Hegemonic masculinity predicts 2016 and 2020 voting and candidate evaluations.

This work examined whether the endorsement of the culturally idealized form of masculinity-hegemonic masculinity (HM)-accounted for unique variance in men's and women's support for Donald Trump across seven studies (n = 2,007). Consistent with our theoretical backdrop, in the days (Studies 1 and 2) and months (Studies 3 through 6) following the 2016 American presidential election, women's and men's endorsement of HM predicted voting for and evaluations of Trump, over and above political party affiliation, gender, race, and education. These effects held when controlling for respondents' trust in the government, in contrast to a populist explanation of support for Trump. In addition, as conceptualized, HM was associated with less trust in the government (Study 3), more sexism (Study 4), more racism (Study 5), and more xenophobia (Study 6) but continued to predict unique variance in evaluations of Trump when controlling for each of these factors. Whereas HM predicted evaluations of Trump, across studies, social and prejudiced attitudes predicted evaluations of his democratic challengers: Clinton in 2016 and Biden in 2020. We replicate the findings of Studies 1 through 6 using a nationally representative sample of the United States (Study 7) 50 days prior to the 2020 presidential election. The findings highlight the importance of psychological examinations of masculinity as a cultural ideology to understand how men's and women's endorsement of HM legitimizes patriarchal dominance and reinforces gender, race, and class-based hierarchies via candidate support.

Donald J. Trump’s history-making ascension from nonpolitician to president of the United States has been explained in terms of an array of factors. Support for President Trump has been found to be associated with the antiestablishment, antielitist, and nativist populism of Trump voters (1), as well as voters’ sexism (2–9), racism (10, 11), homophobia, and xenophobia (2). In addition, many of the factors that predict support of President Trump are confounded with group membership. Men (versus women), White people (versus non-White people), those with relatively less (versus more) education, and Republicans (versus Democrats and Independents) are both 1) higher in racism, sexism, nationalism and 2) stronger supporters of President Trump (12). Status threat, or the increase in cultural diversity that threatens the status quo, is a broader factor that predicts support for Trump and may account for many of the aforementioned associations [specifically, those who supported Trump in the 2016 election were those who felt the hierarchy was being upended and those who perceived more discrimination against White than Black people, Christians than Muslims, and men than women (13)].

As social theorists long have noted, the state and state-sponsored institutions reflect the ideology of dominant groups (14), promoting the broad endorsement and acceptance of cultural ideologies that reinforce and maintain the status quo (15–17). Given that men have more physical, social, and economic power than do women (18), masculinity and manhood are valued and normalized, whereas femininity and womanhood are othered and in need of explanation (19). As de Beauvoir noted, the othering of women (or the gender binary) is at the heart of hierarchical systems that oppress women (19); this includes women with various intersecting identities (20), as well as men who belong to marginalized ethnic, economic, religious, and sexual groups (21). As a result, the state institutionalizes a male point of view (22, 23), and masculinity long has been embedded in the political discourse of the United States (24), perpetuated by candidates who strategically symbolize masculine ideals while attempting to emasculate their opponents (25). In fact, since the 1980s, Republicans have defined their party as masculine by femininizing the Democratic Party and running campaigns based on strength and protection (26). Therefore, presidential elections were, up until 2016, conducted to decide which man was best able to protect the United States from various threats—real or constructed (27). In 2016, Trump epitomized aggressive masculine traits and waged masculinity competitions (28, 29) against his fellow Republican opponents via imputations of failed masculinity (26), accentuating Clinton’s gender [e.g., “she’s playing the woman card” (30)], and her implicit threat to the presidency given gender role incongruity (31).

Our theory and research examined whether United States citizens’ endorsement of culturally valued and idealized forms of masculinity account for unique variance in support of Trump, over and above the variance accounted for by the various other factors noted at the outset of the article. To elaborate our theoretical backdrop and elucidate our hypotheses, we discuss the two primary ways in which masculinity has been conceptualized: masculinity as a precarious social identity and masculinity as hegemony. Within our consideration of each conceptualization, we discuss the relation of masculinity to status, power, and threat.

Precarious Masculine Identity

Psychologically speaking, masculinity typically and most frequently has been conceptualized as a precarious social identity (32–34). From this perspective, masculinity is earned and maintained through continual behavioral displays of manhood. As a result, momentary lapses in behavioral displays of masculinity have the potential to threaten masculinity, with important intra- and interpersonal consequences. Consistent with this notion, threats to masculinity have been produced experimentally by leading men to believe that they are like women in actions (e.g., braiding hair; 35), knowledge (36), personality (37), and/or cognitive performance (33, 38). Beyond documenting the causes of situational and chronic threats to masculinity (39), the precarious masculinity literature examines the consequences of threats to masculinity. For instance, findings show that the threat of being like women (versus men) inspires anger (40) and concerns about how one looks in the eyes of others (35, 36, 38) as well as compensatory dominance that reestablish one’s status as a good man. These compensatory acts include physical aggression (35), sexual dominance (38), intimate partner violence (41), the sexualization and harassment of women (34, 38), and interpersonal violence against lesbian, gay, bisexual, transgender, and queer people (42). Importantly, most studies have found that gender threats inspire compensatory acts of dominance in men but not women (43), consistent with the notion that masculinity is a valued and precarious social identity for men in a way that femininity is not for women (33).

Although research on precarious masculine identity (PMI) has primarily examined content and relevance of masculine self-concepts (33, 35) in intra- and interpersonal contexts, threats to masculine social identities may have political implications. For instance, threatening men by leading them to believe that they were more like women (versus men) resulted in the greater justification of social inequality (37), less support of gender equity (44), and more expression of sexism (38) and homophobia (45). Situational threats to masculinity have also been linked to support of aggressive policies—like support for the Iraq war (45) and gun enthusiasm (46).

Regardless of the potential political consequences (47), PMI cannot fully account for the political rise of Trump. As noted, women do not exhibit parallel threat responses upon receiving feedback that they are gender atypical. As a result, PMI findings can explain the political candidate support and voting patterns of some men but not women (13).

Theoretically and empirically, our work suggests that hegemonic masculinity (HM), more so than PMI, predicts support of Trump. In fact, a point often overlooked by psychological scholars is that conceptualizing masculinity in hegemonic terms implies, as discussed below, that culturally exalted forms of masculinity may be accepted and endorsed by most people—men, women, and gender nonbinary people—which has broader implications and offers a more parsimonious explanation of the linkages between masculinity, political thought, and candidate support.

HM

HM (48) refers to the form of masculinity that, within cultures, is exalted above all others. In the United States, idealized conceptualizations of masculinity prescribe that men should be high in 1) power, or the potential to control outcomes and influence others in psychologically meaningful ways (49–51), and/or 2) status, or the ability to elicit admiration given one’s accomplishments and social standing (52). Ideally, men should also be mentally, physically, and emotionally tough (52, 53), able to persist, unaffected, in the face of physical pain and/or emotional challenges. Given the gender binary and stereotypic prescriptions of men and women as opposites (54), to achieve this masculine ideal, one must repudiate and distance from all that is feminine, gay, or otherwise unmanly (53, 55, 56).

Whereas PMI is a social identity of relevance to individual men and the people with whom they interact, HM is a cultural ideology that is separable from male bodies (57). HM is an ideology that links success and power to men (not women) but is endorsed and accepted as personally beneficial by most members of a given culture—men and women (48). As a result, HM justifies and legitimizes the power of dominant men (i.e., White, straight, upwardly mobile, and able-bodied men) over women and marginalized men (i.e., non-White, gay, disabled, and poor men). Endorsement of HM elevates masculinity and male dominance by othering femininity and womanhood (19) and reinforcing the gender binary (54). Likewise, the endorsement of HM legitimizes and justifies notions of dominant group supremacy, which reinforces and maintains the othering and marginalization of racial minority, nonstraight, physically disabled, religious minority, elderly, and immigrant men. Therefore, HM is related to but distinct from sexism and prejudice, allowing processes of hegemony to operate (15) by veiling sexism and prejudice and subtly contributing to the reinforcement of male dominance and dominant group supremacy.

Interestingly, the psychological implications of HM have received little empirical attention. Theoretically, however, processes of hegemony have been hypothesized to be powerful tools in the reinforcement and maintenance of the status quo (15) given their influence on political thought and support of state-sponsored institutions (14, 23).

HM and Support for President Trump

If HM is a cultural ideology, then its endorsement should be broad, consensual, and predictive of both men’s and women’s support for the status quo. In his 2016 presidential campaign, Trump embodied HM while waxing nostalgic for a racially homogenous past that maintained an unequal gender order. Trump performed HM by repeatedly referencing his status as a successful businessman (“blue-collar businessman”) and alluding to how tough he would be as president (26). Further contributing to his enactment of HM, Trump was openly hostile toward gender-atypical women, sexualized gender-typical women, and attacked the masculinity of male peers and opponents. For instance, Trump consistently referred to the Obama administration as “weak,” proclaiming himself to be the masculine protector who could successfully restore America from its feminized state (26, 58). As a result, HM should predict support for Trump, operationalized as positive evaluations of and voting for Trump, over and above the factors that, as previously described, have been found to be associated with support of Trump (Studies 1 through 7).

Our theoretical conceptualization also positions populism as an effect of HM. As noted at the outset, Trump’s 2016 victory was initially explained in terms of a populist explanation: Trump was an antiestablishment, political outsider, who resonated with voters who felt that the government was no longer representing their interests (1). By contrast, as masculinity scholars have suggested, populists presumably have nostalgia for a racially homogenous, male-dominant past in which women and minorities were often blamed for taking jobs (that “should” belong to White men) and breaking apart the family (59). If populism is an effect of HM, then HM should predict voting for and evaluations of Trump over and above trust in government, as well as political party affiliation, gender, race, and education (Studies 3 and 7).

In sum, HM is a cultural ideology that legitimizes and reinforces idealized notions of masculinity and manhood as dominant over femininity and women and dominant over marginalized and subordinated masculinities (48). HM is related to but distinct from sexism and prejudice toward marginalized groups, allowing processes of hegemony (15) to subtly reinforce male dominance and dominant group supremacy (48). From this perspective, HM is tied to negative attitudes and prejudice toward women and marginalized groups perceived as competing for resources to which dominant men have historically had unencumbered access (59). Consistent with this notion, masculinity has been found to be associated with sexism (60), racism (61), xenophobia (62), homophobia (63), and racial and religious outgroups, more generally (59). However, HM should predict support for Trump over and above sexism (Studies 4 and 7) and prejudiced attitudes (Study 5 through 7), as well as other demographic variables.

Results

In seven studies (n = 2,007), undergraduate or nonstudent men and women reported their past or intended voting, candidate evaluations, and demographic information, including gender, race, level of education, and political party affiliation. One’s own level of education was the predictor used in the three samples involving nonstudent men and women, (i.e., two Mechanical Turk [MTurk] samples and a nationally representative Prolific sample). However, because there was no variance in one’s own level of education in the undergraduate student samples, parental education was used as the predictor in the four student samples. Participants in each study indicated their endorsement of HM. Consistent with other research (64), we operationalized HM using the Male Role Norms Scale (MRN, 53). We chose the MRN as our measure of HM because MRN measures dominant and normative sociocultural masculine ideologies that can be endorsed and accepted by all, regardless of gender identification (53; see also 65–67), rather than descriptive traits of individual men that can be embodied and experienced by some (68, 69). We operationalized PMI using the Masculine Gender Role Stress Scale (69); it measures one’s stress as the thought of failures to embody masculine characteristics. The above-mentioned variables were measured in Study 1 and Study 2.

We estimated two series of hierarchical regressions. The first set of regressions examined whether HM accounted for unique variance in support for Trump, over and above political party, gender, race, and education. In these analyses, political party affiliation was entered in Step 1, the demographic variables of gender, race, and education were entered in Step 2, and the masculinity variables—HM and PMI—were then entered in Step 3. Finally, the interactions that involved the moderation of HM and/or PMI were then entered in Step 4. In the second set of hierarchical regressions, a step was added to the model to examine whether HM predicted support for Trump over and above populism, sexism, and prejudice toward marginalized groups, as well as demographic variables. In these analyses, after political party affiliation and the demographic variables (i.e., gender, race, and education) were entered in Steps 1 and 2, respectively, social attitudes (i.e., trust in government, sexist attitudes, or prejudiced attitudes toward marginalized groups) were entered in Step 3. The masculinity variables were then entered in Step 4 and the interactions were entered in Step 5. Across analyses, we found no evidence of significant interactions that qualified the information presented below. Therefore, although we present significant interactions in the tables, interactions are not discussed in the main text; instead, interactions are presented and discussed in the .

Consistent with our conceptualization, across variables and studies, HM predicted support for Trump, over and above the contributions of political party, gender, race, and education. In fact, the endorsement of HM predicted both voting for and evaluations of Trump in the days (Studies 1 and 2, Table 1) and months (Studies 3 through 6; see ) following the 2016 election. Fifty days prior to the 2020 presidential election (Study 7, Table 1, right panel), in a nationally representative sample, HM also accounted for unique variance in intended 2020 voting and positive evaluations of Trump. Specifically, inclusion of the masculinity variables in Step 3 was associated with a consistent increase in ΔR2 (Table 1 and ), which was significant in six studies and marginally significant in the remaining study. Across studies, HM, but not PMI, predicted support for Trump in Step 3, and these effects were not qualified by any interactions in Step 4.

Table 1.

Results of hierarchical regression analyses for voting for (binary logistic) and evaluations of Trump (linear) for Study 1 (MTurk sample), Study 2 (undergraduate sample), and Study 3 (nationally representative sample)

	Study 1 (days after 2016 election)		Study 2 (week after 2016 election)		Study 7 (50 days before 2020 election)
Independent variables	Vote† 2016	Trump evaluation	Vote† 2016	Trump evaluation	Vote† 2020	Trump evaluation
	OR	β	OR	β	OR	β
Step 1: R2	0.600***	0.381***	0.555***	0.425***	0.677***	0.509***
Political party	5.28***	0.62***	4.57***	0.65***	6.92***	0.71***
Step 2: ∆R2	0.649**	0.013	0.596**	0.032**	0.690	0.0120.069
Political party	5.99***	0.61***	4.81***	0.62***	6.90***	0.71***
Gender	1.06	0.08	2.06**	0.17***	0.98	−0.01
Race	1.21	−0.03	1.30	0.07	1.630.073	0.10*
Education	0.55**	−0.08	0.82	−0.04	0.84	0.04
Step 3: ∆R2	0.6640.087	0.041***	0.622*	0.022**	0.721**	0.064***
Political party	5.44***	0.53***	4.24***	0.57***	5.98***	0.60***
Gender	0.89	0.03	1.84*	0.14**	0.84	−0.080.062
Race	1.38	0.01	1.45	0.090.092	1.85*	0.13**
Education	0.55**	−0.06	0.79	−0.04	0.84	0.03
PMI	0.96	0.03	1.56	0.090.084	0.98	0.02
HM	1.89*	0.22***	1.850.069	0.100.075	2.49**	0.27***
Step 4: ∆R2	0.705	0.022	0.634	0.035	0.766	0.025

*P < 0.05. **P < 0.01. ***P < 0.001.

OR = odds ratio; R values for binary logistic models refer to Nagelkerke R associated with each step.

We additionally tested for PMI effects in two ways. First, in the hierarchical regressions, a PMI effect could be evidenced by a significant ΔR2 and a significant PMI × gender interaction associated with Step 4. The PMI X gender interaction emerged as significant in two cases () but exhibited contradictory patterns of findings (). Second, PMI effects might also be revealed if PMI predicted support of Trump for men but not women. We conducted analyses separately for men and women (). These analyses failed to reveal evidence of precarious masculinity effects; HM, but not PMI, predicted support for Trump for both men and women. These analyses did, however, point to the possibility of stronger HM effects in men than women, which could be related to PMI effects to which we return attention in the discussion.

Consistent with our conceptualizations of HM, and in contrast to populist explanations of Trump’s 2016 victory, HM also predicted voting for Trump and support for Trump, over and above trust in government, political party affiliation, gender, race, and education (left panel of Table 2, Step 4). Study 3 was conducted following the 2016 election and results revealed that trust in government did not predict 2016 voting for or evaluations of Trump (Steps 3 and 4 in left panel of Table 2). Whereas trust in government and evaluations of Trump were negatively correlated prior to the 2016 election (r = −0.209, P < 0.001), the direction of the relation reversed by the end of President Trump’s first term in office: 50 days prior to the 2020 election (Study 7), trust in the government was positively associated with evaluations of President Trump (r = 0.183, P = 0.002). Therefore, it is not surprising that the findings of Study 7 (right panel of Table 2) show that trust in the government predicted 2020 evaluations of Trump and marginally predicted intended voting for Trump (Step 3). Consistent with our theoretical conceptualization, however, HM continued to account for unique variance in both voting for and evaluations of Trump, over and above the effects of trust in the government, political party affiliation, gender, race, and education (Step 4).

Table 2.

Results of hierarchical regressions for voting for Trump (binary logistic) and evaluations of Trump (linear), including trust in the government, for Studies 3 and 7

	Study 3		Study 7
Independent variables	Vote† 2016	Trump evaluation	Vote† 2016	Trump evaluation
	OR	β	OR	β
Step 1: R2	0.605***	0.463***	0.677***	0.509***
Political party	5.82***	0.68***	6.92***	0.71***
Step 2: ∆R2	0.613	0.004	0.690	0.0120.069
Political party	5.79***	0.67***	6.90***	0.71***
Gender	1.08	0.02	0.98	−0.01
Race	1.40	0.06	1.630.074	0.10*
Education	0.94	−0.03	0.84	0.04
Step 3: ∆R2	0.620	0.002	0.6990.080	0.024***
Political party	5.35***	0.66***	7.02***	0.71***
Gender	1.08	0.02	0.93	−0.02
Race	1.39	0.06	1.660.072	0.10*
Education	0.96	−0.02	0.79	0.02
Trust in government	0.23	−0.05	9.530.077	0.16***
Step 4: ∆R2	0.639*	0.035***	0.725*	0.054***
Political party	5.14***	0.60***	6.01***	0.61***
Gender	0.99	−0.02	0.82	−0.08*
Race	1.480.091	0.070.078	1.86*	0.12**
Education	0.99	−0.01	0.81	0.02
Trust in government	0.41	−0.01	5.01	0.12**
PMI	1.22	−0.04	0.96	0.03
HM	1.64*	0.22***	2.40*	0.25***
Step 5: ∆R2	0.678	0.034*	0.744	0.019
HM × race	0.30*			0.10*
HM × party		0.17**		0.10*
HM × PMI		−0.14**

*P < 0.05. **P < 0.01. ***P < 0.001. For discussion of interactions, please see .

OR = odds ratio; R2 values for binary logistic models refer to Nagelkerke R2 associated with each step.

In the present data, HM was also strongly related to prejudice (i.e., sexism, racism, homophobia, xenophobia, and Islamophobia). We present meta-analytic summaries (70) of correlations between the variables measured in more than one study, as well as the zero-order correlations that emerged in instances where variables were measured in a single study (e.g., Islamophobia)(). The top panel of presents correlations between variables assessing demographics, masculinity, and support for Trump. Relations of those variables to measures of sexism, racism, homophobia, xenophobia, and Islamophobia are presented in the bottom half of . We found large correlations between HM and sexism. Endorsement of HM was associated with more 1) benevolent sexism, or stronger beliefs that gender-stereotypic women should be cherished and protected, and 2) hostile sexism, or antipathy toward gender-violating women (71). In addition, the correlations between HM and sexism were larger than the correlations between PMI and sexism. Similarly, HM was more strongly associated with racism than was PMI. The greater endorsement of HM was associated with less pro-Black attitudes, or positivity toward African Americans as victims of past injustices, and stronger anti-Black attitudes, or negativity toward African Americans who are viewed as pushing for unearned rewards (72). Greater endorsement of HM (compared to PMI) was also more associated with greater dislike of gay men (homophobia), people from other countries (xenophobia), and Muslims (Islamophobia). When correlations were estimated separately for men, similar patterns emerged, although correlations between PMI and prejudice (e.g., sexism, racism, and homophobia) were slightly stronger.

Consistent with our theoretical conceptualization, HM was also distinct from sexism, racism, and prejudice (i.e., homophobia, xenophobia, and Islamophobia). In fact, HM predicted evaluations of Trump across Studies 4 through 7, over and above prejudiced attitudes. Inclusion of the masculinity variables in Step 4 was associated with a significant ΔR2 in evaluations of Trump, and HM predicted evaluations of Trump, over and above sexism (Table 3), racism (Table 4), xenophobia, homophobia, and Islamophobia (Table 5), as well as the other demographic variables. Contrary to predictions, however, prejudiced attitudes (i.e., sexism, racism, homophobia, xenophobia, or Islamophobia) but not HM predicted voting for Trump. Specifically, on voting, Step 4 as not consistently associated with a significant ΔR2 and when it was, HM was only a marginally significant predictor of voting. These effects replicated in a nationally representative sample were gathered 50 days prior to the 2020 presidential election (right panel of Tables 3 and 4); as shown in Table 5 (see Step 4), HM, as well as political party affiliation and prejudice (xenophobia, homophobia, or Islamophobia), predicted evaluations of Trump, whereas prejudice (xenophobia, homophobia, or Islamophobia) and political party affiliation predicted intent to vote for Trump in the 2020 presidential election. We present the findings of Study 7 in Table 5 because Study 7 included measures of prejudice, HM, and PMI, whereas PMI was omitted from Study 6 ().

Table 3.

Results of hierarchical regressions for voting for Trump (binary logistic) and evaluations of Trump (linear), including sexism, for Studies 4 and 7

	Study 4		Study 7
Independent variables	Vote† 2016	Trump evaluation	Vote† 2020	Trump evaluation
	OR	β	OR	β
Step 1: R2	0.593***	0.498***	0.677***	0.508***
Political party	5.02***	0.71***	6.92***	0.71***
Step 2: ∆R2	0.627*	0.0160.074	0.689	0.0120.072
Political party	5.45***	0.69***	6.89***	0.71***
Gender	0.96	0.090.076	0.98	−0.01
Race	1.28	0.04	1.630.073	0.10*
Education	0.66**	−0.10*	0.84	0.04
Step 3: ∆R2	0.668**	0.030**	0.709*	0.053***
Political party	5.40***	0.61***	6.01***	0.60***
Gender	1.00	0.06	0.89	−0.070.073
Race	1.58	0.07	1.720.056	0.12**
Education	0.63	−0.10*	0.86	0.04
Benevolent sexism	2.62**	0.090.094	1.18	0.11*
Hostile sexism	1.05	0.13*	1.49*	0.19***
Step 4: ∆R2	0.673	0.015*	0.723	0.018**
Political party	5.37***	0.60***	5.89***	0.57***
Gender	0.89	0.02	0.86	−0.09*
Race	1.670.084	0.08	1.81*	0.13**
Education	0.63**	−0.090.061	0.84	0.04
Benevolent sexism	2.39*	0.05	0.93	0.04
Hostile sexism	0.92	0.08	1.21	0.11*
PMI	0.79	−0.08	0.95	0.00
HM	1.57	0.17*	2.27*	0.20**
Step 5: ∆R2	0.711	0.051*	0.758	0.025
HM × race		−0.13*		0.10*
HM × party		0.19**
PMI × gender		0.17**

*P < 0.05. **P < 0.01. ***P < 0.001. For discussion of interactions, please see .

OR = odds ratio; R values for binary logistic models refer to Nagelkerke R associated with each step.

Table 4.

Results of hierarchical regressions for voting for Trump (binary logistic) and evaluations of Trump (linear), including racism, for Studies 5 and 7

	Study 5		Study 7
Independent variables	Vote† 2016	Trump evaluation	Vote† 2020	Trump evaluation
	OR	β	OR	β
Step 1: R2	0.639***	0.475***	0.677***	0.509***
Political party	5.05***	0.69***	6.92***	0.71***
Step 2: ∆R2	0.645	0.041*	0.690	0.0120.069
Political party	4.93***	0.64***	6.90***	0.71***
Gender	0.90	0.19**	0.98	−0.01
Race	1.36	0.08	1.630.074	0.10*
Education	0.88	−0.05	0.84	0.04
Step 3: ∆R2	0.695*	0.062***	0.769***	0.122***
Political party	4.00***	0.50***	5.11***	0.45***
Gender	0.88	0.15*	0.91	−0.03
Race	0.96	0.02	1.32	0.05
Education	0.97	−0.02	0.71	0.02
Pro-Black	0.42*	−0.17*	0.50*	−0.29***
Anti-Black	1.980.071	0.23**	2.02*	0.20***
Step 4: ∆R2	0.727*	0.042**	0.7820.094	0.022***
Political party	4.24***	0.49***	5.20***	0.42***
Gender	0.61	0.08	0.79	−0.05
Race	1.17	0.02	1.30	0.070.066
Education	0.93	−0.05	0.710.094	0.014
Pro-Black	0.35*	−0.16*	0.46**	−0.30***
Anti-Black	1.68	0.14*	1.94*	0.12*
PMI	0.250.080	−0.03	2.12	0.070.078
HM	2.760.069	0.24***	1.29	0.13**
Step 5: ∆R2	0.839*	0.059	0.796	0.023
PMI × Race	0.01*
HM × Party		0.20*
PMI × Gender	0.01*

*P < 0.05. **P < 0.01. ***P < 0.001. For discussion of interactions, please see .

OR = odds ratio; R values for binary logistic models refer to Nagelkerke R associated with each step.

Table 5.

Results of hierarchical regressions for voting for Trump (binary logistic) and evaluations of Trump (linear), including xenophobia, homophobia, and Islamophobia, for Study 7

	Xenophobia		Homophobia		Islamophobia
Independent variables	Vote† 2020	Trump evaluation	Vote† 2020	Trump evaluation	Vote† 2020	Trump evaluation
	OR	β	OR	β	OR	β
Step 1: R2	0.677***	0.508***	0.677***	0.508***	0.678***	0.509***
Political party	6.90***	0.71***	6.92***	0.71***	6.90***	0.71***
Step 2: ∆R2	0.689	0.0120.071	0.689	0.0120.072	0.691	0.0120.079
Political party	6.89***	0.71***	6.89***	0.71***	6.88***	0.71***
Gender	0.98	−0.01	0.98	−0.01	0.99	−0.02
Race	1.630.076	0.10*	1.630.073	0.10*	1.650.069	0.10*
Education	0.84	0.04	0.84	0.04	0.85	0.04
Step 3: ∆R2	0.729***	0.097***	0.768***	0.106***	0.747***	0.096***
Political party	5.60***	0.53***	5.33***	0.48***	5.90***	0.55***
Gender	0.92	−0.04	0.96	−0.05	0.94	−0.03
Race	1.46	0.05	1.58	0.11**	1.56	0.07
Education	0.79	0.04	0.77	0.03	0.84	0.04
Prejudice	1.88***	0.37***	3.37***	0.40***	2.17***	0.35***
Step 4: ∆R2	0.738	0.015**	0.771	0.010*	0.751	0.013*
Political party	5.43***	0.50***	5.36***	0.48***	5.84***	0.53***
Gender	0.85	−0.070.084	0.90	−0.070.057	0.89	−0.06
Race	1.59	0.08*	1.61	0.12**	1.63	0.09*
Education	0.80	0.04	0.77	0.03	0.84	0.04
Prejudice	1.62*	0.29***	3.08***	0.33***	1.92**	0.28***
PMI	1.07	0.02	1.27	0.03	1.16	0.01
HM	1.68	0.15**	1.25	0.11*	1.39	0.14**
Step 5: ∆R2	0.754	0.014	0.801	0.016	0.782	0.014

*P < 0.05. **P < 0.01. ***P < 0.001. For discussion of interactions, please see .

OR = odds ratio; R values for binary logistic models refer to Nagelkerke R associated with each step.

Whereas HM consistently predicted evaluations of Trump in 2016 and 2020, evaluations of Trump’s democratic challengers—Clinton in 2016 and Biden in 2020—were more consistently and robustly predicted by prejudiced attitudes. As shown in Table 6 (Clinton) and Table 7 (Biden), the inclusion of trust in government and/or prejudiced attitudes (i.e., sexism, racism, xenophobia, homophobia, or Islamophobia) in Step 3 consistently predicted evaluations of both Clinton and Biden, as evidenced by a significant ΔR2 with one exception: racist attitudes (pro-Black or anti-Black) did not predict evaluations of Clinton in Study 5. Greater endorsement of pro-Black attitudes predicted more positive evaluations of Biden (but not Clinton), and greater endorsement of anti-Black attitudes predicted more negative evaluations of Biden (but not Clinton). In other words, prejudiced attitudes, but not HM, predicted evaluations of both Clinton and Biden. In fact, there was only a single instance in which the inclusion of the masculinity variables (Step 4) accounted for additional variance in evaluations of the democratic candidate; HM accounted for unique variance in evaluations of Clinton in Study 3, over and above trust in government, political party affiliation, gender, race, and education. In addition, in the same Study, as evidenced by a significant ΔR2 in Step 3, trust in government predicted positive evaluations of Clinton, while HM and Republican Party affiliation predicted more negative evaluations. Interestingly, regardless of whether prejudiced attitudes were included in the model (Tables 6 and 7) or omitted from the model (), HM did not consistently predict evaluations of Clinton.

Table 6.

Results of hierarchical regressions for evaluations of Clinton (Studies 3 through 6), including social prejudice

Independent variables	Trust government	Sexism	Racism	Xenophobia	Homophobia
Step 1: R2	0.407***	0.344***	0.246***	0.261***	0.264***
Political party	−0.64***	−0.59***	−0.50***	−0.51***	−0.51***
Step 2: ∆R2	0.019*	0.0230.052	0.024	0.040***	0.039***
Political party	−0.62***	−0.58***	−0.49***	−0.48***	−0.49***
Gender	−0.12**	−0.07	−0.11	−0.18***	−0.17***
Race	0.00	0.00	0.03	0.04	0.04
Education	0.080.093	0.14*	0.10	0.09*	0.090.064
Step 3: ∆R2	0.044***	0.024*	0.027	0.013*	0.037***
Political party	−0.55***	−0.51***	−0.41***	−0.42***	−0.36***
Gender	−0.12**	−0.05	−0.08	−0.16***	−0.090.075
Race	0.00	−0.03	0.06	0.05	0.01
Education	0.06	0.14*	0.08	0.09*	0.080.067
Attitude 1	0.23***	−0.04	0.08	−0.13*	−0.25***
Attitude 2	—	−0.15*	−0.160.052	—	—
Step 4: ∆R2	0.015*	0.009	0.022	0.004	0.000
Political party	−0.51***	−0.50***	−0.42***	−0.41***	−0.36***
Gender	−0.09*	−0.02	−0.10	−0.13*	−0.09
Race	−0.01	−0.03	0.07	0.04	0.01
Education	0.06	0.12*	0.07	0.09*	0.080.068
Attitude 1	0.21***	−0.04	0.05	−0.110.059	−0.24***
Attitude 2	—	−0.150.057	−0.170.060	—	—
PMI	0.00	0.110.075	−0.150.054	—	—
HM	−0.13*	−0.06	0.03	−0.08	−0.02
Step 5: ∆R2	0.010	0.026	0.110	0.010	0.008

*P < 0.05. **P < 0.01. ***P < 0.001.

Table 7.

Results of hierarchical regressions for evaluations of Biden (Study 7), including social prejudice

Independent variables	Trust government	Sexism	Racism	Xenophobia	Homophobia	Islamophobia
Step 1: R2	0.356***	0.358***	0.356***	0.356***	0.358***	0.356***
Political party	−0.60***	−0.60***	−0.60***	−0.60***	−0.60***	−0.60***
Step 2: ∆R2	0.007	0.007	0.007	0.007	0.007	0.007
Political party	−0.59***	−0.59***	−0.59***	−0.59***	−0.59***	−0.59***
Gender	0.05	0.05	0.05	0.05	0.05	0.05
Race	−0.04	−0.04	−0.04	−0.04	−0.04	−0.05
Education	0.05	0.05	0.05	0.05	0.05	0.05
Step 3: ∆R2	0.036***	0.019*	0.072***	0.016**	0.035***	0.023**
Political party	−0.59***	−0.55***	−0.39***	−0.52***	−0.46***	−0.51***
Gender	0.04	0.05	0.05	0.06	0.07	0.06
Race	−0.04	−0.03	0.00	−0.03	−0.05	−0.03
Education	0.03	0.05	0.06	0.05	0.05	0.04
Attitude 1	0.19***	0.11*	0.27***	−0.15**	−0.23***	−0.17**
Attitude 2	—	−0.17**	−0.10	—	—	—
Step 4: ∆R2	0.005	0.001	0.003	0.001	0.005	0.003
Political party	−0.56***	−0.55***	−0.40***	−0.52***	−0.47***	−0.52***
Gender	0.05	0.05	0.04	0.06	0.06	0.05
Race	−0.05	−0.03	0.01	−0.02	−0.04	−0.02
Education	0.03	0.04	0.06	0.04	0.05	0.04
Attitude 1	0.20***	0.120.070	0.28***	−0.16*	−0.28***	−0.20**
Attitude 2	—	−0.17*	−0.130.054	—	—	—
PMI	0.03	0.04	−0.02	0.03	0.02	0.04
HM	−0.09	−0.04	0.08	0.01	0.08	0.04
Step 5: ∆R2	0.034	0.039	0.049	0.047†	0.042†	0.052†

*P < 0.05. **P < 0.01. ***P < 0.001.

A significant step at P < 0.05. All interactions for Study 7 are presented and fully discussed in the .

Discussion

Consistent with our theoretical backdrop, findings across seven studies revealed that HM predicted voting for and evaluations of Trump, over and above the variance accounted for by political party affiliation, gender, race, and education. These findings were documented in the days following the 2016 American presidential election (Studies 1 and 2) and replicated in the months following the 2016 election (Studies 3 through 6), as well as the days preceding the 2020 presidential election (Study 7). In fact, HM predicted voting for and evaluations of Trump equally well for women and men, White and non-White participants, Democrats and Republicans, and across levels of education. The same patterns emerged when controlling for trust in government (Study 2), as well as demographic variables, indicating that the HM effect could not be attributed to a populist perspective.

As conceptualized, HM was related to prejudiced attitudes toward women and marginalized groups in America. In the present data, HM was associated with more benevolent and hostile sexist attitudes (Studies 4 and 7), as well as the weaker endorsement of pro-Black attitudes and greater endorsement of anti-Black attitudes (Studies 5 and 7). HM was also related to more xenophobic attitudes (Studies 6 and 7), homophobic attitudes (Studies 6 and 7), and Islamophobic attitudes (Study 7). Although HM correlated with prejudiced attitudes toward women and marginalized groups, HM continued to predict positive evaluations of Trump when controlling for these prejudiced attitudes. Because gender- and race-based hierarchies are embedded within the culturally exalted form of masculinity (48), the endorsement of HM may mask the operation of dominant group supremacy. Analogous to how benevolent feelings of paternalism can mask open gender- and race-based inequities from the view of discriminators (73), the endorsement of culturally idealized forms of masculinity may mask the homophobic-, xenophobic-, and Islamophobic-based inequities from the view of actors. For example, one’s exaltation of good men may make it difficult for a given person to see one’s own race- and/or ethnicity-based antipathy toward and dehumanization of men who belong to marginalized and/or subordinated groups and are being judged to be bad men. The elevation of idealized forms of masculinity may, like the paternalistic desire to protect nonthreatening women, contribute to one’s self-veiling and self-motivated lack of awareness of one’s own misogynistic acts of aggression toward gender and status-quo threatening women (71) and/or one’s dominant group supremacist acts of violence and dehumanization toward men who belong to marginalize and subordinated groups.

HM was also distinct from prejudiced attitudes in terms of the outcomes they predict. While HM consistently predicted positive evaluations of Trump, prejudiced attitudes–and not HM–more reliably predicted voting and evaluations of Trump’s Democratic opponent (Clinton in 2016 and Biden in 2020). In other words, while HM influenced support for Trump, prejudiced attitudes were more influential in evaluations of those who are questioning the status quo (Clinton and Biden) than those who expressed open vitriol toward women and marginalized men (Trump). By predicting evaluations of different candidates in unique ways, processes of hegemony may operate more effectively to justify and legitimize hegemonic masculine dominance over women of intersecting identities and marginalized men. As noted, when included in the models and contrary to predictions, prejudiced attitudes were more consistent predictors of voting than HM. In hindsight, HM may be more directly related to support for Trump, given suggestions that some people in the 2016 and 2020 elections hid their support of Trump in light of awareness of Trump’s open misogyny, racism, and nationalism (74). By contrast, perhaps HM is not needed to veil the role of prejudice in voting, as decisions for whom to vote are naturally veiled under candidates’ differential stances on a host of issues if import beyond prejudice. Additional research is needed to fully understand the outcomes predicted by HM versus open prejudice.

Our findings also point to the potential importance of context on the nature and consequences of prejudiced attitudes toward various groups in America. As noted above, in the United States of America, prejudice and civil rights movements have been defined almost exclusively in terms of racism toward Black Americans and sexism, which, until 2016, raised social desirability concerns about appearing to be prejudiced. For instance, contemporary theories of racial prejudice assume that White people have ambivalent attitudes toward racial minorities, including sincerely positive attitudes related to egalitarian principles and unacknowledged and negative attitudes of which they are often unaware (73, 75, 76); as a result, contemporary racism is theorized to be particularly influential when race-based biases in judgment and behavior can be attributed to something other than racist attitudes, masking White people’s negative attitudes (75). Contemporary theories of sexism are also conceptualized in terms of ambivalent attitudes, with feelings of benevolent and paternalistic protections directed toward nonthreatening and gender-stereotypic women and feelings of hostility and aggression directed toward-threatening women who violate gender stereotypes (71); the adoration of stereotypic “good” women (e.g., wives, daughters, and service providers) masks the antipathy toward threatening “bad” women (e.g., well-performing colleagues and bosses). However, other prejudiced attitudes, like those toward immigrants, gay men, and transgender women (particularly transgender women of color), are better characterized by antipathy and open hostility in the absence of any substantive positive attitudes. For example, in recent years, there has been an increased visibility and prevalence of immigrant detention practices, as well as increased incidences of hate crimes toward gay men and transgender women in the wake of lesbian, gay, bisexual, and transgender civil rights gains (77).

The present theory and research point to the need to articulate the relation between HM and prejudices toward various marginalized groups with different intersecting identities, different histories, and different contemporary stations. Future research examining the role of HM and prejudiced attitudes in politics outside of the United States would also add to our understanding of how context affects the present findings.

Comparing findings across studies highlights the importance of considering when and with what consequences HM and/or PMIs influence political thought and candidate evaluations. Presently, little attention is focused on PMI because we directed attention only to those effects that were replicated across studies and that included no PMI effects. It is important to note that we measured chronic PMI, or individual differences in the degree that people were concerned about masculine shortcomings. As noted, much of the work on PMI examines the consequences of situational threats to masculinity, which were beyond the scope of the present work. Importantly, some PMI effects did emerge in the present data but in isolated instances. Because studies were also conducted across time, it is possible that PMI becomes pertinent to particular social contexts. A primary question raised by the present work is when and with what consequences do HM and/or precarious masculinity predict status quo maintaining consequences?

The present work also raises critical questions about whether the aggressive and violent acts that follow from threats to the PMIs of individual men reinforce and maintain HM and vice versa. More generally, theoretical perspectives on HM position the racialized gender binary at the core of systems of inequity (19, 62). HM is a legitimating ideology that justifies and reinforces gender and racial hierarchies. This is consistent with influential theories of system justification (16) and social dominance orientation (17), with the current findings pointing to another potentially critical legitimating ideology (or hierarchy enhancing ideology) that operates within the context of those theories: HM. However, the present theory and research also points to HM as an unique ideology that unifies 1) dominant men and marginalized men in their perceived superiority and rightful dominance over women, 2) dominant men and dominant women in their perceived superiority and dominance over marginalized people, and 3) dominant men, marginalized men, and women in their perceived superiority and rightful dominance over immigrant, foreign, and indigenous people who are dehumanized and seen as less human. This functionally ties groups of men and women of lower ranks to dominant men in different contexts, reinforcing and justifying existing hierarchies—as the unions of some groups assure dominance over others. In addition, because the subordinate groups that align with dominant groups vary across contexts, HM may be a particularly effective ideology that functionally prevents the formation of alliances and collective action (78) across different groups of marginalized people.

Materials and Methods

All studies described were approved by the authors’ Institutional Review Board. Across all seven studies, participants first read the consent statement and, by continuing with the survey, implied their consent to participate. Participants were able to skip questions and/or withdraw their participation at any time. All of our data and syntax are available at https://osf.io/u3g2n/ (79).

Participants.

We recruited 2,007 participants from both Amazon’s MTurk and The Pennsylvania State University’s subject pool between November 10, 2016, and December 6, 2017, and from Prolific on September 14, 2020. Participants recruited from MTurk were compensated $0.50 for their participation; participants recruited from the subject pool were given partial course credit for their participation and participants recruited through Prolific were compensated an average of $9.66/h. The includes additional information about each sample.

Procedure.

Studies 1 and 2.

After reading consent statements, participants indicated their endorsement of HM (53), their PMI (69), and political affiliation. They also indicated for whom they voted in the 2016 presidential election (or, if they did not vote, for whom they would have voted), evaluated Trump and Clinton, and provided demographic information. The include a full description of each measure.

Studies 3 through 6.

Similar to Studies 1 and 2, participants indicated their endorsement of HM and their PMI (participants did not complete the Male Gender Role Stress Scale in Study 6). Participants then reported either their trust in the government (Study 3), sexism (Study 4), racism (Study 5), or homophobia and xenophobia (Study 6). Participants indicated their political affiliation, their vote in the 2016 presidential election, and evaluated Trump and Clinton before providing demographic information. The includes a full description of each measure.

Study 7.

Identical to the previous studies, participants indicated their endorsement of HM and their PMI. They then reported, in random order, their trust in the government, sexism (80), racism, homophobia, xenophobia, and Islamophobia (81) before indicating their political affiliation, their vote in the 2016 presidential election, and their anticipated vote in the 2020 presidential election. Finally, they evaluated Trump and Biden before providing demographic information. The includes a full description of each measure.

Data analysis.

To test whether HM accounted for unique variance support for Trump over and above demographic variables known to have affected political behavior in the 2016 election (82), we conducted two hierarchical regressions on each dependent variable: voting for and evaluations of Trump. Parallel analyses were also performed on evaluations of Democratic challengers: Clinton in 2016 and Biden in 2020.

Hierarchical binary logistic regression was used to analyze voting for Trump (vote for Trump = 1, vote for another candidate = −1) and hierarchical linear regression was used to analyze evaluations of Trump, evaluations of Clinton, and evaluations of Biden. In the first set of hierarchical regression analyses, predictors were included in four steps. First, political party (higher numbers indicate stronger Republican affiliation) were entered in Step 1. Second, gender (1 = male, −1 = female), race (1 = White, −1 = non-White), and level of education (higher numbers indicate higher education) were included in Step 2. In Step 3, both HM and PMI were included. Finally, the two-way interactions of HM with each of the three demographic variables (i.e., HM × Gender, HM × Race, HM × Education), the two-way interactions of PMI with each of the three demographic variables (i.e., PMI × Gender, PMI × Race, and PMI × Education), the two-way interaction between HM and PMI, and the three-way interactions of HM, PMI, and the three demographic variables (i.e., HM × PMI × Gender, HM × PMI × Race, HM × PMI × Education, and HM × PMI × Party) were added in Step 4.

To examine whether hegemony was distinct from related social and prejudiced attitudes, a second set of hierarchical regressions analyses were performed on each variable in Studies 3 through 7, in which social or prejudiced attitudes were measured. We conducted parallel hierarchical regressions (binary logistic for voting for Trump and linear for candidate evaluations) to test the HM hypothesis alongside alternative explanations of the findings and/or the conceptualization of masculinity. Again, political party was entered in Step 1. Gender, race, and level of education were entered in Step 2. Social or prejudiced attitudes of interest—trust in the government, sexism, racism, xenophobia, homophobia, and Islamophobia—were entered in Step 3. HM and PMI were entered in Step 4. All two-way interactions that included HM and PMI were included in the final step.