A (Political) Battle of the Sexes: Gender Polarization on Gender‐Related Attitudes Across 14 Societies

ABSTRACT

Although it is well‐established that women typically endorse more progressive gender‐related attitudes compared to men, key questions remain about the magnitude of this gender difference and whether it systematically varies in relation to levels of gender inequality across societies. In the current research, we aim to address these questions by examining endorsement of traditional (overt) and modern (indirect) forms of sexism across 14 countries, including those with relatively low levels of gender equality (e.g., Saudi Arabia, India, and Egypt) and those with relatively high equality (e.g., Sweden, the United Kingdom, and the United States). Results show that endorsement of old‐fashioned sexism is inversely related to country‐level gender equality for women and men, whereas modern sexism endorsement is inversely related to gender equality for women, but not men. Thus, gender polarization is largest on modern sexist attitudes in relatively high gender‐equal countries, and this is because in such contexts modern forms of sexism are especially likely to be rejected by women.

An examination of gender‐related attitudes across 14 countries showed that endorsement of old‐fashioned forms of sexism (men's superiority and traditional gender roles) is inversely related to country‐level gender equality for women and men, whereas modern sexism (denial of gender discrimination and belief that men are disadvantaged) was inversely related to gender equality among women, but not men. Thus, modern forms of sexism in egalitarian societies showed the largest gender gaps.

1. Introduction

Recent media reports have sounded dramatic alarms over how women and men are more politically polarized than ever before [1, 2, 3, 4]. A Financial Times article, for instance, highlighted striking gender gaps in political attitudes across the United States, the United Kingdom, Germany, and South Korea [5], suggesting a global trend and one that is potentially driven by rising gender equality. Indeed, reports delving into gender gaps in political attitudes in the United States argue that this so‐called gender war appears to be fueled by divisions in gender‐related attitudes [6, 7]. Many of these accounts posit that women's expanding opportunities—and men's perceived decline in relative societal standing—have provoked a conservative backlash among men [2, 7, 8]. This is surprising because most social scientific analyses suggest that increasing gender equality tends to foster more egalitarian attitudes among both men and women [9, 10, 11, 12, 13, 14, 15, 16, 17].

The present research seeks to better understand the relationship between societal‐level gender equality and individuals’ gender‐related attitudes across various domains, with a particular focus on the extent of polarization between women and men. Although it is well‐established that women typically endorse more progressive gender beliefs than men, key questions remain about the magnitude of this gender difference and whether it systematically varies in relation to levels of gender inequality across societies. Toward that end, we examine various gender‐related attitudes in samples from very diverse (and often understudied) countries, including those with relatively low levels of gender equality (e.g., Saudi Arabia, India, and Egypt) and those with relatively high equality (e.g., Sweden, the United Kingdom, and the United States).

Our general thesis is that the association between gender equality and the extent of gender gaps in attitudes will differ depending on the form of the gender‐related belief. More specifically, we expect that traditional (or “old‐fashioned”) sexist beliefs about gender roles will be inversely related to societal levels of gender equality and to roughly the same extent for men and women, whereas more contemporary versions—for example, denial of discrimination against women, or the belief that men are now the disadvantaged group—will be more strongly rejected by women, especially in societies with relatively high levels of gender equality. Thus, our prediction is in line with the supposition that the gender gap in (at least some) gender‐related attitudes will be larger in places with relatively high (vs. low) gender equality, but it diverges from the media accounts described above that the gap will be driven by men's attitudes. Instead, we expect that women living in more (vs. less) equal societies—who presumably have greater awareness around gender‐related injustice—will find the denial of discrimination toward women (and claims of discrimination against men) especially untenable.

1.1. Progress Toward Gender Equality?

In recent decades, women around the world have made unprecedented progress in educational, economic, and political equality relative to men. Since 1995, countries around the globe have collectively adopted over 1500 legal reforms advancing gender equality, and women's political representation has more than doubled [18]. The global gender gap in educational attainment is now over 95% closed, and the gap in economic participation and opportunity has closed by 60% [19]. A host of studies have documented how system‐level shifts toward gender equality have been associated with more progressive gender‐related attitudes and beliefs for both men and women, on both implicit and explicit measures [9, 10, 11, 12, 15, 16, 17, 20]. This association appears to be driven by reciprocal forces—that is, egalitarian attitudes about gender lead to the adoption of policies that promote gender equality [14, 21, 22] and greater inclusion of women in the economic and political spheres leads to more gender‐egalitarian attitudes [23, 24, 25].

Despite these advances, gender disparities—and the beliefs that serve to justify them—persist. No country has reached full gender parity, and projections suggest it will take 134 years to do so [26]. Countries in Europe and North America typically rank highest on various measures of gender equality, but their recent progress has been stagnant [27, 28]. For instance, from 2006 to 2024, economic gender parity stayed relatively constant in Europe and slightly declined in North America [19, 27, 28]. Evidence that the Western gender revolution has stalled also emerges from work examining attitudes, which found rapid shifts toward more egalitarian attitudes in previous decades, and small or negligible shifts in more recent years [9, 10, 13]. At the same time, other regions of the world are undergoing rapid change. For instance, in 2017, women made up just 17% of Saudi Arabia's labor force; by 2024, that share rose to 36% [29, 30]. Thus, although the Saudi Arabian workforce includes fewer women compared to most Western nations, it is also experiencing one of the fastest and most dramatic shifts in female economic participation anywhere in the world. One question is how attitudes (and gender polarization in attitudes) in these rapidly transforming places compare to those in societies where gender equality has been relatively stable in recent years.

1.2. Gender‐Related Attitudes

Decades of social science research have documented the myriad beliefs that people hold that serve to explain or legitimize the existence (and persistence) of gender inequality. So‐called old‐fashioned forms of sexism assessed the extent to which people overtly endorse traditional gender roles in the public (men in leadership roles) and/or private (e.g., women in domestic roles) spheres [31, 32], whereas more modern forms of justification for the gender system legitimize women's low status relative to men by more subtly implicating women's shortcomings—for instance, by denying the continuation of discrimination against women, or asserting discrimination against men.

1.2.1. Old‐Fashioned Forms of Sexism

One of the most influential early measures of gender‐related attitudes in social psychology was the “attitudes toward women” scale (ATW) [17, 33, 34]. Developed during the second‐wave feminist movement in the United States, where traditional gender roles were being questioned and challenged, the scale assessed people's beliefs about appropriate societal privileges and responsibilities for women versus men. (The authors of the scale have referred to the original title as “something of a misnomer” [17] and have conceded that the ATW scale is more accurately described as a measure of attitudes toward women's rights, as opposed to attitudes toward women more generally. For this reason, in our study, we refer to it as an endorsement of men's superiority.) Items thus asked participants the extent to which they agreed with statements such as “a woman should not expect to go to exactly the same places or to have quite the same freedom of action as a man,” and “the intellectual leadership of a community should be largely in the hands of men.” This measure was employed in approximately 400 psychological studies from its inception until 1990, making it easily the most commonly used gender‐related attitudes measure of its time [17, 34, 35]. Over time, this scale fell out of popularity because both women and men became less and less likely to endorse such blatant assertions of men's superiority, at least in the United States [17, 36]. However, recent studies conducted in less gender‐equal contexts—for example, rural Bangladesh [37]; India [38, 39]; Pakistan [40]; Saudi Arabia [41]; and Turkey [42]—have found that the scale has good predictive power. Across all contexts, women consistently score more egalitarian than men. Indeed, an analysis of 57 nations included in the World Values Survey showed that gender and gender inequality appear to have additive effects on the endorsement of these types of overtly sexist statements—that is, endorsement of men's superiority is higher among men (vs. women) across countries and inversely related to national gender inequality overall [21].

Around the same time as the ATW scale was developed, other (predominantly US‐based) social and political scientists were also developing measures to assess egalitarian attitudes that were more focused on the domestic/family context [9, 10, 31, 43, 44, 45]. As married American women began entering the workforce in record numbers, researchers were especially interested in the perceived conflict between women's participation in the workforce and their traditional domestic responsibilities [9, 10, 14]. Statements such as “working women put a strain on the family” have thus been staple items in US surveys since the 1970s, and continue to be included in international surveys today. Analyses of these survey questions have found that men consistently show greater endorsement of traditional domestic roles compared to women. Some work suggests that the gap between women and men on these questions is especially pronounced in places with low (vs. high) gender equality [46, 47]. However, time‐series analyses on US samples have found that men and women's attitudes have become more egalitarian over time, and roughly to the same extent, but the rate of change among both groups has slowed since the early 2000s [9, 10]. Similarly, recent analyses of international samples found that implicit and explicit stereotypes that associate men with careers and women with family have slowly but consistently decreased over the past 12 years, and to an equal extent for men and women [20].

Although overt prescriptions of women's domestic roles may be considered old‐fashioned in the Western world today, they may nonetheless miss the mark in places where the controversy is not just a perceived work/family conflict for women, but women being in the workplace at all. Research conducted in Saudi Arabia, Kuwait, and rural India, for instance, has shown that most people think other people disapprove of women working outside the home, regardless of their own beliefs [48, 49, 50]. Questions assessing attitudes toward women in the workplace are important indicators of gender‐related attitudes in surveys conducted across Africa and the Arab world [51, 52] but are virtually absent from most social psychological assessments, which tend to be formulated in Western settings. In one exception to this, a measure of egalitarian attitudes developed in Japan in the 1980s [45, 53] included a “vocational egalitarianism” subscale, which tapped the belief that women can achieve independence through working (e.g., “There will be much social progress and development when more women work.”) [see also Ref. 54]. This idea of encouraging women to work as part of a progressive agenda may be irrelevant (or nonsensical) in contemporary Western societies, where the vast majority of women are in the labor force [19]. However, these types of questions may very well be important measures of gender‐related attitudes for women and men in societies where structural realities and societal norms prevent many women from entering the workforce, and thus doing so, in and of itself, might be perceived as a radical act.

1.2.2. Modern Forms of Sexism

In more egalitarian societies, people tend to disavow overtly hostile intergroup attitudes, but beliefs that indirectly serve to perpetuate and maintain inequality tend to persist. Social psychologists have documented how justifications for racial and gender inequality have evolved from old‐fashioned beliefs that overtly assert White/men's superiority to more modern narratives that, while less blatant, still reinforce the existing social hierarchy [55, 56, 57, 58]. For instance, the Modern Sexism Scale [57] assesses the extent to which respondents deny the persistence of discrimination against women and resent “special favors” that aim to redress gender disparities. Although modern sexists do not express sentiments that explicitly derogate women, the denial of discrimination indirectly promotes the narrative that women are responsible for their own relatively low societal status—that is, if gender discrimination does not exist, then extant gender disparities must be stemming from legitimate sources (e.g., differences in effort, ability, or choice) [59, 60]. Thus, asserting that “discrimination against women is no longer a problem” allows individuals to oppose gender‐egalitarian policies without expressing overtly negative (or seemingly outdated) attitudes. Work has shown that the more anti‐women bias there is in a society, the more that members of that society deny that discrimination against women is a problem [16, 61].

Recently, scholars have honed in an even newer manifestation of gender‐related attitudes—namely, the belief that men, not women, are the disadvantaged gender group. This belief in a sexism shift [36] (also called “male victimhood ideology” [62]) contends that modern societies have unjustly favored women, casting men as the new victims of discrimination and exclusion. Like denial of discrimination against women, claims of discrimination against men do not explicitly belittle women; instead, they indirectly downplay women's ability and effort by implying that advances in gender equality have been obtained illegitimately (e.g., through favoritism and anti‐men bias). Research has demonstrated that the claim of discrimination against men is positively associated with other traditional and modern forms of sexism, but less subject to social desirability compared to other measures [36], which is in line with the proposition that it serves as a less overt (and thus more acceptable) form of sexism.

1.3. Gender‐Related Attitudes in a Global Context

Across all of these measures of gender‐related attitudes, decades of research from around the world consistently show that men's attitudes tend to be less egalitarian compared to women, and that people's attitudes become more egalitarian as gender equality increases—either over time or across contexts [9, 12, 13, 14, 15, 16, 17, 20, 21]. Few analyses, however, have examined different gender‐related attitudes in tandem, raising questions about the extent to which they are endorsed across contexts, by both men and women. For instance, research conducted in North America and Europe has shown that people are more likely to endorse modern (vs. old‐fashioned) forms of sexism [36, 57, 63, 64]; is this also true in less gender‐equal places? If modern forms of sexism have emerged as a replacement for more blatant sexism in relatively equal places where overt assertions of male superiority are no longer tolerated, the opposite pattern should emerge in unequal places—that is, people should be more likely to endorse old‐fashioned (vs. modern) forms of sexism in societies with low levels of gender equality. This seems unlikely, however, given research showing that old‐fashioned and modern sexism tend to co‐occur—for instance, the more people in a country endorse overtly sexist norms, the higher they are to deny the existence of discrimination against women, which is the cornerstone of modern forms of sexism [12, 13, 16]. Thus, it seems plausible that the gap between old‐fashioned versus modern forms of sexism will be greater in countries that are relatively high (vs. low) in gender equality.

A second, more focal, question about the interaction between gender, gender‐related attitudes, and context is about the extent of gender polarization. More specifically, although evidence shows that men robustly score higher than women across different measures of sexism, does the magnitude of this gap vary across attitudes and across contexts? Although media reports have claimed there is an emerging gender war (in terms of highly polarized gender‐related attitudes) in places that have achieved high levels of gender equality [2, 5, 8], academic studies have tended to show the opposite to be true—finding that sexism within a society decreases over time (as gender equality increases), and roughly to the same degree for men and women [9, 10, 20]. Findings from cross‐national examinations of gender polarization are less consistent—with some studies finding that gender polarization on assertions of men's superiority [21] and on endorsement of traditional domestic roles [46] does not vary as a function of gender equality, and other analyses using similar measures finding marked levels of gender polarization in countries with relatively low gender equality, and virtually no gender polarization on attitudes among those living in countries with relatively high equality [47].

None of these studies seem to suggest that gender polarization will be higher in places with high (vs. low) gender equality. However, most of these cross‐national examinations have focused on old‐fashioned sexist beliefs, because these types of measures are included in large international surveys. It could be the case that gender polarization in high gender‐equal countries may be more pronounced on modern forms of sexism. Although research has shown that modern sexism endorsement (i.e., denial of gender discrimination) is inversely related to gender equality [16, 61], no work that we are aware of has looked at polarization on these types of attitudes, cross‐nationally or over time. Research on ambivalent sexism theory, however, offers some clues about how gender polarization may differ for more overt (vs. subtle) forms of sexism. More specifically, social psychologists have shown that the two components of ambivalent sexism—namely, hostile sexism (HS; derogatory attitudes toward women who defy traditional gender roles) and benevolent sexism (BS; positive but patronizing attitudes toward traditional women)—are higher in more sexist places, and that gender gaps tend to be smaller on benevolent (vs. hostile) sexism [12, 65]. However, recent longitudinal and time‐series analyses conducted in a relatively high gender‐equal society (i.e., New Zealand) have found that women (but not men) are increasingly more likely to reject BS over time [66, 67]. In addition, an analysis of data from 81 countries from studies conducted between 1996 and 2023 found that although both forms of sexism decreased over time, the slope of decline for HS was roughly the same for women and men, whereas the decline of BS was steeper among women compared to men [13]. Taken together, these findings suggest that the gender gap on more modern (or indirect) forms of sexism may be growing, at least in more gender‐equal places.

Indeed, it is conceivable that women in more gender‐equal places are gaining greater awareness of the way that sexism operates and thus are more able to identify (and reject) sexism even in its subtler forms [68]. In line with this idea, a study among adolescent girls showed that those who had learned about feminism and/or those who recognized societal pressures for gender conformity were more likely to identify personal discriminatory experiences as sexism [69]. If shifts toward equality bring greater awareness, women in relatively egalitarian societies might be especially attuned to the discrepancy between the current and ideal state in terms of gender equality [70], and thus, they might find the denial of discrimination against women, or claims of discrimination against men, especially dubious.

1.4. Overview of Current Research

The current work aims to better understand the manifestation of different forms of (sexist) gender‐related attitudes across different societies, with a particular focus on attitude polarization between societies and between women and men. More specifically, we examine the interactive effects of attitude type, gender, and gender equality to shed light on the extent to which women and men living in places with varying degrees of gender equality endorse both overt (or old‐fashioned) and indirect (or modern) forms of sexism. Toward this end, we conducted surveys across 14 countries that assessed various forms of gender‐related attitudes—specifically, endorsement of men's superiority, traditional domestic roles for women, opposition to women in the workplace (which we call “vocational anti‐egalitarianism”), denial of discrimination against women, and belief in sexism shift (i.e., the perception that men are now the primary victims of discrimination). The countries included in the survey are listed in Table 1, and the gender‐related attitude items included are listed in Table 2. As seen in Table 1, the sample included respondents from countries such as Egypt, India, Indonesia, and Saudi Arabia—places that are severely underrepresented in previous psychological studies of gender attitudes, despite the extensive attention paid to their gender policies and practices [71, 72, 73, 74, 75].

TABLE 1.

Demographic data and gender equality indices for countries surveyed, ordered from highest to lowest in overall gender equality.

						Gender inequality indices
	n	% Female	Age, M (SD)	SES, M (SD)	% Married	GGGI	WPS	GII	SIGI	Total
Sweden	487	49.6	47.60 (16.85)	5.72 (1.99)	56.0	0.823	0.895	0.977	0.910	0.901
The United Kingdom	451	53.0	49.62 (16.13)	5.65 (1.78)	62.2	0.775	0.888	0.902	0.880	0.861
The United States	482	49.3	48.35 (17.18)	6.21 (2.01)	55.9	0.763	0.861	0.821	0.810	0.814
South Korea	485	49.2	45.17 (13.85)	5.57 (1.76)	57.2	0.676	0.827	0.933	0.800	0.809
Hungary	487	51.6	46.24 (14.67)	5.55 (1.81)	65.4	0.688	0.790	0.779	0.850	0.777
Japan	443	49.9	49.94 (15.27)	5.34 (2.09)	60.0	0.656	0.823	0.917	0.670	0.766
Russia	487	53.0	43.72 (13.19)	5.52 (2.11)	72.0	0.708	0.770	0.797	0.790	0.766
China	497	48.2	39.24 (12.70)	6.33 (1.75)	84.4	0.682	0.722	0.808	0.730	0.735
Brazil	481	47.0	42.20 (15.81)	6.35 (2.10)	54.8	0.695	0.734	0.610	0.780	0.705
Turkey	484	49.4	40.42 (12.64)	5.96 (2.06)	65.2	0.628	0.693	0.728	0.760	0.702
Indonesia	480	49.4	39.24 (11.99)	6.96 (1.79)	70.6	0.688	0.707	0.556	0.620	0.643
Saudi	462	39.8	36.71 (10.74)	6.52 (2.24)	77.6	0.590	0.703	0.753	0.410	0.614
India	480	48.8	37.61 (13.84)	7.30 (1.96)	66.4	0.625	0.597	0.510	0.570	0.575
Egypt	481	48.0	38.11 (12.28)	6.16 (1.96)	70.8	0.637	0.627	0.557	0.440	0.565

Note: n represents the number of respondents with available data. GII and SIGI were reverse‐coded so higher scores indicate higher equality for all indices in the table.

Abbreviations: GGGI = Global Gender Gap Index; GII = Gender Inequality Index; SIGI = Social Institutions and Gender Index; WPS = Women, Peace, and Security.

TABLE 2.

Gender‐related attitudes questions and latent factor loadings.

	Est. (SE)	p
Men's superiority
Women should worry less about their rights and more about becoming good wives and mothers	1.00 (–)
A woman should not expect to go to exactly the same places or to have quite the same freedom of action as a man	0.91 (0.02)	<0.001
In general, the father should have greater authority than the mother in the bringing up of children	0.93 (0.02)	<0.001
The intellectual leadership of a community should be largely in the hands of men	1.01 (0.02)	<0.001
Traditional domestic roles
Working women put a strain on the family	1.00 (–)
Bringing up children is the most important job for a woman	2.54 (0.10)	<0.001
Domestic chores should be shared between husband and wife*	2.36 (0.02)	<0.001
Vocational anti‐egalitarianism
Women should work even if they are not in need economically*	1.00 (–)
There will be much social progress and development when more women work*	2.03 (0.23)	<0.001
Denial of discrimination against women
Discrimination against women is no longer a problem	1.00 (–)
It is rare to see women treated in a sexist manner on television	0.77 (0.02)	<0.001
Belief in sexism shift
If anything, men are more discriminated against than women these days	1.00 (–)	<0.001
Giving women more rights often requires taking away men's rights	0.93 (0.02)	<0.001
Under the guise of equality for women, men are actually being discriminated against	1.00 (0.02)	<0.001

Note: * indicates that item was reverse scored.

The measures of men's superiority and traditional domestic role endorsement are classic assessments of old‐fashioned sexism (in the public and private spheres, respectively), whereas denial of discrimination and belief in a sexism shift are considered more modern forms of sexism. The measure of vocational anti‐egalitarianism is less commonly employed in social science research, but we included it here because we wanted to extend the scope of the attitudes assessed, given the wide‐ranging differences in gender experiences across the countries included in our sample.

We expect that gender‐related attitude endorsement will be affected by gender equality, respondent gender, and their interaction. First, we hypothesize that the endorsement of sexist attitudes will be inversely related to country‐level gender equality (hypothesis 1a), but that the relative mean levels of endorsement of the different attitudes will vary as a function of gender equality. Specifically, we expect that old‐fashioned beliefs (men's superiority and traditional domestic role prescriptions) will be endorsed to a lesser extent than modern sexist beliefs (denial of discrimination against women and belief in sexism shift) in countries with high levels of gender equality (hypothesis 1b), but to roughly the same extent in countries with low levels of gender equality (hypothesis 1c). Second, we predict that men's attitudes will be more sexist compared to women's across measures (hypothesis 2a), but that the magnitude of this gender gap will vary as a function of belief type and country‐level gender equality. More specifically, we expect that endorsement of old‐fashioned sexism beliefs will be inversely related to gender equality roughly to the same extent for men and women (hypothesis 2b), whereas women in high (vs. low) gender‐equal societies will be especially likely to be reject modern forms of sexism (hypothesis 2c). Thus, we predict that gender polarization will be largest for modern sexist attitudes in places with relatively high gender equality. We should note that, although we center many of our findings around gender polarization (i.e., the gap between women and men in terms of attitudes), all of our results are based on slope estimates for men and women, not difference scores, which makes our findings less susceptible to incorrect or misleading inferences [76].

2. Methods

2.1. Participants

Individual‐level variables were collected in 2021 by Ipsos (https://www.ipsos.com/en-us), a multinational marketing and public opinion research firm. The countries included in the survey were based on availability through Ipsos within our budget, with aims of maximizing variation in both country‐level gender equality and geographical representation (see Table 1). The sample size (approximately 500 respondents per country) was determined by Ipsos. According to Ipsos, in countries with high levels of internet penetration (i.e., Hungary, Japan, Sweden, the United Kingdom, and the United States), the samples are nationally representative, whereas the samples in the other countries are more educated and/or urban than the general population. (Specifically, in India and Indonesia, the samples are representative of the urban population; in Brazil, Russia, and South Korea, the samples are representative of the population with primary education or higher; in Egypt, Saudi Arabia, and Turkey, they are representative of the population with secondary education or higher.) Table 1 summarizes demographics of the current samples.

2.2. Gender‐Related Attitudes Measures

As part of a larger survey (see https://osf.io/qjysp), respondents answered 14 questions assessing gender attitudes (see Table 2), with responses recorded on a scale from 1 (strongly disagree) to 7 (strongly agree). Four items were taken from the ATW scale [33], assessing belief in men's superiority (α = 0.85, ω_{t o t a l} = 0.86). Three items were taken from scales assessing support for traditional gender roles in the domestic sphere [45, 77] (α = 0.54, ω_{t o t a l} = 0.60). We included two additional items from this scale, both assessing vocational egalitarianism, or the belief that women can achieve independence through working [45, 54] (α = 0.62, r = 0.46). This scale was reverse‐coded so that it reflects vocational anti‐egalitarianism, aligning its semantic direction with the other sexism measures. Two items were taken from the Modern Sexism Scale. (One additional item from the Modern Sexism Scale was included in the survey, namely, “It is easy to understand why women's groups are concerned about societal limitations of women's opportunities.” This item did not hang together with the other two items, seemingly because it was difficult to understand. Thus, we omitted it from the analyses.) [57], assessing denial of discrimination against women (α = 0.65, r = 0.49). We used three items from the Belief in Sexism Shift Scale [36] (α = 0.80, ω_{t o t a l} = 0.81), which assesses the perceptions of discrimination against men. A confirmatory factor analysis supported the 5‐factor categorization, showing acceptable fit, CFI = 0.951; RMSEA = 0.066, 95% CI [0.063, 0.069]; SRMSR = 0.051; see Table 2 for factor loadings. Country‐by‐country correlations between these measures are provided in Table S1.

2.3. Demographic Predictors

Respondent gender was dummy coded (0 = male; 1 = female). We statistically adjusted for demographic variables that tend to be related to gender attitudes, including age, socioeconomic status, and marital status. Age was grand‐mean centered and divided by 10. Ipsos did not provide data on respondents’ education or income for all countries, so we used a measure of subjective socioeconomic status, where respondents are asked to rate their social standing relative to other people in their country, where 0 = “worst off” and 10 = “best off”; this was centered around the grand mean. Finally, marital status was measured in country‐specific ways, and data from Egypt, Japan, and Saudi Arabia did not separately record divorced versus widowed. Thus, in our models, we adjusted for marital status with two dummy codes—single (vs. married) and divorced/separated/widowed (vs. married).

2.4. Country‐Level Gender Equality

We considered several measures of country‐level gender equality (see Table 1). The Global Gender Gap Index (GGGI) was developed by the World Economic Forum [19], and evaluates gender parity across four key dimensions: economic participation and opportunity; educational attainment; health and survival; and political empowerment. Scores range from 0 to 1, where 1 indicates perfect gender parity. The Women, Peace, and Security (WPS) index is published by the Georgetown Institute for WPS [78], and assesses nations on inclusion (women's economic, social, and political status), justice (formal and informal sources of gender discrimination), and security (lack of household, community, and societal gender‐based violence). Like the GGGI, the WPS ranges from 0 to 1, with 1 indicating women's status is perfectly equal with men. The Gender Inequality Index (GII) is a composite metric published by the United Nations Development Programme [79] and reflects gender‐based disadvantage in reproductive health, empowerment, and the labor market. It ranges from 0 (perfect equality) to 1 (perfect inequality); we reversed this so that it is comparable to the GGGI and WPI, such that higher numbers indicate more equality. Finally, we used the Social Institutions and Gender Index (SIGI) [80], which assesses discrimination toward women and girls embedded in formal and informal laws and social norms across four domains: discrimination in the family; restricted physical integrity; restricted access to resources; and restricted civil liberties. The SIGI ranges from 0 (0% of women experiencing discrimination) to 100 (100% of women experiencing discrimination). In order to make this comparable to the other indices, we rescaled it (so it ranged from 0 to 1) and reverse scored it (so that higher numbers indicate more equality). Because our survey data were collected in 2021, we used the gender inequality indices from that same year for GGGI, WPI, and GII. The SIGI was computed in 2019 and 2023; we used the 2023 index because the 2019 index was not available for all the countries included in our survey. Because these measures were highly correlated (α = 0.87, ω_{t o t a l} = 0.86), we created a composite Gender Equality index by averaging them to reduce measurement error associated with any single measure. The main analyses relied on this index.

2.5. Preliminary Analyses

2.5.1. Sensitivity Analysis

A Monte Carlo simulation–based sensitivity analysis [81] was conducted to estimate statistical power for detecting cross‐level interaction effects (gender × gender equality). Simulations were based on the empirical design (14 countries, approximately 500 individuals per country), and parameter estimates were derived from the fitted models. Gender equality was simulated using its observed standard deviation (SD = 0.099), and power was estimated across a range of interaction effect sizes using 500 replications per condition.

Results indicated that approximately 80% power was achieved for cross‐level interaction effects of b ≈ |3.0| for denial of discrimination and b ≈ |2.5| for belief in sexism shift (corresponding to approximately 0.30 and 0.25 points in the outcome per one standard deviation increase in gender equality, respectively). The observed interaction effects (b = 2.67 for denial of discrimination; b = 1.94 for belief in sexism shift) were therefore somewhat below these thresholds, indicating moderate—but not high—statistical power. The observed interaction effects for the old‐fashioned sexism measures were substantially smaller (bs = 0.16–0.44) than the effects required to achieve 80% power (b ≈ 1.0–1.8). However, interactions were not predicted for these outcomes. The observed interaction effect for vocational anti‐egalitarianism (b = 1.78) fell slightly below the b ≈ 2.0 threshold for 80% power, but we had no prediction regarding this outcome.

Because the models were underpowered to detect cross‐level interactions, Figures 2 and 3 show the country‐specific gender gaps in attitudes alongside the estimated interaction effects. These country‐specific estimates are based on large within‐country samples (approximately 500 respondents per country) and are therefore well powered to detect gender differences within countries. See the Supporting Information for country‐specific analyses, including the effect sizes for the gender differences (Table S2) and regression estimates (Table S3).

FIGURE 2.

Estimates from a multilevel model (top panel) and from country‐specific regression models (bottom panel) of endorsement of old‐fashioned forms of sexism, as a function of country‐level gender equality and respondent gender.

FIGURE 3.

Estimates from a multilevel model (top panel) and from country‐specific regression models (bottom panel) of endorsement of modern forms of sexism, as a function of country‐level gender equality and respondent gender.

2.5.2. Measurement Invariance

Cross‐country analyses of attitudinal scales pose a well‐known challenge—the issue of measurement invariance. Investigations involving many groups often violate the assumption of scalar invariance (i.e., equality of item intercepts), a conventional criterion that permits valid mean comparisons. However, violations of (scalar) invariance are not strictly diagnostic of measurement bias: They may arise from theoretically expected sources; see [82] for an extensive discussion. In particular, groups with highly homogeneous responses—such as those in which attitudes are tightly shared—tend to distort intercept estimates and lead to apparent non‐invariance. We therefore conducted exploratory analyses and multigroup confirmatory factor analyses (see the Table S4). We found acceptable fit for configural and metric invariance for the composite score (which sums of all the sub‐scales). Scalar invariance was not fully supported, largely due to groups with low individual variability (e.g., Egypt and Saudi Arabia), indicative of culturally homogeneous response patterns. We conclude that these deviations do not compromise the current interpretation and refer readers to the intercorrelations among outcome measures (Table S1).

3. Results

We conducted a 3‐level model with the five gender‐related attitudes (men's superiority, traditional domestic roles, vocational anti‐egalitarianism, denial of discrimination against women, and belief in sexism shift) nested in respondents, and respondents nested in countries. We included random effects for the intercept and the slope for gender. This analytic strategy allowed us to compare the cross‐level interactions between gender equality, gender, and attitude type, to examine: which gender‐related attitudes are more or less likely to be endorsed across contexts (hypothesis 1a–c; “H1a‐c”), and whether men's and women's endorsement of each attitude varies as a function of gender equality (hypothesis 2a–c; “H2a‐c”).

Table 3 reports the results from the multilevel model predicting the gender‐related attitude with attitude type (four dummy codes for the comparison of five attitudes), gender, gender equality, and all the two‐ and three‐way interactions, adjusting for age, subjective socioeconomic status, and marital status (two dummy codes) and their interactions with gender. To facilitate the presentation of the results, we report the unstandardized coefficients for the model using each gender‐related attitude as the reference group.

TABLE 3.

Unstandardized coefficients from a multilevel model using each of the five attitude types as the reference group.

	1: Men's superiority		2: Traditional gender roles		3: Vocational anti‐egalitarianism		4: Denial of discrimination against women		5: Belief in sexism shift
	b (SE)	p	b (SE)	p	b (SE)	p	b (SE)	p	b (SE)	p
Intercept	3.55 (0.12)	<0.001	3.33 (0.12)	<0.001	3.17 (0.12)	<0.001	3.84 (0.12)	<0.001	3.82 (0.12)	<0.001
Gender equality (GE)	−6.75 (1.20)	<0.001	−3.60 (1.20)	0.011	0.55 (1.21)	0.655	−2.30 (1.20)	0.080	−2.65 (1.20)	0.047
Female (vs. male)	−0.61 (0.05)	<0.001	−0.33(0.05)	<0.001	−0.18(0.05)	0.001	−0.47(0.05)	<0.001	−0.76(0.05)	<0.001
GE  $\times$ female	−0.44 (0.49)	0.375	−0.16 (0.49)	0.744	1.78 (0.49)	0.001	−2.67(0.49)	<0.001	−1.94 (0.49)	0.001
1			0.22 (0.03)	<0.001	0.37 (0.03)	<0.001	−0.29 (0.03)	<0.001	−0.28 (0.03)	<0.001
2	−0.22 (0.03)	<0.001			0.16 (0.03)	<0.001	−0.51 (0.03)	<0.001	−0.49 (0.03)	<0.001
3	−0.37 (0.03)	<0.001	−0.16 (0.03)	<0.001			−0.67 (0.03)		−0.65 (0.03)	<0.001
4	0.29 (0.03)	<0.001	0.51 (0.03)	<0.001	0.67 (0.03)	<0.001			0.01 (0.03)	0.622
5	0.28 (0.03)	<0.001	0.49 (0.03)	<0.001	0.65 (0.03)	<0.001	−0.01 (0.03)	0.622
GE $\times$1			−3.14 (0.27)	<0.001	−7.30 (0.28)	<0.001	−4.45 (0.27)	<0.001	−4.09 (0.27)	<0.001
GE $\times$ 2	3.14 (0.27)	<0.001			−4.16 (0.28)	<0.001	−1.31 (0.27)	<0.001	−0.95 (0.27)	<0.001
GE $\times$ 3	7.30 (0.28)	<0.001	4.16 (0.28)	<0.001			2.85 (0.28)	<0.001	3.20 (0.28)	<0.001
GE $\times$ 4	4.45 (0.27)	<0.001	1.31 (0.27)	<0.001	−2.85 (0.28)	<0.001			0.36 (0.27)	0.193
GE $\times$ 5	4.09 (0.27)	<0.001	0.95 (0.27)	0.005	−3.20 (0.28)	<0.001	−0.36 (0.27)	0.193
Female $\times$ 1			−0.28 (0.04)	<0.001	−0.43 (0.04)	<0.001	−0.14 (0.04)	<0.001	0.15 (0.04)	<0.001
Female $\times$ 2	0.28 (0.04)	<0.001			−0.15 (0.04)	0.001	0.14 (0.04)	<0.001	0.43 (0.04)	<0.001
Female $\times$ 3	0.43 (0.04)	<0.001	0.15 (0.04)	0.001			0.29 (0.04)	<0.001	0.57 (0.04)	<0.001
Female $\times$ 4	0.14 (0.04)	<0.001	−0.14 (0.04)	0.003	−0.28 (0.04)	<0.001			0.29 (0.04)	<0.001
Female $\times$ 5	−0.15 (0.04)	<0.001	−0.43 (0.04)	<0.001	−0.57 (0.04)	<0.001	−0.29 (0.04)	<0.001
GE $\times$ female $\times$ 1			−0.28 (0.39)	0.473	−2.22 (0.40)	<0.001	2.23 (0.39)	<0.001	1.49 (0.39)	<0.001
GE $\times$ female $\times$ 2	0.28 (0.39)	0.473			−1.94 (0.40)	<0.001	2.51 (0.39)	<0.001	1.77 (0.39)	<0.001
GE $\times$ female $\times$ 3	2.22 (0.40)	<0.001	1.94 (0.40)	<0.001			4.45 (0.40)	<0.001	3.72 (0.39)	<0.001
GE $\times$ female $\times$ 4	−2.23 (0.39)	<0.001	−2.51 (0.40)	<0.001	−4.45 (0.40)	<0.001			−0.73 (0.39)	0.062
GE $\times$ female $\times$ 5	−1.49 (0.39)	0.003	−1.77 (0.40)	<0.001	−3.72 (0.40)	<0.001	0.73 (0.39)	0.062

Note: Each dummy variable for attitude type coded the target type as the reference category (=0), and the other types as comparison groups (=1). Age, subjective SES, marital status, and their interactions with gender are omitted for the presentation purpose.

3.1. Comparing Attitude Endorsement Across Contexts

Looking across the columns in Table 3, country‐level gender inequality is negatively related to endorsement of men's superiority and traditional domestic role attitudes (ps < 0.012), unrelated to vocational anti‐egalitarianism (p = 0.655), and negatively (albeit not significantly) associated with denial of discrimination (p = 0.080) and belief in sexism shift (p = 0.047). These findings provide modest support for H1a.

The main effects of each attitude type (rows labeled 1–5) show the relative levels of endorsement across attitudes. For instance, in the first column of Table 3 (where men's superiority beliefs are the reference group), the negative main effects listed in Rows 2 and 3 show that endorsement of men's superiority is higher than traditional domestic roles and vocational anti‐egalitarianism endorsement; the positive main effects for the last two attitudes (Rows 4 and 5) show that men's superiority endorsement is lower than denial of discrimination and belief in sexism shift. As seen in Table 3, almost all of the main effects were qualified by gender equality, indicating that the relative mean levels of endorsement differ across contexts, with the exception of the two modern sexism measures (see Figure 1).

FIGURE 1.

Mean levels of endorsement of gender‐related attitudes in countries with low (−1 SD) and high (+1 SD) gender equality.

Using simple slopes analyses, we examined the relative attitude endorsement for low (−1 SD) and high (+1 SD) gender‐equal places, collapsing across gender (i.e., probing gender equality × attitude type interactions). Results show that, in countries with high levels of gender equality (+1 SD), denial of discrimination against women was higher than men's superiority, b = |0.69|, SE = 0.03, p < 0.001, traditional gender roles, b = |0.44|, SE = 0.03, p < 0.001, and belief in sexism shift, b = |0.16|, SE = 0.03, p < 0.001, but not significantly different from vocational anti‐egalitarianism endorsement, b = |0.02|, SE = 0.03, p = 0.589. Belief in sexism shift endorsement was significantly lower than vocational anti‐egalitarianism, b = |0.14|, SE = 0.03, p < 0.001, but higher than both old‐fashioned forms of sexism, including men's superiority, b = |0.54|, SE = 0.03, p < 0.001, and traditional domestic roles, b = |0.28|, SE = 0.03, p < 0.001. Endorsement of men's superiority was lower than endorsement of traditional domestic roles, b = |0.25|, SE = 0.03, p < 0.001, and both of these old‐fashioned forms of sexism were less likely to be endorsed compared to all others (ps < 0.001). Thus, in line with H1b, modern (vs. old‐fashioned) forms of sexism are more strongly endorsed in high gender‐equal societies.

In low gender‐equal countries (−1 SD), endorsement of men's superiority and denial of discrimination did not differ, b = |0.02|, SE = 0.03, p = 0.456, and both were significantly higher than traditional domestic roles, vocational anti‐egalitarianism, and belief in sexism shift, ps < 0.001. Traditional domestic roles endorsement was higher than vocational anti‐egalitarianism, b = |0.59|, SE = 0.03, p < 0.001, but lower than the modern forms of sexism, including denial of discrimination, b = |0.43|, SE = 0.03, p < 0.001, and belief in sexism shift, b = |.28|, SE = 0.03, p < 0.001. These findings provide only partial support for H1c.

3.2. Gender and Gender Equality as Predictors of Each Gender‐Related Attitude

Men endorsed all of the sexist attitudes more strongly than women, bs range from 0.18–0.76, ps < 0.001, supporting H2a (see Table 3). Further, for both old‐fashioned forms of sexism (see Table 3, Columns 1–2), there were no significant interactions between gender and gender equality, supporting H2b (see Figure 2).

For vocational anti‐egalitarianism (for which we did not have a hypothesis), a significant interaction between gender and gender equality emerged, such that in relatively low gender‐equal places (−1 SD), women's endorsement was significantly lower than men's, b = −0.36, SE = 0.07, p < 0.001, whereas in places with high equality (+1 SD), there was no gender gap in endorsement, b = −0.01, SE = 0.07, p = 0.940. Looking at this interaction another way, men's attitudes were unrelated to gender equality, b = 0.55, SE = 1.21, p = 0.647, whereas women's were positively related, b = 2.33, SE = 1.15, p = 0.043.

For modern forms of sexism (denial of discrimination against women and belief in sexism shift), the main effects of gender were both qualified by significant negative interactions with gender equality (see Figure 3). Simple slopes analyses confirmed H2c, namely, that the gender gap in modern sexist attitudes was smaller in countries with low gender equality (b = −0.20, SE = 0.07, p = 0.003 for denial of discrimination; b = −0.56, SE = 0.07, p < 0.001 for belief in sexism shift) compared to high gender equality (b = −0.74, SE = 0.07, p < 0.001 for denial of discrimination; b = −0.95, SE = 0.07, p < 0.001 for belief in sexism shift). Looking at the interaction another way, both measures of modern sexism were more strongly associated with gender equality among women (b = −4.97, SE = 1.15, p < 0.001 for denial of discrimination, and b = −4.59, SE = 1.15, p < 0.001 for belief in sexism shift) compared to men (b = −2.30, SE = 1.21, p = 0.057 for denial of discrimination, and b = −2.65, SE = 1.20, p = 0.028 for belief in sexism shift).

The three‐way interactions across attitude types (last five rows of Table 3) show that the pattern of results between the two old‐fashioned measures did not differ from each other, b = 0.28, SE = 0.39, p = 0.473, but did differ from the other sexism measures (ps < 0.001). Similarly, there was no significant three‐way interaction between the two modern sexism measures, b = −0.73, SE = 0.39, p = 0.062, but the pattern for the two measures differed from that of the other measures (ps < 0.001).

3.3. Comparing Gender Polarization Across Contexts

In our final analysis, we directly compared the magnitude of gender polarization across the different attitude types for places with low and high gender equality. As seen in Figure 4, the largest gap in men's and women's attitudes (men's minus women's endorsement) was for belief in sexism shift among people in high gender‐equal countries (b = −0.95, SE = 0.07, p < 0.001). Pairwise comparisons confirmed that this gender gap was significantly larger than the gaps on the other attitudes in both high and low equal contexts (ps < 0.001). In low gender‐equal countries, the gender gaps on endorsement of men's superiority (b = −0.57, SE = 0.07, p < 0.001) and belief in sexism shift (b = −0.56, SE = 0.07, p < 0.001) did not significantly differ, M_D  = 0.00, SE = 0.05, p = 0.969, whereas all other gender gaps were significantly different from each other, ps < 0.001. In countries with relatively high gender equality, the gender gap on men's superiority (b = −0.65, SE = 0.07, p < 0.001) did not significantly differ from the gap on denial of discrimination (b = −0.73, SE = 0.07, p < 0.001), M_D  = 0.08, SE = 0.06, p = 0.140, whereas all other gender gaps were significantly different from each other (ps < 0.001).

FIGURE 4.

Estimated gender polarization (men's endorsement minus women's) on gender‐related attitudes, as a function of low (−1 SD) and high (+1 SD) country‐level gender equality.

4. Discussion

The present research examined gender polarization in gender‐related attitudes across 14 diverse societies, focusing on how the type of attitude (old‐fashioned vs. modern sexism) and the level of societal gender equality jointly predict the magnitude of gender differences. Our analyses reveal several overarching patterns regarding the overall rates of gender‐related attitude endorsement, as well as attitude polarization between women and men across societies.

First, consistent with prior research, endorsement of sexist attitudes was inversely associated with country‐level gender equality: individuals in more (vs. less) gender‐equal societies were less likely to endorse sexist beliefs overall (supporting H1a). However, this pattern was not uniform across attitude types. Old‐fashioned forms of sexism—such as endorsement of men's superiority and traditional domestic roles—showed clear and robust negative associations with gender equality. In contrast, modern forms of sexism—namely, denial of discrimination against women and belief in a sexism shift favoring women—were not robustly related to country‐level gender equality, suggesting that these ideologies persist even in relatively egalitarian contexts. Indeed, as predicted (H1b), results also showed that modern forms of sexism were more strongly endorsed compared to old‐fashioned sexism in places with relatively high gender equality. Although we predicted that rates of endorsement across attitude types would not vary in countries with relatively low gender equality (H1c), results showed some mean differences—for instance, that men's superiority, denial of discrimination, and belief in a sexism shift were more strongly endorsed than traditional domestic roles.

Second, women consistently reported lower sexism than men (supporting H2a), but the magnitude of this gender gap varied across belief types and societal context. For old‐fashioned sexism, gender differences were relatively stable across levels of gender equality, consistent with H2b. For modern sexist attitudes, however, gender gaps were more pronounced in places high (vs. low) on gender equality (supporting H2c). This pattern was driven primarily by women's attitudes: women in more (vs. less) gender‐equal contexts were especially likely to reject beliefs denying discrimination or portraying men as the disadvantaged group, whereas men's attitudes were comparatively less sensitive to variation in gender equality.

Because we used a three‐level design, we were able to further test how the pattern of results varied across gender equality, gender, and attitude type (i.e., the three‐way interactions in the model). Overall, these interaction results confirmed that the two old‐fashioned forms of sexism exhibited a common pattern, as did the two modern forms, and that these two sets of attitudes differed systematically from one another, consistent with theoretical expectations. Taken together, the greatest polarization across all attitudes and contexts was the difference between women and men in high gender‐equal countries on a modern form of sexism—namely, the belief that sexism has shifted, and that women are now the advantaged group—where women scored a near full one‐scale point lower than men on average.

We also examined a measure we termed vocational anti‐egalitarianism, which assessed opposition to women achieving independence through participation in the workforce. Conceptually, this construct captures a form of gender‐related attitude that is meaningful primarily in highly gender‐segregated contexts, where women's employment remains socially contested or symbolically charged. In such settings, the endorsement of women's economic participation represents a genuine stance on gender equality, as working outside the home challenges established gender norms. In more gender‐equal societies, by contrast, women's employment has long been the normative status quo; thus, these items may no longer function as meaningful indicators of gender ideology. For example, disagreement with statements such as “women should work even if they are not in need economically” may reflect general beliefs about work and leisure, rather than opposition to women's employment per se. In line with this reasoning, our results showed gender polarization on vocational anti‐egalitarianism only in countries with lower levels of gender equality—precisely those contexts where encouraging women's participation in the workforce constitutes a progressive, equality‐promoting stance. The inclusion of this measure in the current work highlights the importance of contextualizing attitudinal measures: beliefs that index gender ideology in one sociocultural setting may be interpreted in entirely different ways in another.

Indeed, one key strength of this study lies in its inclusion of countries that vary widely in cultural, economic, and political conditions—including several that are rarely represented in psychological research (e.g., Egypt, Indonesia, and Saudi Arabia). Across these diverse settings, we observed both universal patterns (e.g., lower sexism among women compared to men) and context‐specific nuances (e.g., the relevance of vocational egalitarianism in societies where women's labor participation remains contested). This highlights the importance of moving beyond Western‐centric frameworks when studying gender attitudes.

The findings also have practical importance for gender researchers. For example, endorsement of men's superiority—derived from the classic ATW scale developed in 1973— showed the largest correlation with national gender equality, with relatively high endorsement in less gender‐equal countries and low endorsement in more equal countries. Although this scale has fallen out of favor among (Western) researchers because many contemporary participants tend to eschew these overtly sexist statements [17, 36], the current work shows that this is not the case across the board.

We also found that a relatively new measure of sexism, belief in sexism shift [36]—which assesses the perception that men (not women) are the disadvantaged gender group in modern societies—was endorsed across the board, and even more strongly in countries with low (vs. high) gender equality. This is somewhat counterintuitive, as one might expect that beliefs about men's victimhood would not emerge unless gender equality is close to being actualized. In some of these countries—like Saudi Arabia and Indonesia—women still lag far behind men, but their societal status is undergoing rapid advancement. Thus, it seems plausible that rapid or visible changes in women's status, rather than absolute levels of equality, may fuel perceptions of status threat. Even in societies where women remain structurally disadvantaged, shifts toward greater inclusion may be interpreted as a loss of relative standing among men. This is in line with research showing that victimhood ideology among men is unrelated to their actual socioeconomic status, but positively associated with a perceived decline in status [62].

The findings from this research also contribute to theoretical debates on the social‐psychological consequences of gender equality. Contrary to popular narratives of a gender war driven by men's backlash, our pattern of results suggests that widening gender gaps in some contexts may be a function of women's increasingly progressive attitudes rather than a rise in men's conservatism. Although we cannot make any longitudinal (causal) claims with these cross‐sectional data, the finding that women's (but not men's) endorsement of modern sexist attitudes is inversely associated with gender equality is at least consistent with theories of awareness and recognition [68, 70], which suggest that as women gain social and economic empowerment, they may also become more perceptive of persistent structural inequities and more resistant to ideologies that deny them.

Another related explanation for why men and women show greater divergence on modern sexist attitudes in societies with high (vs. low) gender equality may be related to women's needs to cognitively cope with overt injustice in places where women's oppression is more normative. This idea aligns with research on system justification theory, which has demonstrated how the endorsement of beliefs that frame the system as fair (like the denial of gender discrimination) can serve a palliative function for members of disadvantaged groups, insofar as these beliefs are associated with better subjective well‐being, particularly when inequality is high [16, 83]. Thus, women living in less equal environments may face greater psychological costs when rejecting these types of sexist ideologies.

At the same time, the persistence of modern sexist attitudes among men in highly egalitarian contexts illustrates how these less direct forms of bias, which reframe inequality as stemming from women's choices or bias against men, persist even when overtly sexist ideas are deemed normatively unacceptable [36, 57]. The gender polarization on these more modern forms of sexism in countries with relatively high levels of gender equality may be fueling political polarization more broadly. In line with this, a recent study in Spain showed that women's (vs. men's) higher levels of rejection of modern sexism explained the gender difference in voting for a far‐right political party [84]. Although overt sexism continues to decline, divides around perceptions of fairness and victimhood may sustain political and cultural polarization. In this sense, these modern forms of sexism may constitute the new frontier of resistance to gender equality.

It is important to emphasize that the current research is cross‐sectional, and thus precludes causal inference about the directionality between gender equality and attitudes. Longitudinal designs could help clarify temporal dynamics of polarization. Indeed, it may be misguided to compare gender equality changes over time to cross‐national differences in gender equality. Given the global diffusion of social media narratives about anti‐male bias and wokeness, it seems conceivable that transnational information flows would be affecting gender‐related attitudes across distinct cultural contexts. If so, we would not expect that the trajectory that occurred in the Western world following the women's liberation movement of the 1960s would necessarily replicate in the same way in contemporary contexts.

This leads to a broader theoretical point that attitudes within a society are likely not only affected by local structures but are also embedded in a global and historical context. For instance, scholars have documented how in Islamic societies, many individuals are ambivalent toward Western gender relations, which they associate with sexualization of women, family dissolution, and moral decline [85, 86]. Further, gender equality initiatives can be perceived through a lens of Western imperialism or part of a colonial feminist agenda [87, 88]—the imposition of Western values with disregard for cultural traditions. It is thus conceivable that the determinants of sexist attitudes (or resistance to egalitarian attitudes) in any given society could be unique, dependent on certain intersectional frameworks.

Another important limitation to point out is that our measure of gender equality was based on national indices, which are clunky for several reasons. First, these country‐level measures obscure any within‐country heterogeneity in local norms and institutional practices. Second, although we combined multiple indices that aim to capture the prevalence of institutional and informal discrimination across multiple domains, which gives us a more reliable measure of country‐level equality, these indices may nonetheless fail to fully reflect psychological, cultural, or symbolic gender dynamics. For instance, a recent report by Gallup [89] found that in 2012, approximately 75% of American women and men under 45 years old agreed that “women in the United States are treated with respect and dignity”; by 2022, agreement with this statement dropped to 63% among men and, stunningly, to 41% among women. Countries, like Greece, Albania, and North Macedonia, showed similarly dramatic increases in gender polarization in perceived respect for women, whereas polarization was virtually eliminated over the course of these 12 years in other countries (like France, Germany, Estonia, and Saudi Arabia). The orthogonality of these patterns with national gender equality underscores how national indices are not well equipped to capture the social atmosphere—for example, perceptions of intergroup threat, backlash, or shifting norms. As such, two societies with similar index scores may differ substantially in how gender relations are experienced and perceived by their citizens.

A final limitation to mention is that our reliance on online panels, which overrepresented urban and educated populations in some countries, could potentially lead to underestimating overall sexism levels. Further, although we adjusted for age, socioeconomic status, and marital status and their interactions with gender in our models (see Table S3), we did not thoroughly explore those results. We should note, however, that there were only a few, inconsistent interactions between gender and age predicting any of the gender‐related attitudes, suggesting that the patterns we observed are not driven by any particular age group. (Age‐by‐gender interactions emerged in Japan, where age was positively correlated with denial of discrimination among women but unrelated among men; in South Korea, where age was negatively correlated with belief in sexism shift among men but uncorrelated among women; and in Turkey, where age was negatively correlated with men's superiority among men but uncorrelated among women. Detailed results are provided in Tables S3 and S3a.) Subjective socioeconomic status, by contrast, consistently interacted with gender in predicting both old‐fashioned and modern forms of sexism in both China and India, such that it was positively associated with attitude endorsement among women, but weakly or unrelated to men's attitudes. This could indicate that, in some places, women may embrace sexist ideologies at least partly due to economic reasons. More broadly, future research would benefit from a more thoughtful intersectional examination of how different social categories, in addition to gender, may affect people's gender‐related attitudes.

In sum, this research demonstrates that gender polarization in gender‐related attitudes depends on both attitude content and the societal context in which it is embedded. Although overt forms of sexism decline with increasing gender equality, more indirect forms of gender bias persist, especially among men. Thus, in the most egalitarian societies, men and women are especially likely to diverge on their perceptions of gender discrimination. Recognizing and addressing the more nuanced beliefs and perceptions that continue to sustain inequality is thus a critical step in reducing gender polarization, and ultimately, in progressing toward a more gender‐equal world.

Author Contributions

Kodai Kusano and Jaime L. Napier developed the research concept. Kodai Kusano wrote the initial draft of the article. Kodai Kusano and Jaime L. Napier performed data analyses. Jaime L. Napier secured the funding for the project, oversaw data curation, and wrote the final draft. All authors contributed to the interpretation of the results and approved the final version of the manuscript.

The survey was approved by the Institutional Review Board at NYU Abu Dhabi (Protocol number: Legacy‐107‐2018).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material: nyas70373‐sup‐0001‐SuppMat.docx

Data Availability Statement

The data and R scripts used in this article are openly available at https://osf.io/du39f.