of the Bem Sex Role Inventory (BSRI). Following prior studies (Auster & Ohm, 2000), we include six masculine (agentic) attributes that guide the self and one’s own command and that strive for accomplishment: being ambitious, competitive, and self-confident; having a strong personality; and being tolerant of conflict and risk. In turn, the stereotypical feminine (communal) traits are oriented toward others and their comfort: being affectionate, attentive, compassionate, empathic, kind, and warm. The exact wording of the survey item is: “Now think about the following traits that define people’s personalities. To what extent does each trait define your own personality? Please rate each trait on a scale from 0 (I am not like that at all) to 10 (I am totally like that).”

In panel A of Fig. 1, we estimate mean differences in self-placement on each trait by gender (with 95% confidence intervals). Women are more likely to describe themselves as empathetic, attentive, and compassionate (p < 0.05), whereas men are more likely to describe themselves as confident, risk-seeking, ambitious, competitive, and conflictual (p < 0.05). For four qualities – being friendly, warm, affective, and having a strong personality – we do not find statistically significant differences in self-rating by gender.

Fig. 1
figure 1

Source: Study 2. December 2020

Mean differences in A Self-placement, by gender, and B Traits attributed to men and women in society. Mean difference estimates with 95% confidence intervals

Full size image

In addition to self-assessment, we ask respondents to identify men and women in society on the agentic-communal scale. In panel B of Fig. 1, negative values indicate that the trait is more likely to be attributed to women, while positive values indicate that the trait is more likely to be attributed to men. For each characteristic, we find statistically significant differences (p < 0.05) in the expected direction. Communal traits are more often attributed to women, while agentic traits are more often attributed to men. This suggests that despite the mass entry of women into the labor market and higher education, and the increasing representation of women at the top levels of Spanish government, Spaniards continue to attribute feminine and masculine traits in accordance with H1. Moreover, the data suggest considerable within-group variation in how men and women place themselves on the social role model. Such variation is ripe for testing H2 below.

Next, we test the implications of gendered alignment along the agentic-communal scale for the gender gap in internal political efficacy. We measure two key facets of self-efficacy: beliefs in one’s ability to play an active role in the political arena and understanding of the political realm (Niemi et al., 1991). Empirically, we measure these facets by means of agreement with a battery of items typically used to gauge self-efficacy: (i) “I feel capable of taking an active role in a group that deals with political issues.” (ii) “I possess the qualities needed for politics.” And (iii) “I often feel that politics is so complicated that I do not really understand what is going on.” Responses range from “not at all” (coded as 0) to “fully capable” (4), with the last item reversed. In the following analyses, we sum responses to the three items into a continuous additive scale, ranging from 0 to 12, and rescale it from 0 to 11 due to very few observations in the final category.Footnote 2 Figure 2 shows the distribution of scores by gender: while women are overrepresented in the lower range of the scale (from 0 to 5), men are overrepresented in the upper range (from 6 to 11).

Fig. 2
figure 2

Source: Study 2

Histogram of internal political efficacy, by gender. December 2020

Full size image

To test for the association of adherence with feminine and masculine personality traits and self-efficacy (H2), we follow Coffé and Bolzendahl (2021) and McDermott (2016) and create a pair of indices for agentic and communal traits. Both indices are additive and based on respondents’ self-placement on the six agentic and six communal traits displayed in Fig. 1-A.Footnote 3 The agentic traits index ranges from 0 to 18, with a mean of 15.38 (17.15 for men and 13.72 for women, p < 0.001).Footnote 4 The communal traits index has a mean value of 15.77 (16.43 for women and 15.07 for men, p < 0.001).Footnote 5

Table 1 presents unstandardized coefficient estimates from a set of OLS regression models on internal political efficacy.Footnote 6 Model 1 tests for the gender gap in self-efficacy, controlling for relevant sociodemographic differences.Footnote 7 On average, women express significantly lower levels of self-efficacy (-0.89***). Substantively, the gender gap represents 8.1 percent of total variation in self-efficacy. This offers empirical support for the presence of a sizeable gender gap in self-efficacy. Model 1 also shows that education positively predicts self-efficacy; a unit increase in education (the variable ranges from 0 to 4) is associated with an average increase in self-efficacy of 0.38 (p < 0.001). This implies a maximum rise of 1.52 units in self efficacy if we compare an individual with the lowest education level (value 0) to an individual with the highest education level (value 4), or around 14 percentage points of total variation in self-efficacy. Model 1 shows a negligible effect of having children on self-efficacy, thus contradicting previous studies on the negative effects of care work on women’s self-efficacy (Gidengil et al., 2008).

Table 1 Internal political efficacy: OLS estimates
Full size table

Model 2 in Table 1 includes the agentic and communal indices as predictors of self-efficacy. Including the two indices in our model reduces the size of the association between gender and self-efficacy, suggesting that variation in trait adherence explains some of the gender gap in self-efficacy. Adherence to the agentic model is positively associated with feelings of self-efficacy. A unit increase in the agentic index (from 0 to 33) is associated with an average increase in self-efficacy of 0.09*** (see Model 2 in Table 1), all else equal. Thus, when we compare a person with the lowest attachment to agentic traits with a person with the highest attachment, our model predicts a difference of 2.97 units in self-efficacy, which corresponds to 27 percentage points of total variation in self-efficacy. The size of the effect is therefore substantial and larger than that of education. This offers empirical support for the agentic nature of self-efficacy (H2). In contrast, we find no empirical association between the communal index and self-efficacy (see the coefficient corresponding to Communal Index = -0.01 in Model 2).

We contend that women exhibit lower levels of self-efficacy because they perceive politics as a realm that does not match their personal qualities. One implication of this argument is that women who adhere more strongly to agentic qualities also feel more politically efficacious (H2). We test this hypothesis with an interaction term between gender and adherence to agentic traits in Model 3. The estimated coefficient corresponding to the interaction term is statistically significant and positively signed (p < 0.01), suggesting that women who identify more strongly with agentic traits have higher self-efficacy than women who identify weakly with agentic traits. Figure 3 shows predicted self-efficacy based on Model 3 of Table 1 as a function of gender and level of adherence to agentic traits. When men and women place in the mid- to high range on the agentic index, the gender gap in self-efficacy vanishes. Gender differences are no longer different from zero beyond values of 19 on the agentic index. In contrast, the interaction term between communal characteristics and gender is not statistically different from zero (see the coefficient for Women*Communal = -0.01 in Model 4, Table 1).

Fig. 3
figure 3

Source: Study 2. December 2020

Predicted internal political efficacy for women and men by level of adherence to the agentic index. Predicted values based on Model 3 in Table 1. Control variables are held at their mean values

Full size image

Taken together, the results suggest that the documented mismatch between the communal traits with which women identify and the agentic traits ascribed to the political realm is an important part of the story of the gender gap in internal political efficacy. Women who align more strongly with the agentic dimension are just as likely as men to believe that they possess the qualities needed to participate in the political process. Next, we complement this finding by testing the extent to which the relationship between agentic traits and self-efficacy is malleable to reframing politics in congruence with the communal dimension.

Testing for Agentic Perceptions of the Political Realm

Next, we test the extent to which the political realm evokes traits aligned with agentic traits (H3). The sparse literature to date draws on data from the United States and provides evidence consistent with H3: Americans perceive the political realm as congruent with qualities and abilities on the agentic pole of the social role model (Schneider et al., 2016). We extend this empirical test to the Spanish case. We use two survey questions to tap into perceptions about politics with the following wording: (1) “To what extent do you identify each of the following words with the political realm? Struggle for power; reaching agreements; personal promotion; solving citizens’ problems; competition; service to citizens”. (2) “In general, to what extent would you say that politicians aim for the following? Achieving power; personal promotion; service to citizens; getting rich; solving citizens’ problems; contributing to the improvement of the world in which we live”. For both questions, respondents gave answers on a four-point scale ranging from “not at all” to “very much”.

Responses plotted in Fig. 4 suggest that politics and political motivation strongly align with the agentic dimension of the social role model. For the majority of respondents in panel A, politics very much evokes power (75%), competition (60%), and personal promotion (56%), and only a small proportion of respondents associate politics with seeking agreement (33%), solving people’s problems (24%), or serving citizens (24%). The alignment along the agentic pole is even more pronounced when respondents describe politicians’ motivations. In panel B, most respondents associate politicians with the motivation of gaining power (88%), getting recognition (74%), or getting rich (72%), while only a very small percentage identify political motivation with serving citizens, solving people’s problems, or improving citizens’ lives (5%).Footnote 8 Overall results suggest that the “masculine ethos” of politics holds outside the United States, thus offering support for H3.

Fig. 4
figure 4

Source: Study 2. December 2020

Percent who identify each word with A Politics and B Politicians

Full size image

Reframing Politics on the Communal Dimension: a Survey Experiment

We offer a final test of our argument that the incongruence between feminine traits and the masculine attributes evoked by political ambition and leadership accounts for women’s depressed internal political efficacy. We argue that shifting perceptions of what motivates politicians along the agentic-communal dimension could shape how citizens relate to the political realm. Our final hypothesis is that framing political motivation in line with the communal dimension will likely boost women’s self-efficacy and depress men’s self-efficacy, thereby reducing the size of the gender gap (H4a-b).

Our experimental treatment aims to mimic the mass media’s portrayal of political leaders along the agentic-communal dimension through fictitious interviews with politicians (cf. Lay et al., 2021). We vary our interviewees’ personal motivations to pursue a career in politics using either typical communal or typical agentic traits. In the first interview (Treatment 1: Agentic/Power Frame), working in politics is described as a competitive struggle for power in which political opponents often launch personal attacks against the interviewee and in which the interviewee often takes risks. The headline reads, “To make a difference in politics, you must be willing to compete and take risks: this is the only way the best ideas can triumph” (the full text of the fictitious interview is available in the Appendix). This framing fits well with agentic, self-oriented, and power-seeking motivations and presents a realistic picture of what is perceived to drive politicians. Political careers are commonly considered to fulfill self-promotion and seek power (Schneider et al., 2016), while running for office is seen as a competitive and combative endeavor (Kanthak & Woon, 2015).

In the second fictitious interview (Treatment 2: Communal/Service Frame), the interviewee’s political motivation is framed as a public service in line with the communal attribute of “other-oriented”. Here, the interviewee stresses the importance of cooperating with other political actors to solve people’s problems and improve citizens’ lives. The headline reads, “Politics is simply a service we all do to solve people’s problems.”

Participants in our survey were randomly assigned to one of the two treatments, in the format of an interview on the back cover of a typical Spanish daily newspaper (280 words each in the Spanish original),Footnote 9 or to a control group in which no interview was shown. Importantly, no other characteristics of the interviewee, such as age, gender, name, or political party, were revealed. In both interviews, the exact same questions are raised to the interviewees, and the responses are the same in style and length, except for the adjectives describing politics. The appendix contains the English translation of the exact wording (see Appendix Section I).

To test whether the manipulation worked, we asked respondents to associate politics, as portrayed by the politician interviewed, with three pairs of words: conflict versus agreements; power versus public service; and competition versus cooperation. The wording of the questions is as follows: “According to the interview you read, which of the following words in the scales best portrays the political world described by the politician?” The three scales appeared on the screen in random order. Respondents in the power treatment were more likely to associate politics with conflict, power, and competition; differences in mean values between the two groups were substantial (ranging from 2.46 to 3.18 points on an 11-point scale, p < 0.001). These differences suggest that the treatments effectively transmitted two clearly distinct ideas about the political realm. The manipulation check appeared immediately after the battery of three self-efficacy questions to avoid any influence of the response to this manipulation check on our dependent variable.

In addition to the effectiveness of our treatments, we also test the credibility of the interviews. Overall, 61.4 percent of respondents thought the interviewee could be considered a typical politician, while another 23 percent thought the politician was atypical but could identify him or her with a real political figure. Overall, 84 percent of participants thought the interview was credible.

We test further for attention with a factual question about the content of the interview. Overall, 82 percent of respondents gave a correct answer (83 percent of total participants in the communal frame vs. 81 percent of total participants in the agentic frame). The wording of the question was: “Thinking about the content of the newspaper interview you just read, which of the following topics was not mentioned in the interview? (i) day-to-day life of a politician; (ii) corruption in politics; (iii) political vocation; (iv) advice for those who have political ambition.” Corruption was the only topic not mentioned in the interview. The four topics were offered to participants in random order.

Our results suggest that political motivation, along with self-oriented attributes (Treatment 1), is perceived as more realistic. When asked if the interviewee was a typical politician, 71 percent of respondents in the Agentic Frame affirmed so, while only 52 percent of respondents in the Communal Frame did so. This result corroborates evidence from Schneider et al. (2016) that politicians are generally perceived to pursue self-promotion and power. An implication of this finding is that the gender gaps in self-efficacy might be of a similar magnitude in the Agentic Frame and the control group.

To test H4a-b, we regress gender, experimental treatment (Agentic or Communal Frame versus Control), and their interaction on self-efficacy while controlling for relevant sociodemographic characteristics. Equation 1 represents the estimated functional form of the model:

$$Self - efficacy = {\beta _0} + {\beta _1}*Gender + {\beta _2}*Agentic{\rm{ }} + {\beta _3}*Communal + {\beta _4}*Agentic*Gender{\rm{ }} + {\beta _5}*Communal*Gender + {\beta _k}*Control{s_k} + e$$
(1)

We limit the analyses to the sample of respondents who correctly identified the topic of the interview and who took at least forty seconds to read it. Table 4 in the Appendix presents full regression results with controls, while Fig. 5 presents the main results. Our findings provide only relative support to our expectations that framing political motivation along the communal dimension boosts women’s self-efficacy (H4a) and dampens men’s self-efficacy (H4b). Table 4 in the Appendix indicates that neither the main effect of the communal treatment, nor its interaction with gender are indistinguishable from zero. However, they are consistent with the hypothesized decrease in the size of the gender gap in efficacy. The uncertainty in these estimates is likely related to the loss of statistical power we suffer when we split our analyses by three sub-samples (control, agentic and communal treatment). However, when we focus on the magnitude and direction of the interaction by calculating the effect of treatment, an interesting pattern is revealed. Figure 5 suggests that gender differences in self-efficacy are no longer statistically significant for the subsample of respondents who were exposed to the communal frame. While the predicted value of self-efficacy for women exposed to the communal frame (5.11) is slightly greater than those exposed to the agentic frame (4.88), it is exactly the other way around for men: 5.68 for men exposed to the communal frame versus 6.02 for men exposed to the agentic frame. As a result, the small but statistically significant gender differences in efficacy found in the case of both the control and the agentic subsamples are no longer statistically different from zero when political motivation is framed as emphasizing communal traits.

Fig. 5
figure 5

Source: Study 1. June 2020

A Predicted internal political efficacy for men and women by treatment. B Predicted internal political efficacy for men and women by treatment with 83.4% confidence intervals. Estimates are based on Model 2 in Table 4 in the Appendix while holding control variables at their mean values

Full size image

Although we cannot say with absolute certainty that exposure to communal treatment produces a different effect for men and women due to the low statistical power of these tests, the results are suggestive in this regard. Following Knol et al. (2011), we replicate Fig. 5 using a stricter level of 83.4 percent to further test for an overlap in the confidence intervals for men’s and women’s self-efficacy between treatment and control groups. Our results are robust to this more stringent test. See Fig. 5-Panel B.

These results suggest a potential second way to reduce the gender gap in self-efficacy: reframing politics in line with the traits and motivations associated with communality. While observational evidence shows that the gender gap in self-efficacy disappears when women deviate from their typical attachment to communality (H2) and instead associate with the agentic traits attributed to politicians, evidence from the survey experiment suggests that there might be a second way to close the gender gap in internal political efficacy. Namely, reframing political drive in congruence with traits of communality.

Conclusion

In this article, we have focused on the psychological component of the distance women feel from politics. We contribute to an emerging line of research that applies social role theory to the persistence of multiple gender gaps in politics (Schneider & Bos, 2019). More specifically, we document an incongruence between feminine characteristics and the characteristics associated with politics, and we show that this incongruence carries negative implications for how efficacious women feel in politics. Our empirical analyses help identify two pathways to closing the gender gap in self-efficacy. First, the gap vanishes when women express high levels of confidence, assertiveness, and ambition consistent with the masculine ethos of politics. Second, when politics is framed as a public service to fellow citizens consistent with communal traits, women appear to show comparable levels of self-efficacy to men.

Our results have broader implications for the study of gender differences in the political realm. Scholars of political behavior should pay close attention to both gendered socialization and gendered political socialization because they account for (a) a range of different traits for boys and girls and (b) an association of the political realm with masculine traits (Bos et al., 2022). Early childhood socialization lays the foundation for the gender gap in politics. School curricula and missed opportunities to engage in sociopolitical learning in adolescence widen further the divide between young men and women (Beaumont, 2011). Although the early years are critical to the formation of the gender gap (see for instance Fraile & Sanchez-Vitores, 2020), we know relatively little about gendered political socialization, particularly outside the United States (but see Bos et al., 2022; Lay et al., 2021). Our study speaks to the need to expand our knowledge of gendered socialization as it carries important implications for people’s future engagement in politics.

Our findings also carry implications for how we conceptualize internal political efficacy, whether as a crystallized attitude or as one malleable over the life course. Some research holds that self-efficacy develops early in life and is difficult to change after childhood socialization (Gimpel et al., 2003), a view consistent with gendered political socialization and women’s low self-efficacy. Other research suggests several mechanisms of sociopolitical learning are at work and that political efficacy may be malleable in adulthood (Beaumont, 2011). The results of our study are consistent with enduring patterns of gendered socialization, yet they also indicate that internal political efficacy is not set in stone. Namely, the results of our experiment suggest that self-efficacy might not be a fully crystallized attitude but can be malleable to how politics is framed. Beyond internal political efficacy, political attitudes, and by extension, political representation, may be shaped by the way ordinary people perceive the political world. Studies using an experimental research design also find that political attitudes are malleable. For example, Preece (2016) shows that the gender gap in political interest diminishes when women receive positive feedback on their political knowledge.

Finally, and perhaps most importantly, our findings have implications for how we may close the gender gap in politics. One line of intervention should aim to overcome traditional gender stereotypes and minimize gendered perceptions of social traits. This is the main goal of existing initiatives in the U.S., such as candidate training programs for women and youth, which have spread around the world (Piscopo, 2019). Such initiatives provide information, technical skills, and access to networks, but the latent idea of these training programs is to improve women’s and youth’s typical agentic attributes (such as confidence, competitiveness, and assertiveness) and their sense of belonging in the political world.

These programs may have boosted participating women’s self-efficacy, but the gender gap in society endures. One reading of our results is that the gender gap in politics persists because existing policies place responsibility solely on women. As our study makes clear, an alternative way to closing the gender gap is by implementing measures that help change the perception that politics is a man’s world. Citizens need to be confronted with other ideas about the political realm, such as the concern for others or the willingness to solve people’s daily life problems. Here, the social science and humanities curricula have an important role to play as do the mass media more generally.