Mosaic of beliefs: Comparing gender ideology in China across generation, geography, and gender

Modernization accounts of cultural change hold that economic development drives liberalization of social values, including gender beliefs. Some comparative gender scholarship suggests that societal affluence is often accompanied by the growth of gender-essentialist beliefs, and that these beliefs coexist comfortably alongside gender-egalitarian values. The multidimensional conceptualization of gender ideology that is required to assess these competing claims has been applied so far mostly to Western societies. China is an obvious case for extending knowledge of these relationships, given its rapid economic growth and its recent history of state-imposed gender-egalitarian discourses. Applying latent class analysis to the Chinese General Social Survey (2010–2017), this study links different tenets of gender ideology in China to temporally and spatially specific histories and gendered interests. The results show that the relative importance of modernization and gender accounts depends on the generational, regional, and gendered identities being examined. Unlike in the West, moreover, egalitarian and essentialist beliefs do not always coincide in contemporary China. The friction between these beliefs reflects the resilience of male-primacy ideology.

The multidimensional approach that is required to assess different theoretical claims on gender ideology development has mostly been applied to post-industrial Western societies (e.g.Knight and Brinton, 2017;Scarborough et al., 2019).The rapidly modernizing and formerly socialist Chinese society provides an optimal case to test how different tenets of gender ideology evolve.Qualitative work on contemporary gender culture in China delineates complex trends, including market logics, socialist feminist legacies, Confucian patriarchy, and the authoritarian state, that crosscut with each other to set and shape gender norms and discourses (Evans, 2021;Ji, 2017;Song, 2011;Wang, 2017).These forces together constitute a "patchy patriarchy" (Evans, 2007(Evans, , 2021) ) or "mosaic temporality" (Ji, 2017) in which contemporary Chinese navigate their gender subjectivities and behaviors.
How is this complex mosaic of gender culture mirrored at the public attitudinal level?There is competing evidence to support both "modernization progress" and "post-socialist regression" accounts.On one hand, empirical research finds that higher levels of development are linked with more liberal values about work and individual freedom in China (Cao, 2020;Pan and Xu, 2018).On the other hand, focusing on the temporal shift from the 1990s to the 2010s, concerned feminists show that Chinese people's attitudes toward women's domestic roles regress toward traditionalism with marketization sweeping away socialist egalitarian values and practices (Gu, 2014;Yang et al., 2014).I argue that the various drivers of gender ideology imply that both modernization-induced liberalization and post-socialist traditionalization could be unfolding simultaneously in China.A multidimensional approach to systematically map and compare gender ideology in contemporary China is in dire need.
Using nationally representative survey data, this study explores the multidimensional structure of gender ideology that has born out of recent Chinese history.I argue that this multidimensionality is linked to historical and socioeconomic features of political generations, uneven development across the urban-rural divide, and gendered interests.I leverage five waves of Chinese national surveys from 2010 to 2017 and use latent class analysis (LCA) to map out how different types of gender beliefs cluster.Rather than pointing to a single narrative, this study reveals five ideological profiles that blend varied positions on gender essentialism, male primacy, and beliefs in women's employment.Results point to the predominance of essentialist gender beliefs in China that are compatible with support for women's employment but haven't moved away from male primacy.I further show how the structure of gender beliefs varies along three intersecting axes-political generation, the urban-rural regional divide, and gender status.Generation in this study refers to people who were affected by distinctive gender politics that characterize five periods in recent Chinese history-the early socialist era, the Cultural Revolution, the early reform, the radical reform, and the post-reform periods.As the identity formation of a generation is not only age-dependent but also place-specific (Griffin, 2004), I highlight the intersection between generation and the urban-rural divide in assessing the imprints of temporally and spatially specific histories.Given men's and women's divergent interests in achieving gender equality (Bolzendahl and Myers, 2004), I explore how gender crosscuts with political generation in shaping variation of gender beliefs.Results show that the relative explanatory power of modernization and feminist "post-socialist regression" accounts depends on the generational, regional, and gendered identities being examined.

Egalitarianism, essentialism, and the multidimensionality of gender ideology
After waves of women's movements in the West, substantial gaps exist between gender equality outcomes in families and workplaces, with women's incorporation in the public spheres outpacing men's participation at home (England et al., 2020).A similar divergence in attitudes is well-established, with greater support for equal opportunities in work and politics than for de-gendering of private lives (Cotter et al., 2011;Pepin and Cotter, 2018).This gender-equality paradox leads scholars to propose a cultural framework of "egalitarian essentialism," an ideology combining feminist equality ideals and beliefs in fundamental gender traits that are horizontally different but appear to be vertically equal (Charles, 2011;Charles and Grusky, 2004;Cotter et al., 2011).It argues that the post-industrial West has gradually moved away from male primacy, but a more thorough penetration of gender-egalitarianism is held back by prevalent beliefs in men's and women's essentialist difference.The compatibility of egalitarianism and essentialism propels researchers to investigate the multidimensional nature of gender ideology.
One way to address multidimensionality is by comparing beliefs about equality in different institutions notably family and the labor market (e.g.Pepin and Cotter, 2018).But this does not tell us whether and how egalitarianism and essentialism covary.Another approach focuses on how individuals' positions on varied gender issues cluster and form belief patterns (e.g.Knight and Brinton, 2017;Scarborough et al., 2019).This line of research, generally based on LCA, identifies gender-attitude classes that cannot be subsumed into a single framework or be captured by "egalitarian essentialism."For instance, in some East European and East Asian countries, a prevalent ideology prescribes women's labor market participation and their primary roles in the family (Brinton and Lee, 2016).This normative pro-work belief system is weak in the United States and Western Europe where espousal of individual autonomy goes against the requirement that women should work.But due to limited survey measures, existing research generally relies on measures of motherhood and the feminine self and attitudes toward working women to tap gender essentialism and excludes broader beliefs in gendered traits and capacities.As I show later, the Chinese dataset I use goes beyond this limit.

Gender attitudes in China
A multidimensional conceptualization is particularly important for understanding gender attitudes in China.Since the market reforms introduced in the late 1970s, Chinese society has witnessed an increasing separation of the public and private spheres and deterioration of gender inequalities (Ji et al., 2017;Song, 2011).The privatization of care strengthens ideologies of family solidarity and women's domesticity, which were briefly attacked by socialist collectivization-but to little avail (Evans, 2007;Hershatter, 2014).Institutional changes along with cultural backlash against socialist feminism together contribute to a widespread rise of gender-essentialism that emphasizes the biology of sex differences (Ji et al., 2017).Paradoxically, women's paid employment, which enjoys natural legitimacy that dates to the socialist years, becomes even more necessary for most families in the market economy.Meanwhile, belief in male primacy looms large.In the domestic arena, ethnographic research finds that female family members largely adhere to the authority of male ones (Evans, 2007).In the public arena, employers reject women applicants based on assumptions of women's lack of work commitment due to domestic duties and some blatantly post male-only job advertisements.More tellingly, patriarchal values undergird Chinese courts divorce adjudication that disproportionately denies petitions from women plaintiffs despite evidence of domestic violence (Michelson, 2019).
Echoing the complex ideological landscape, research on gender attitudes in China reveals three distinctive trends based on population average of survey measures.First, attitudes toward gender equality in work, politics, and education become more liberal across time and cohorts (Qian and Li, 2020;Shu and Zhu, 2012).Second, evidence from three decennial cross-sectional surveys from 1990 to 2010 suggests rising emphases on women's domestic roles (Gu, 2014;Xu, 2016;Yang et al., 2014).Third, support of women's dual roles of caregiver and wage-earner has been continuously strong and is less subject to either period or cohort variation (Shu and Zhu, 2012).
The dual beliefs in gender equality in the public sphere and women's domesticity lead recent scholars to apply an "egalitarian essentialism" descriptor to the Chinese context, and ascribe essentialism and egalitarianism to each domain of the private-public dichotomy (Qian and Li, 2020).The theoretical dichotomization fits well with China's post-socialist distinction of the two spheres but fails to address how people understand the relationship between earning and caring, and ignores other dimensions of gender attitudes.How widespread is this egalitarian-essentialist ideology identified in China as compared with other potential types of gender-belief systems?Is it conceptually equivalent to the Western frame in which essentialist beliefs about gender differences deny women's lower status and worthiness?If not, then what forms of gender-egalitarianism have taken shape and what forms of gender-traditionalism have survived during China's socioeconomic transformations?
To answer these questions, we need to attend to culturally relevant principles of gender beliefs.I rely on previous research to identify the following three salient aspects of gender attitudes in contemporary China.First, domesticity that traditionally defines Chinese "women's identity and raison d'être" survives the socialist state's advocacy of equal rights between men and women (nannü pingdeng) despite challenges from state feminists during the early Mao era (Evans, 2007;Ko, 1994: 117;Wang, 2017).This naturalized belief of gender difference is solidified by the unrelenting salience of family in post-socialist China where a state welfare system is lacking (Ji et al., 2017).Second, values of male authority and primacy remain largely intact during the socialist gender revolution and persistently countervail liberalization brought by socioeconomic modernization and global diffusion of cultural norms (Hershatter, 2019;Michelson, 2019).Third, though the rate of women's paid employment declines after economic reforms, its legitimacy gets consolidated and even moralized (Zhou, 2020).In this study, I examine meanings that emerged out of the relations between gender essentialism, male primacy, and beliefs about women's employment, and I explore logics that govern the coupling and decoupling of theses gender principles with one another.

Gendered history in China: rupture and continuity
This section outlines historical rationales for the multidimensionality of gender ideology in China.Based on the reasoning that historical circumstances during people's formative years shape sociopolitical attitudes that remain relatively stable over the life course (Kiley and Vaisey, 2020;Mannheim, 1952), I theorize five Chinese political generations that came of age under the early socialist era, the Cultural Revolution, the early reform, the radical reform, and the post-reform periods, which are characterized by distinctive political, economic, and cultural contexts that have important gender implications.I demonstrate how and why these generations vary in gender beliefs.
Early socialist era (1949)(1950)(1951)(1952)(1953)(1954)(1955)(1956)(1957)(1958)(1959)(1960)(1961)(1962)(1963)(1964)(1965).The early socialist period was characterized by imbalanced stateimposed feminism.In the party-state propaganda, women's labor force participation was made equivalent to their liberation.Both men and women were degendered state persons equally obligated to socialist construction (Zuo, 2005).In practice, however, women's disproportionate share of household labor and gender inequalities in wage work and collective farming were made invisible.The assumption that home is primarily women's responsibility was left unchallenged.Despite strong advocacy of gender equality, early socialist China was still male dominated, especially in the rural areas.Incomplete as the gender revolution was, feminist propaganda left imprints on a generation of men and women.Oral histories with early socialist women-both urban and ruraldocument their nostalgia for the period when they were valued for labor contribution (Hershatter, 2014;Rofel, 1999;Zuo, 2005).A study of urban married couples finds that men who got married during this period were less approving of division of gender roles into separate spheres than both younger and older cohorts (Pimentel, 2006).
Cultural Revolution (1966Revolution ( -1976)).The social and political disruptions brought by the Cultural Revolution made it difficult to assess its impact on people's ideas about gender.With women's liberation being reduced rhetorically to an established fact, gender ceased to define individual identity in public life (Yang, 1999).Nonetheless, the party-state adeptly used female images in political mobilization where slogans such as "times have changed, men and women are the same" were widespread.A militant and masculinized imagery of women predominates post-Mao interpretations of gender practices in the radical era.Memories of this tumultuous period are unsurprisingly mixed and shaped by the present socioeconomic positions of this generation.Retrospective accounts of elite urban women report exhilaration and autonomy when they were sent down to the countryside for "socialist re-education" (e.g.Ye and Ma, 2005).Similar feelings are rarely found among studies of worse off people and rural women where nihilism and resentment figured prominently (Hershatter, 2014;Xu, 2019).Men's grievance can be directly detected from prevalent post-Mao attacks on socialist gender practices.
Early reform period (1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992).The end of the turbulent Cultural Revolution in 1976 and the beginning of economic reforms in the late 1970s marked a new era of gender norms and relations.In the emergent space for public discussion, male intellectuals bombarded Maoist feminism for emasculating men and masculinizing women and attributed a masculinity crisis to women's liberation and socialist repressions of sexual desires (Dai, 1995;Zhang, 2015).Yet shadowed by criticism against the socialist patriarchy, more widespread male domination escaped intellectual reflections (Dai, 1995).Re-sexualizing women and pushing them back to the family were pursued as ways to rectify the wrong.Emerging urban unemployment in the early 1980s added economic incentives to this backlash.Popular media encouraged women to pay excessive attention to their appearances and homemaking to reconnect with feminine selves.Notions of women's biological differences and intellectual inferiority permeated Chinese society (Honig and Hershatter, 1988).In public minds, women's "low quality" (di su zhi) and lack of competitiveness justified growing gender and class inequalities accompanying the reforms (Jacka, 2014).
Radical reform period (1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003).More radical economic reform policies from late 1992 to 2003 focused on the privatization of state-owned enterprises (SOEs), which led to massive layoffs of women from SOEs and institutionalized efforts to push women back into the private sphere and informal sectors (Cook and Dong, 2011;Song, 2011).Consequently, the market more thoroughly replaced the state in shaping mainstream gender culture and fueled discourses on women's domesticity as the state withdrew from welfare and the subsequent privatization of care (Song, 2011;Wu, 2009).Onechild policies also facilitated rising ideologies of intensive mothering among the emergent Chinese middle class (Greenhalgh and Winckler, 2005).Meanwhile, rights-based feminist understandings of equality are introduced to China around the 1995 Fourth World Conference on Women in Beijing and made inroads into public discourse.Growing interactions with the world after China's 2001 accession to the World Trade Organization also introduced more liberal social values.-reform era (2004-present).A globalized market economy took hold in China after three decades of reforms.Economic development and labor surplus propelled an expansion of tertiary education around the turn of the century.Studies have demonstrated a consistent robust effect of education, especially college education in disseminating values about gender-equal opportunities (Davis and Greenstein, 2009;Shu, 2013).Internet access-another potential conduit of gender liberalism diffusion in the age of digital globalization (Charles, 2020)-also expands during this period.The post-reform generation, many of whom are singletons, lived through early adulthood with these egalitarian forces.Yet gender traditionalism has by no means been eradicated.Marriage remains nearly universal and is essential to womanhood (Davis and Friedman, 2014).Patriarchal traditions such as hypergamy dominate the marriage market, stigmatizing single highly educated women as "left-over."Having been brought up in a seemingly gender-neutral way, the singleton generation grew up to find themselves constrained by gender norms in private lives and occupational choices (He and Zhou, 2018).

Post
The gendered history has unfolded differently across urban and rural areas.The imprints left on each generation are likely to be contingent upon the deeply entrenched urban-rural divide.In what follows, I describe how urban-rural differences are likely to factor into the patterning of gender attitudes variation.

Urban-rural inequality, migration, and gender relations
The urban-rural divide is the most fundamental demographic axis of social inequality in China by measures such as income and educational attainment.The institutionalized divide through the household registration (hukou) system also fosters a cultural hierarchy partially based on differing gender relations, with "urban" associated with modernity and "rural" with the feudal past (Yan, 2008).Intense bureaucratic control through the danwei (working sector) system, along with industrialization, steered urban China away from traditional patriarchy as early as the early socialist era (Whyte, 2020).In the countryside, the deeply rooted patriarchy was less weakened by socialist collectivization.The urban-rural inequality also meant that socialist egalitarian ideals were less widely practiced in the rural region.For example, the public-subsidized childcare system available in urban China was not accessible to most rural Chinese (Hershatter, 2014).Modernization forces such as educational expansion, economic development, and globalization are also felt more keenly by urban residents.
There are reasons to expect that the urban-rural gap in gender attitudes narrows across generations, however.The de-collectivization of land and the rise of small enterprises created employment opportunities for women outside agriculture locally and inter-regionally (Jacka, 2014;Judd, 1994).It facilitated the rise of conjugal family that weakened patriarchy along the generational axis (Yan, 2003).But the egalitarian impacts brought by rural reforms should not be exaggerated.For one thing, village governance reforms pushed women out of leadership and revived an androcentric political structure (Judd, 1994).For another, family planning policy solidified son-preference culture in many rural areas.
Increasing internal migration adds another layer of complexity to how gender beliefs might vary.The first wave of internal migration that started in the mid-to-late 1980s and 1990s is seen as empowering for unmarried female migrants (Jacka, 2014;Pun, 2005), who constituted the bulk of migrants alongside males of more diverse age groups.The discrimination faced by these women also pushed them to culturally mimic the urbanites (Otis, 2011;Pun, 2005).Starting from the 2000s, the new dynamic of growing couple migration and left-behind children and grandparents gradually change the dichotomy of breadwinning and caregiving into a generational division and less a gendered one (Santos, 2017).When moving to cities, many migrant couples make pragmatic family adaptations that appear egalitarian (Choi and Peng, 2016).

Research hypotheses
Comparative research has revealed the multidimensional nature of gender ideology in Europe and North America.In China, market transitions along with the state-facilitated separation of the private and public spheres have been argued to shape a contemporary gender culture that normalizes women's paid work, naturalizes women's domesticity, and coexists with traditional patriarchal values (Ji et al., 2017;Song, 2011;Wu, 2009), the complexity of which will reflect in popular gender beliefs.I thus expect that: Individuals in contemporary China cluster into groups representing varied combinations of beliefs about gender essentialism, male primacy, and women's employment, with some exhibiting consistently egalitarian or traditional beliefs and others embracing hybrid gender-belief patterns that do not fit into the egalitarian-traditional axis (H1).Some forms of gender-essentialism may reject any male supremacist claims, while other variants may hold onto beliefs about vertical gender differences and accept some degrees of male supremacy.Simultaneously, with Chinese women's long involvement in the sphere of earning, not all forms of essentialist gender ideologies would be located within a separate-spheres framework where earning and caring are entirely separate.
China's temporally and spatially specific gender politics also suggest that gender-belief patterns are likely to vary across populations that are impacted by different histories and different interests.Modernization accounts imply that gender-attitude schemas will be increasingly egalitarian across generations.This leads to the prediction that: More recent political generations will display more egalitarian gender-attitude schemas than earlier generations (H2a).However, the post-socialist revival of gender traditionalism suggests that the modernization effect may be tempered for generations came of age under the reform era.Therefore, The change toward more egalitarian genderattitude schemas will be small for the reform generations (H2b).In addition, cross-national literature suggests that women are more gender-egalitarian than men, I expect that: The move toward more egalitarian gender-attitude schemas across generations is faster for women than for men (H2c).
The strict long-term enforcement of hukou and rural-urban inequality have led to far more rapid modernization process-and thus strengthened more liberal social values-in cities than the countryside.Therefore, Individuals residing in urban areas, including the rural-to-urban migrants, will be more likely to be gender-egalitarian than individuals living in rural areas (H3a).However, increasing geographic mobility and economic development in rural areas also suggests that: The urban-rural gap in endorsing more egalitarian gender-attitude schemas will narrow across generations (H3b).

Data and methods
This study uses data from the nationally representative pooled Chinese General Social Survey (CGSS) in the years 2010, 2012, 2013, 2015, and 2017.The five waves were conducted by the Renmin University of China and employed a hierarchical design and multistage probability proportional to size random sampling.Since the 2010 wave, the CGSS has consistently administered five questions pertaining to gender attitudes.CGSS does not exclusively focus on gender issues and is less subject to social desirability bias.Its large sample size and geographic coverage of China's 31 provinces and municipalities allow fine-grained investigation of attitudinal variation across demographic axes.Some of the gender-attitudinal measures appear in international surveys as well and some are particularly relevant to the Chinese society.The measures are: Q1.Men should put career first, and women should put family first.
Q2.For women, marrying well is better than working well.
Q3. Men were born more competent than women.
Q4. Women should be fired first during economic recessions.
Q5. Husband and wife should share household labor equally.
The first item measures normative beliefs in the primacy of men's breadwinner role and women's caregiving role, which are developed from the belief in gendered separate spheres.Most Chinese scholarship has relied on this singular item to assess change in gender attitudes over time.The second item originates from a popular saying in the reform era, the popularity of which supported its inclusion in the 2000 Survey on the Social Status of Women in China and multiple Chinese surveys since then.The common though contested understanding of "marrying well" primarily refers to women achieving upward social mobility by marrying "up" with men of higher social status and more wealth.By ascribing greater value to "marrying well" in comparison to "working well," this item attaches feminine self and worth to marriage and deemphasizes women's independence through paid employment.Q3 taps into the central component of gender status beliefs that ascribe men with greater status and worth on the alleged basis of greater competence (Ridgeway, 2011).The word "born" signals a biologically-based belief in men's greater competence and agency.The sexist framing is rarely seen in international surveys, providing a great opportunity to examine vertical forms of gender-essentialism with male supremacist connotations.Q4 pertains to overt employment discrimination against women in the form of male entitlement to work during job scarcity and indicates acceptance of male privilege.Q5 taps beliefs about division of household labor and is also connected to relations between men and women in heterosexual households.Similar items such as "sharing chores in marriage is important" and "men should take as much responsibility as women for the home and children" have shown to align well with familism and intensive parenting ideologies in Europe-based research (Grunow et al., 2018;Knight and Brinton, 2017).Together-though with overlaps-these five items tap into attitudes about roles that men and women are expected to perform within heterosexual marriages (Q1, Q5), acceptance of male primacy (Q3, Q4), and both hostile and benevolent types of gender essentialism (Q1, Q2, Q3). 1  Responses to these items are coded on a 1-5 scale, from "strongly disagree" to "strongly agree" and the midpoint "3" administered as "neither agree nor disagree" or "unsure." 2 I reverse-code Q5 so that the higher score reflects traditionalism, which is consistent with the other items.I collapse "strongly agree" and "agree" responses to indicate a traditional answer and "strongly disagree" and "disagree" to indicate an egalitarian stance.I retain the midpoint response, which in political opinion research is often interpreted as conveying respondents' nonknowledge, ambivalence, or sometimes contempt toward survey questions (Baka et al., 2012).Since these questions generally do not require domain-specific knowledge, an interpretation of ambivalence is adopted in this study.My related interview-based project on gender and family values in China which used these five CGSS items find that most midpoint responses negotiate a dilemma between "theory" and "practice" (e.g.wrong social value but true in practice), especially with Q2 and Q4, which further confirms the ambivalent nature of middle answers to these attitudinal items. 3 Given the research purpose of investigating how different dimensions of gender ideology cluster, I use LCA to identify groups in the pooled sample that share similar response patterns on these five items.Due to the exploratory nature of LCA, models with different numbers of classes are computed to reach the best-fitting model.The selection of an optimal five-class solution is made based on a series of fit statistics, the principle of parsimony, and theoretical interpretability.After model selection, I run a series of three-step multinominal logistic regression models that classify individuals into classes in a probabilistic manner and regress posterior probabilities of individual class assignment on demographic covariates (Vermunt, 2010).
The regression model also controls sociodemographic traits that plausibly predict gender attitudes, including education (elementary education or less, secondary education, and college equivalent and above), employment status (agricultural job, non-agricultural job, non-employed), marital status (currently married = 1), parenthood (yes = 1), ethnicity (Han = 1), religiosity (religious = 1), and communist party membership (yes = 1).To answer hypotheses on how generation intersects with region and gender in shaping the variation of gender attitudes, I computed multinomial regression models that include interaction terms between generation and residency/hukou status, as well as generation and gender.
The repeated cross-sectional pooled set of CGSS makes it difficult to distinguish the effects of age, generational cohort, and year.This issue is not completely solvable given that the current data-though best available-is not longitudinal and covers a 7-year span only.To address this issue, I estimate three reduced multinomial models, that is, a gross age/generation/survey year effect model, three reduced two-factor models (age + generation, age + year, generation + year), and a full model where all three variables are simultaneously controlled.I compared the fit statistics of each model to identify the best-fitting one.The penalized-likelihood model selection criteria, namely, the Akaike information criterion and sample-size adjusted Bayesian information criterion suggest that the reduced model that controls generation and survey year best summarizes the data.The reduced model also aligns with this article's theoretical framework.Only results of the best-fitting regression model are reported.
Cases with missing values on any of the attitudinal measures and the covariates are deleted.The listwise deletion generates 53,638 valid cases.The weighted means of gender attitudinal measures are presented in Supplemental Appendix 1A.The descriptive statistics of demographic covariates by political generation are presented in Supplemental Appendix 1B.Survey weights are used in both LCA measurement and regression models to adjust for sampling design.All analyses are performed using the software Latent Gold 6.0.

Gender attitude classes
Table 1 shows conditional item response probabilities of the five latent classes that emerge from LCA modeling.Goodness-of-fit statistics of the measurement models are presented in Supplemental Appendix 2. I label the five classes as (1) full egalitarianism, (2) egalitarian essentialism, (3) familist essentialism, (4) ambivalent essentialism, and ( 5) ambivalent.Consistent with Hypothesis 1, results defy unidimensional conceptualizations of gender ideology in China in the sense that no clear continuum emerges, and most respondents' gender attitudes are not uniformly egalitarian or traditional.Apart from full egalitarians and the ambivalent, the other classes do not evince uniformly egalitarian or traditional attitudes.To varying degrees, these three "essentialist" classes combine beliefs in intrinsic gender differences and beliefs in a neotraditional arrangement of household labor where both husband and wife should contribute.No class reports traditional responses on all items.
Full egalitarian describes 24.4 percent of the pooled sample.This label captures egalitarian responses to all items.This is the only group where the majority disagrees that men prioritize career and women family (Q1, P egalitarian = 0.84) and that for women marrying well is better than working well (Q2, P egalitarian = 0.71).With 77.6 percent of individuals in this class supporting equal division of household chores (Q5), full egalitarian constitutes the only class that supports joint spheres on caring and earning and the only class that does not stress women's primary role at home.In congruence with the anti-essentialist and egalitarian attitude, members in this class also disapprove that men were born more capable than women (Q3, P egalitarian = 0.94).Accordingly, nearly all members reject male primacy, disagreeing with the statement that undermines women's working rights (Q4, P egalitarian = 0.95).
Egalitarian essentialist comprises 36.6 percent of the sample and is the largest class.This group shows dual beliefs in women's labor rights (Q4, P egalitarian = 0.86) and the priority of their domestic roles (Q1, P traditional = 0.69).Associated with support of women's dual roles of primary caregiver and secondary breadwinner, 77 percent of individuals in this class express support for equal housework division that requires men's participation.The neotraditional orientation resonates with an ambiguity on the marriage centrality statement, which straddles between traditional and egalitarian (Q2, P traditional = 0.40; P egalitarian = 0.43).But this class differs from the Western version of egalitarian essentialism because the essentialism it adheres to does not necessarily exclude the cultural tenet of male primacy.Only half of individuals in this class explicitly oppose the statement on men's superior competence over women (Q3, P traditional = 0.36; P egalitarian = 0.50).
Familist essentialist captures 16.3 percent of the pooled sample.Its vast majority believes in women's domesticity (Q1, P traditional = 0.95) and agrees that a good marriage is better than a good career for women (Q2, P traditional = 0.89).At the same time, this class shows the highest level of support for equal housework division among all classes (Q5, P egalitarian = 0.81).The strong support for women's domesticity, the centrality of marriage to women's identity and personal worth, and men's participation in unpaid work together attests to this group's familistic orientation.Familist essentialists report a relatively low rate of opposition to laying off women first during economic recessions (Q4, P egalitarian = 0.51).Their agreement with this statement is higher than any of the other classes (Q4, P traditional = 0.37).Their high rate of support for superior male competence (Q3, P traditional = 0.89) suggests a possible explanation for the compromised stance on male entitlement to jobs.
Ambivalence essentialism comprises 15.9 percent of the respondents.Like the familist class, members in this class report high levels of adherence to the hostile stereotype of men's superior competence (Q3, P traditional = 0.79).Yet they do not overtly endorse male privilege in job security (Q4, P traditional = 0.15).Instead, they are selectively ambivalent toward this issue, using the evasive midpoint response "neither agree nor disagree" (Q4, P ambivalent = 0.51). 4The majority of this group strongly supports separate-spheres division of gender roles (Q1, P traditional = 0.89).Their attitude toward the marriage centrality statement is overwhelmingly traditional or ambivalent (Q2, P traditional = 0.54; P ambivalent = 0.34).Also, their support for men's participation in housework is relatively low (Q5, P egalitarian = 0.53).This class is therefore close to a watered-down version of gender traditionalism in a linear traditional-liberal continuum.But instead of overtly approving male primacy, this class does not express an explicit stance on it to avoid appearing sexist.
Ambivalent-the smallest class-describes only 6.9 percent of the pooled sample.Members of this class tend to give ambivalent answers to all items.As described further on, the demographics of this class resemble those of gender traditionalists reported by previous research.Concerns of social desirability might lead some gender-traditionalists to hide themselves in ambivalent answers.Another interpretation might be that people in this group are struggling to adapt to a conflicted gender culture in contemporary China (Koo et al., 2020).Also, women are less likely to be in this class than men.This gendered composition suggests that men's invested interests in the existing gender power structure are likely to orient them to indifference toward gender issues or reluctance to admit underlying sexism. 2 presents results of the best-fitting multinomial regression model that regresses individuals' class memberships on generation, survey year, and the other covariates introduced earlier. 5Each class is compared with the other four.For presentation purposes, not all pair-wise comparisons are shown.A noteworthy pattern along the gender and education lines first emerges in distinguishing full egalitarians from the rest, with women and more educated individuals more likely to belong to the full egalitarian class than any of the other classes.Urban residents, party members, and non-religious people are more likely to be members of the full egalitarian or egalitarian essentialist classes than the other classes.By contrast, rural natives, men, and the less educated are more likely to belong to the non-egalitarian essentialist classes.When do people endorse full egalitarianism over egalitarian essentialism?The early socialist generation, women, urban residents, more educated individuals, not religious and non-married people are more likely to be full egalitarians.The demographics of familist and ambivalent essentialist classes largely resemble each other.Yet being a woman, religious, married, and having children increases the odds of endorsing familist essentialism, which further testifies this class's familist orientations. 6In contrast, men are nearly two times more likely to be ambivalent essentialists than familist essentialists (odds ratio = 1.96, p < .001).Men are also more likely to be in the smallest ambivalent class.These gender differences suggest that all else equal, women are more likely to be full egalitarian rather than egalitarian essentialist, and familist essentialist rather than ambivalent essentialist.For men, it is exactly the opposite.In other words, women are more egalitarian than men and women tend to endorse a different type of gender-essentialism compared with men."Traditional" women desire to have men more involved in housework, though they endorse both hierarchical and horizontal forms of essentialism."Traditional" men, however, are marked by their reported ambivalence toward male primacy.

Demographic correlates of class membership. Table
Moving to the attitudinal shift across survey waves, the survey year coefficients show that the latent class composition changes most substantively in 2017 when compared with the reference category of 2010.In 2010, the egalitarian essentialist made up the largest latent class (41.7%) while the full egalitarian class constituted only 17 percent.By 2017, the sizes of these two classes have become comparable, both constituted a third of the sample.The expansion of the full egalitarian class could be partially attributed to the expansion of Internet use and a widespread penetration of social media in the 2010s. 7

Attitudinal disparities by generation, urban-rural divide, and gender
Generational distribution of gender-attitude classes.Zooming in on the distribution of gender-attitude classes across generations, Panels A and B of Figure 1 respectively shows unadjusted class membership by political generation and each cohort's predicted class size based on the multinomial regression model in Table 2.The most salient feature in Panel A is an increasing trend toward egalitarianisms and a decrease in ambivalent essentialism and familist essentialism across generations.Egalitarian essentialism has been steadily on the rise and is dominant among all cohorts except the post-reformers.Full egalitarianism starts to increase rapidly among the two youngest generations.In contrast, familist essentialism declines steeply among them.Ambivalent essentialism has been consistently waning.The proportion of the ambivalent declines moderately.
Examining change among cohorts, a break of attitudes is first observed between the early reform generation and the preceding socialist generations.The major point of differentiation is the growth of egalitarian essentialism and the decrease of ambivalent essentialism.The other classes remain nearly stagnant.Moving on to the radical reform generation, the rise of full egalitarianism and the continued decline of familist and ambivalent essentialisms becomes substantial.A much more radical break in attitudes is observed between the post-reformers and earlier cohorts.Examining the later three cohorts together, I find a more linear trend toward more egalitarian attitudes.Over 40 percent of post-reformers endorse full egalitarianism, surpassing the proportion of egalitarian essentialism.
Panel B presents that net of demographic traits, generational differences in class composition remain mostly unchanged.Again, there is a moderate trend toward liberal gender attitudes among the three early generations and a more rapid increase of full egalitarianism and decline in essentialisms among the two youngest ones.Among these three early generations, no linear trend is observed.The differing practices and memories of the two socialist periods left more egalitarian imprints on the early socialist generation than the Cultural Revolution generation.Coming from a less turbulent era, the early socialist generation is more likely to endorse full egalitarianism (see Supplemental Appendix 4).Again, compared with the socialist generations, the decrease in ambivalent essentialism-the class with the strongest belief in male primacy-among the early reformers is evident.Yet this cohort simultaneously holds firmly to women's domesticity, as evidenced by the sizes of familist and egalitarian essentialist classes.In a nutshell, the liberalizing effect that a market economy may have brought for the early reform generation is mild and is accompanied by strong essentialist sentiment.For the two young generations, there is a clear trajectory toward more egalitarian and less essentialist gender-attitude schemas.But even among these cohorts, the shares espousing hybrid forms of essentialism still surpass that of full egalitarianism.
Since the size of full egalitarianism greatly expands across the five survey waves, I further consider whether the egalitarian move applies for each political generation. Figure 2   Notes: Adjusted results are based on the multinomial LCA regression model in Table 2.
"period effect" on all generations, with a rise of full egalitarianism and a decrease of egalitarian and ambivalent essentialism across generations from 2010 to 2017.The increase of full egalitarians averages 40 percent among the five generations during the 7-year period.But interestingly, the proportion of familist essentialists remains remarkably stable and even increases among the Cultural Revolution and the early reform generations in 2017.Notes: Results are based on the multinomial LCA regression model in Table 2.
Confirming Hypothesis 2a, these results support the modernization notion of progress toward liberal gender attitudes in the sense of increasing opposition to male primacy-induced discrimination-both across generations and survey waves.But "progress" is weakened by the dominance of gender values that feature intrinsic gender differences and operate under logics of male primacy.The resilience of familist essentialism across survey waves further suggests persistent beliefs in female domesticity.
Is the egalitarian trend even across the urban-rural divide?The uneven socioeconomic development within China suggests a further disaggregation of political generations by the urban-rural divide.Figure 3 shows the estimated class size based on a multinomial regression model that includes interaction effects between generation and residency/hukou.Figure 4 presents the distribution of the full egalitarian class among urbanites, migrants, and rural natives across generations.
The trend of gender attitudes across generations follows distinctive trajectories for individuals with different residencies and origins.Rural natives and migrants follow similar pathways toward more liberal attitudes, though noticeably at different paces.Among these subgroups, there has been a rise in full egalitarianism and the egalitarian hybrid of essentialism and a decline in non-egalitarian essentialist classes.The increase in full egalitarianism is found among the two recent cohorts only, with a more substantial rise for migrants than rural natives.The differing pace reflects a much later acceptance of egalitarian attitudes on the full spectrum among rural people than those who migrated to cities.
The egalitarian trend is much flatter among the urbanite cohorts till the post-reform generation, where we see an evident increase in full egalitarianism and decline in egalitarian essentialism.Surprisingly, among the four elder generations, the size of the full egalitarian class varies little.The same features are also found in the distribution of essentialist and ambivalent classes among them.This pattern suggests that the egalitarian impacts of the socialist period and a continued liberalization process weakened by post-socialist feminist regression are observed among urbanites only.As such, Hypothesis 2b is only partially supported as the tempering modernization effects among the reform generations are limited to the urbanites, a group most impacted by the cultural backlash against socialist feminism.
Figure 4 shows that the urban-rural gap in endorsing egalitarian gender attitudes starts off wide and then narrows with the two youngest generations.Echoing with the uneven socialist revolution in the city and the countryside, there is a mirroring urban-rural divide in the predicted class size of full egalitarians among the socialist generations.The urban-rural gap still exists in later generations but shrinks considerably especially between urbanites and migrants among radical reform and post-reform generations.Figure 4 highlights the liberalizing effect that is associated with migration.Among the radical-reform and post-reform generations, the difference in endorsing full egalitarianism between migrants and rural natives is more than 10 percentage points.Together these results support hypotheses about the urban-rural gap in liberal gender attitudes that is shaped by socioeconomic disparities and different histories (H3a) but narrows along with economic modernization and increasing geographic mobility after the economic reforms (H3b).
Is the egalitarian trend the same for men and women?Since women are more likely to be full egalitarians than men, does that pattern hold across generations?To assess gender-attitude class compositions of men and women, I computed a regression model to assess the interaction effects between generation and gender.The posterior class assignment of men and women by generation is presented in Figure 5 and that of full egalitarianism is presented in Figure 6.
Figure 5 shows that the pace of change toward more liberal gender attitudes diverges greatly for men and women.For instance, though the proportion of ambivalent essentialism declines monotonically with each successive generation for both men and women, the decline is much sharper for women: down from 26.2 percent to 3.2 percent compared with 29 percent to 14.8 percent for men.In addition, it is women of the young generations who drive the growth of full egalitarianism.Likewise, the decline in egalitarian essentialism is found only among the post-reform women, not men.Compared with the accelerating egalitarian trend among radical-reform and post-reform women, the egalitarian trend for their male counterpart is much flatter.It confirms Hypothesis 2c about a stronger liberalizing effect for women than men, due partially to women's greater interests in egalitarian gender relations and the closing of gender gap in educational attainment.
Figure 6 illustrates the widening gender gap in liberal gender attitudes among recent cohorts.In stark contrast to women's surge in full egalitarianism, the proportion of full egalitarian men remains stagnant across cohorts.Among the socialist generations, there is no marked gender difference in the likelihood of being full egalitarians.However, a gender gap has been enlarging since then.Among the post-reform generation, 55.5 percent of women are predicted to endorse full egalitarianism while that for men is 26.8 percent.Compared with their male peers, post-reform women are over two times more likely to be full egalitarian than egalitarian essentialist (odds ratio = 2.146, p < .001).Further disaggregation shows that the gender gap is most evident among the urban natives, with 68.5 percent of women predicted to be full egalitarians and only 36.6 percent of men.The divergence in understandings of roles and status of men and women may lead more egalitarian young women to delay and forgo marriage as their fallback strategies, which could contribute to current demographic trends of low marriage and fertility rates in China.Meanwhile, it remains to be seen whether marriage and motherhood and subsequent conflicts with employment might undermine the young women's egalitarian attitudes.

Discussion
Based on meanings emerged through the relationships between male primacy, gender essentialism, and beliefs in women's employment, this study reveals five latent classes in China that blend different positions on these aspects of gender attitudes.Two of these classes demonstrate relatively consistent views, with full egalitarians repudiating all gender-essentialist notions and a malemajority ambivalent class that reports indecisiveness toward all issues.The three in-between classes, which are endorsed by the majority of respondents, together suggest that in China, gender essentialism should be understood in relation to the lingering impacts of male primacy and the competing ideologies of family solidarity and individual rights.The largest class egalitarian essentialism combines support for equality in the public sector with essentialist beliefs in women's caregiving qualities.Its adherence to the normative imperative that women's primary role at home and support for their employment rights lends itself closer to the "pro-work conservatism" that places dual emphases on women's wage-earning and care-giving roles and is dominant in East Asian and East European countries (Brinton and Lee, 2016) than the individualist "different but equal" egalitarianism that is prevalent in the United States and other Western countries.The other two types of essentialism-familist essentialism and ambivalent essentialism-to varying extent also believe in more hostile gender stereotypes that support male superiority in ability and working rights.
Given the Chinese state's historic failure to address women's domestic roles and the lack of public care in the post-reform present, responsibilities of social reproduction and care are shifted to individual families and women within families (Ji et al., 2017).Perhaps not surprisingly, female domesticity is still unquestioned by most Chinese today.Equally unrelenting is the normative requirement for men's breadwinning, which is often equated to their marriageability and manhood.To some, emphasizing women's primary role at home might defend a "natural" gender order.To others, it might align cognitively with a pragmatic concession to cope with work-family conflict and achieve family prosperity, albeit based on traditional gendered traits.However, as women's dual roles of wage earner and homemaker have become normative in both behavioral and moral senses (Zhou, 2020), renegotiations or at least reimaginations of unpaid work that entail greater participation by men seem to have occurred more rapidly than changes in assumptions about female domesticity.Consequently, we observe that essentialists of different variants, particularly familist essentialists, are positive about certain degrees of shared unpaid work and see this work as compatible with gender-specific social norms of earning and caring, which arguably carry greater moral and emotional binding power.But it remains to be seen whether gender-egalitarianism in unpaid work translates into men's increasing involvement in housework or remains a superficial commitment.Though Chinese men spend more time on housework than men in other East Asian societies with similar patriarchal cultures, the gender gap in housework shows no sign of decline across cohorts within China (Kan et al., 2022;Luo and Chui, 2018).
For one-third of the sample who endorses ambivalent or familist essentialism, the naturalized belief about female domesticity also mingles with notions of superior male capability to justify gender discrimination in the job market.This suggests that in China understandings of gender differences have not yet been fully subsumed into a liberal egalitarian framework and are still bounded by a hierarchical distinction between masculine and feminine qualities.Again, it differs from the post-industrial West where individual choice and autonomy undergirds an essentialist hybrid of gender-egalitarianism (Charles and Grusky, 2004).These illiberal forms of essentialism indicate the resilience of male primacy in prescribing understandings of gender differences in contemporary China.There are two likely reasons.First, male dominance has never been systematically challenged in China's socialist gender revolution (Wang, 2017).Second, in the post-socialist market economy popular beliefs in women's weaker market competitiveness became dominant and are married with the personal "quality" (su zhi) discourse that draws upon Confucian values of female inferiority, and therefore, lower quality (Wu, 2009).Thus, we are likely to observe that the liberalizing forces including socioeconomic modernization and globalization will be somewhat offset.In China, coupled with logics of familism and male primacy, gender essentialism does not necessarily coexist with ideologies of individual choice.
The heartening news is that across generations, there is increasing endorsement of more liberalegalitarian gender attitudes, with the trajectory of change differing markedly by urban, rural, and migrant identities.Among rural residents and rural-to-urban migrants, gender attitudes are liberalizing gradually across generations, with the pace faster for migrants.For urbanites, shares of egalitarian and essentialist classes vary little among the socialist and reform generations though.This pattern exemplifies both the egalitarian imprints of socialist feminism and the attenuation of a liberalizing process by the resurgence of traditional gender values.Both effects are amplified for the urbanites who are more directly impacted by both processes, though the impacts of resurging traditionalism seem to diminish among the post-reform urbanites.
The distinct demographic compositions across generations have limited power in explaining their differences in latent class memberships.A general cultural shift brought by socioeconomic modernization, globalization, and more recently the Internet penetration provides a potential explanation for the overall egalitarian trend across generations.In particular, the growing claims of individualism are likely to be conducive to stronger support of gender equality in the public sphere (Shu and Zhu, 2019).The egalitarian "period effect" is also evidenced by the temporal shift from 2010 to 2017, during which all generations become more egalitarian.But the process of a cultural shift seems to be deeply gendered.Chinese men's share of full egalitarianism plateaus across generations and post-reform men are lagging far behind their women peers.The gender gap is particularly large among urbanites.Furthermore, the post-reform men's relatively large shares of ambivalent essentialism and ambivalence-two classes that are "indecisive" toward workplace inequality-indicate a general acceptance of male primacy.At the very least, this evidence suggests that the rising understandings of equal access to opportunities is understood and practiced in deeply gendered ways.The forces of individualism-if at work-are less perceptible by young Chinese men in diverting their gender attitudes in an egalitarian and non-essentialist direction.Against the backdrop of the complex gender culture that accompanies China's social transition, young men and women live in different realities and perceptions where new and old ethics are entangled in gender-differentiated ways.The gender gap in endorsing egalitarian gender attitudes would create difficulties for young egalitarian women to find suitable partners and deter them from entering marriage.Between married heterosexual couples, the gender gap in gender-egalitarianism and potential incongruence between gender beliefs and gender practices could negatively affect marital quality, especially when coupled with work-family conflicts (Li et al., 2020).
Due to the availability of data and gender-attitude measures, this study is subject to the following limitations.First, though generation is the best window to assess a temporal shift of gender attitudes with the absence of longitudinal data, the cross-sectional CGSS disallows me to fully disentangle period and generation effects.Second, there is a lack of measures tapping into occupational gender stereotypes, which are widespread among Chinese youth according to studies that utilize small-scale surveys (He and Zhou, 2018;Koo et al., 2020).It is thus not possible to tease out whether the full egalitarianism could in fact be the "different but equal" essentialist egalitarianism.

Conclusion
Mirroring the complex trends that shape mainstream gender culture in contemporary China (Evans, 2021;Ji et al., 2017;Song, 2011;Wang, 2017), this study reveals a mosaic of gender beliefs at the individual attitudinal level.By demonstrating the importance of male primacy as a contextually salient gender principle, results suggest that a different-but-equal version of gender-egalitarianism does not adequately describe dominant gender beliefs in China.Rather, within a social context that strongly emphasizes family solidarity, the mosaic of beliefs takes pieces from lingering male superiority values and from neotraditional combinations of separate spheres ideology and beliefs about egalitarianism in unpaid housework.Future research might examine how attitudes about gender roles in China are configured in relation to the neo-familist values that emphasize family prosperity and the centrality of children in family life (Yan, 2021) and in relation to such traditional family values as patrilineality and filial piety.Moreover, as a multidimensional conceptualization of gender ideology is increasingly applied to non-Western societies (e.g.Charles et al., 2022), this study underlines the importance of attending to regionally salient gender principles.
Findings on the diversity of multidimensional gender-belief patterns within China complicate both the modernization account of steadily evolving egalitarianism and the feminist account of traditionalist revival in post-socialist China.I show that the relative importance of these two accounts depends on the generational, regional, and gendered identities being examined.Modernization-driven progress toward gender-egalitarianism is evidenced among people with rural origins who are less impacted by the socialist gender revolution, and for the same reason, more "immune" to setbacks brought by revivals of traditional gender values.By contrast, we see stalled progress among the socialist-and reform-generation urbanites.Effects of the socialist gender-egalitarian discourses and practices and then the backlash against them appear to have offset a linear progress among them.Also, the modernization account is more applicable to women who have less interest than men in going back to a traditional gender culture.Conversely, much slower progress is observed among men who continue to benefit from these traditions.These intersectional patterns suggest that the gender-traditionalism revival thesis is based largely upon urban and masculine experiences.A shift of focus to the "subalterns" in future research would reveal different paths and intricacies of gender ideology changes in China and beyond.
The structure and diversity of gender attitudes in China together shed light on possible futures of gender ideology development.Implications for China's demographic trends are likely multifaceted.It is encouraging to see more widespread acceptance of gender-egalitarian values among the post-reform generation, which draws them closer to their international millennial peers.Through cohort replacement, economic development, and growing urbanization, we could expect a continued expansion of full egalitarianism and egalitarian essentialism in China.Over time, the in-between variants of essentialisms are likely to evolve into a pro-work and pro-female-domesticity belief pattern that is both essentialist and egalitarian, yet the dominance of which has shown to depress a country's fertility rate across the post-industrial world, notably Japan and Korea (Brinton and Lee, 2016).It is plausible that similar relationships have unfolded in China and will continue to contribute to its declining fertility rate.Another side of the rising tide of gender-egalitarianism is that young Chinese women have far outpaced men, which resonates with the growing gender wars in China's cyberworld.The discrepancies between young men's and women's gender attitudes could further exacerbate declining marriage rates and very low birth rates.
compares the latent class composition of each generation in 2010 and 2017.It suggests a similar liberalizing

Figure 3 .
Figure 3. Predicted class membership by residency/hukou and political generation.(a) Rural native.(b) Migrant.(c) Urban native.Notes: Results are based on a multinomial LCA regression model that includes an interaction term between political generation and residency/hukou status.

Figure 4 .
Figure 4. Proportion of full egalitarianism by residency/hukou and political generation.Notes: Results are based on the same multinomial LCA regression model as in Figure 2.

Figure 5 .
Figure 5. Predicted class membership by gender status and political generation.(a) Male.(b) Female.Notes: Results are based on a multinomial LCA regression model that includes an interaction term between political generation and gender.

Figure 6 .
Figure 6.Proportion of full egalitarianism by gender status and political generation.Notes: Results are based on the same multinomial LCA regression model as in Figure 4.

Table 1 .
Item response probability by latent class.
Notes: N = 53,638.Results are weighted.The largest conditional probability is given in boldface to highlight class features.

Table 2 .
Multinomial three-step LCA regression models predicting gender-attitude latent class membership.
Notes: N = 53,638.Results are computed using personal weight.LCA: latent class analysis; CCP: Chinese Communist Party.