Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Aug 23.
Published in final edited form as: Soc Forces. 2011 Mar 1;89(3):961–982. doi: 10.1353/sof.2011.0011

Cohort Changes in the Socio-demographic Determinants of Gender Egalitarianism

Fred Pampel 1
PMCID: PMC3159871  NIHMSID: NIHMS317392  PMID: 21869848

Abstract

Arguments about the spread of gender egalitarian values through the population highlight several sources of change. First, structural arguments point to increases in the proportion of women with high education, jobs with good pay, commitment to careers outside the family, and direct interests in gender equality. Second, value shift arguments contend that gender norms change with social and economic development among women and men in diverse positions – traditional and non-traditional alike. Third, diffusion arguments suggest that structural change leads to adoption of new ideas and values supportive of gender equality by non-traditional and innovative groups in society, but that the new ideas later diffuse to other groups through cultural processes. Using the General Social Survey from 1977 to 2006 and comparing the determinants of gender egalitarianism across 86 cohorts born from roughly 1900 through 1985, multilevel models support the diffusion arguments.

Gender egalitarianism contributes to progress toward women's freedom and independence in several areas of social life: fertility (Rindfuss, Brewster and Kavee 1996), work choices (Clark, Ramsbey and Adler 1991), job segregation (Charles and Bradley 2002), welfare spending (Bolzendahl and Brooks 2007), politics and voting (Brooks 2000), and family relationships (Amato and Booth 1995; Kaufman 2000). In fact, much progress has been made in moving toward the goal of widespread support for gender equality (Cherlin and Walters 1981; Jackson 1998; Thornton and Young-DeMarco 2001). In the United States, the emergence of new norms during the 1960s and 1970s among second-wave feminists prefigured wider adoption of gender egalitarianism (Mason and Lu 1988).

Studies of changes in gender egalitarianism emphasize a variety of structural and cultural influences, and numerous studies have examined whether the changes occur among groups in positions most prone to adopt feminist attitudes or more widely among diverse groups. A diffusion approach adds another twist by suggesting that structural change provides the impetus for adoption of new ideas but also that widespread diffusion of new values follows. All arguments predict rising levels of gender egalitarianism but specify different patterns of change.

Structural Influences

Changes in gender egalitarianism may occur through increases in the proportion of women with high education, good-paying jobs and commitment to careers outside the family. A shift from industrial economies to post-industrial service and knowledge economies increased the number of jobs traditionally filled by women and the demand of employers for female workers (Huber 1990). As the number of employed women increases to meet this demand, the labor force becomes more gender integrated. Along with a larger female labor force, higher-level, tertiary educational opportunities for women increase, and professional and managerial jobs once filled by men slowly open up to women. Demographic changes reinforce these economic and educational changes (Brooks and Bolzendahl 2004). Trends toward later marriage, fewer children and more divorce strengthen the attraction to careers and goals outside the family, the need of women for independent income, and the opportunity for middle-aged mothers to return to the labor force (Oppenheimer 1976). At the societal level, then, the composition of the population changes in ways that foster gender egalitarianism.

An interest-based mechanism underlies arguments about the economic stake that non-traditional women have in gender equality (Bolzendahl and Myers 2004; Huber and Spitze 1981; Plutzer 1988). Women with high education and special job skills suffer most from gender discrimination in the labor force and feel most deprived by lack of opportunities. Because these women gain the most from equal treatment, they tend to have more egalitarian attitudes. Conversely, women with more traditional commitments to family and children will maintain or perhaps even strengthen their adherence to traditional attitudes in response to structural change (Glass 1992; Kane and Sanchez 1994) and make gender-based political cleavages more salient (Plutzer 1988). Women in homemaking and mothering roles who gain little economically from gender equality have fewer incentives to adopt new attitudes. As Baxter and Kane (1995) argue, women's dependence on men at both the individual and societal levels draws them toward less egalitarian views.

Related mechanisms affect egalitarian attitudes among men – although not to the same extent. In some ways, their interests lie in gender inequality that reinforces their advantage in opportunities, jobs, and incomes. At the same time, however, husbands and family members benefit economically from more egalitarian treatment and higher pay for working spouses, children and relatives (Banaszak and Plutzer 1993a; Morgan and Walker 1983; Warner 1991; Zuo and Tang 2000). Like their wives and daughters, men may become more liberal in their gender attitudes because gender equality favors the financial interest of the household (Smith 1985). Studies thus find that men have stronger egalitarian attitudes when they are part of a dual-earning couple (Cha and Thébaud 2009; Wilkie 1993). In addition, men may respond negatively to obvious inequities and positively to the benefits of equal opportunity at work. Such exposure may increase gender egalitarianism (Bolzendahl and Myers 2004; Davis and Robinson 1991; Kane and Sanchez 1994).

The interest and exposure versions of the structural argument tend to view entrance of women into new education, work and family roles as preceding changes in attitudes. Attitudes in large part follow from behavior rather than the other way around (Oppenheimer 1976): Those in positions to benefit most from equality will adopt egalitarian attitudes more quickly than those in more traditional positions. For example, Rindfuss, Brewster and Kavee (1996) find that entrance into the labor force of mothers with young children occurred during times of normative opposition to the practice and that attitudinal change followed the behavioral change. Much of the change in gender egalitarianism accordingly comes from cohort replacement (Brewster and Padavic 2000; Brooks and Bolzendahl 2004; Firebaugh 1992). Because the positions and roles of many in older cohorts do not change, their attitudes remain traditional throughout their lives. The replacement of older cohorts by younger cohorts, who are more affected by changes in opportunities for women, leads to greater prevalence of gender egalitarianism.

These arguments imply the importance of compositional change for rising gender egalitarianism. If those with education, work, job and family characteristics predisposing them toward egalitarian views steadily become a larger part of the population, then the level of gender egalitarianism will rise as well, even as inegalitarian values persist among traditional groups. Structural position arguments thus emphasize the importance of rising female education, labor force participation and dual-career families for increasing gender egalitarianism.

Value Shifts

Arguments focusing on broad shifts in values suggest that gender norms develop, at least in part, independently of social structural position. Women and men in varied positions of a society – those in traditional and non-traditional roles alike – come to adopt more egalitarian attitudes with social and economic development. Broad changes in values do not erase attitudinal differences across positions – continuing relationships exist between social position and gender egalitarianism. However, value changes may raise egalitarianism similarly across diverse social positions and groups, particularly among younger generations.

Consistent with these claims, within-nation studies find that gender egalitarianism has risen among most groups. Thorton and Young-DeMarco (2001) demonstrate a pervasive trend toward endorsement of gender equality in the United States from the 1960s to the 1990s. Rindfuss, Brewster and Kavee (1996) similarly find that attitudes supportive of working mothers grew across all groups rather than from the changing education, work and age composition of the population. Support for gender equality grew among men as well as women (Bolzendahl and Myers 2004) and among active Protestants with conservative religious views as well as less religious and liberal groups (Petersen and Donnenwerth 1998). In comparing diverse high-income nations, Treas and Widmer (2000) find evidence of cross-national consensus in gender attitudes, and Scott, Alwin and Braun (1996) find some similarities in liberalization of attitudes.

The pervasiveness of value shifts raises questions about why groups in different positions and with different interests adopt similar egalitarian attitudes. Inglehart and Norris (2003) argue that economic prosperity and material security foster a broad cultural shift toward quality of life values that emphasize self-expression and individualism (also see Inglehart and Baker 2000). The changing material conditions and values tend to erode traditional beliefs, family authority and communal obligations and lead to changing views of women's roles. In arguing for the inevitable movement toward gender equality in modern societies, Jackson (1998) makes similar points. He says that the shift of economic and political power from households to business and government institutions in modern societies promotes gender equality. The shift ruptures traditional boundaries between men's and women's roles and weakens the incentives needed to maintain men's power over women. Given the pervasiveness of the change, diverse groups tend to respond with stronger support for gender equality.

Although affecting most socio-economic groups, these value changes emerge generationally. Adoption of new values commonly occurs during adolescence and young adulthood, and cohorts tend to retain these values throughout later adulthood. Inglehart (1989) argues that cohorts raised during the post-World War II decades of material prosperity and economic security widely adopted post-materialist values. The gap in value orientations between generations overwhelms socio-economic difference in values within generations. In support of this claim, Inglehart and Norris (2003) find that secular cohort changes occur in fundamental values and that generation more strongly predicts egalitarian attitudes than sex, class or education.

These arguments treat interests in broad terms (Inglehart 1989). Economic changes toward industrialization and post-industrialism shift motives for action from the search for material needs and the reliance on traditional forms of social organization to the search for self-expression, innovation and justice. Because income grows broadly in post-industrial societies (despite the persistence of inequality), such changes broadly affect norms and values. Interests thus remain a key source of egalitarian values but relate to the larger economic context rather than to the particular positions of individuals. For example, laws and regulations relating to equal opportunity, affirmative action, equal pay, maternity leave and reproductive rights both promote and reflect the widespread acceptance of gender egalitarian views in a society (Alwin, Braun and Scott 1992; Crompton and Harris 1997; Inglehart and Norris 2003; Norris 1987).

Despite the attention to living standards, value shift arguments differ from structural arguments in accounting for rising gender egalitarianism. They posit that, at least among post-World War II cohorts, egalitarian attitudes increase across diverse gender, work, education, income and family groups. While women in non-traditional positions have more egalitarian views than others, economic, cultural and political-legal changes in post-industrial societies raise gender egalitarianism across broad social positions, particularly among younger cohorts.

Patterns of Diffusion

Diffusion arguments highlight a sequence of influences relating to both structural position and value shifts. The key is that contextual changes not only increase the level of gender egalitarianism but also affect social groups differently (Fischer and Hout 2006). The early stages of change most affect the attitudes of educated and working women, those in non-traditional positions and with the strongest interests in gender equality. Reflecting a form of backlash, traditional attitudes among other groups may harden in response. Change thus tends to polarize initially: It widens the gap between non-traditional and traditional women and strengthens the effects of individual characteristics on gender egalitarianism. At later stages of change, however, egalitarianism diffuses vertically from high status, non-traditional innovators to lower status, less innovative, and more traditional groups (Poole and Zeigler 1981). As those with lower levels of education, weaker ties to the labor force, less prestigious jobs and larger families come to adopt attitudes similar to more innovative groups, views on gender egalitarianism tend to become less polarized.

The argument thus predicts a sequence of changes relating first to structural positions occupied by women and men and then to pervasive value change. This interaction takes a non-linear form. The strength of a determinant first increases as innovative groups with strong interests in gender equality adopt egalitarian views and set themselves apart from other groups. The strength of a determinant then decreases as gender egalitarian views diffuse to larger parts of the population with less direct interest in equality. In short, value divergence across groups is followed by value convergence (Fischer and Hout 2006).

Consistent with the argument, Mason and Lu (1988) find that gender egalitarianism grew similarly among most socio-demographic groups in the United States from 1977 to 1985, but that college educated women, who already had high gender egalitarianism in both periods, were an exception. The process of catching-up reflects a narrowing of formerly wide attitude differentials and the spread of new values to less innovative groups. Bolzendahl and Myers (2004) find that decreased polarization has occurred in attitudes toward women's participation in the public sphere and that, as most everyone comes to accept gender egalitarian goals, individual determinants have declining influence. Fischer and Hout (2006) find widening and narrowing by age and city residence in measures of approval for working women from 1936 to 2000.

These changes in relationships fit patterns of diffusion. Montgomery and Casterline (1993) define diffusion as the influence of adoption of innovative ideas and behaviors by some individuals on the likelihood of adoption by others. Diffusion often first occurs horizontally among higher socio-economic groups, as these groups tend to be most innovative and have communication networks across structurally equivalent positions (Strang and Meyer 1993). Vertical diffusion often follows, as lower ranking groups adopt the practices and ideas of more prestigious groups (Fischer 1978; Strang and Soule 1998; Wejnert 2002). Interests or the relative advantage of adoption (Rogers 2003) play a role in the diffusion theory, but other mechanisms of action are important in explaining vertical diffusion. Other groups may follow innovators in adopting gender egalitarianism through processes of class emulation and social learning that sociological arguments about social inequality and cultural distinction highlight (Bourdieu 1994; Simmel 1971[1904]; Veblen (1992[1899]; Weber 1958). Moreover, acceptance of new ideas may become self-sustaining after adopters reach a critical mass (Rogers 2003). At that point, adoption by less innovative groups requires less risk and boldness. Indeed, resistance to increasingly popular ideas becomes more difficult, especially when the ideas come to receive support from the mass media, public policies and legal decisions.

Given the pattern of change, the early adoption of gender egalitarianism by more advantaged groups initially involves innovation that strengthens socio-economic differences, but the vertical diffusion of the values to other groups later reduces socioeconomic differences. However, patterns of diffusion relate importantly to gender. Women who benefit most from gender equality will more quickly adopt new attitudes than men (Ciabattari 2001). The diffusion of gender egalitarianism may occur for men but not as quickly or to the same extent as for women.

Hypotheses

The hypotheses focus neither on overall changes in support for gender egalitarianism nor on group differences in attitudes – topics that have been well studied. Rather, they focus on the combination of the two topics, on whether groups change their attitudes at different rates and produce divergence in views followed by convergence.

H1

Changes in the population composition – the distribution of individual characteristics such as education, employment, occupation, income, marital status, family size and religiosity – account for the trends in gender egalitarianism. Consistent with structural arguments, the hypothesis implies that differences in gender egalitarianism across groups in traditional and non-traditional positions are maintained over time, but that groups in traditional positions decline in size relative to groups in non-traditional positions.

H2

Changes in gender egalitarianism occur similarly across groups. Consistent with value shift arguments, the hypothesis implies that differences in views across groups in traditional and non-traditional positions persist but that egalitarianism rises over time across most socio-economic groups.

H3

Changes first affect innovative, high status groups, thus strengthening the effects of socio-demographic determinants of gender egalitarianism and creating divergence in views, but the changes later affect other groups, thus weakening the effects of the socio-demographic determinants and creating convergence in views. Consistent with diffusion arguments, the hypothesis also implies that the changes occur more quickly and strongly for women than men.

Although most studies of change compare gender egalitarianism across survey years, a cohort-based approach has advantages in testing the hypotheses. Generations are prone to adopt new values during young adulthood that persist through older ages (Brooks and Bolzendahl 2004), thus patterns of adoption of gender egalitarianism should show most clearly in comparisons across cohorts. Studies find that cohort replacement plays an important role in changing gender egalitarianism (Firebaugh 1992; Inglehart and Norris 2003; Schnittker, Freese and Powell 2003), and this role may appear clearly in the changing effects of socio-demographic determinants. Tests of the hypotheses thus come from comparing the group differences in gender attitudes across persons born in different periods of the twentieth century. Given stronger interests in gender equality among women and their greater amenability to adopting liberalizing attitudes and values, the tests also must be done separately for men and women.

Previous studies have analyzed consecutive cross-sectional surveys in the United States, first for periods through the 1970s (Mason Czajka and Arber 1976; Spitze and Huber 1980), then through the 1980s (Mason and Lu 1988; Rindfuss, Brewster and Kavee 1996), and more recently through the 1990s (Bolzendahl and Myers 2004; Brewster and Padavic 2000) and early 2000s (Carter, Corra and Carter 2009). Several examine the influence of cohort on attitudes (Brooks and Bolzendahl 2004; Schnittker, Freese and Powell 2003; Wilkie 1993) and a few compare differences in the influence of determinants of gender egalitarianism across cohorts (Brewster and Padovic 2000; Ciabattari 2001). However, none fully tests all three hypotheses or, in particular, the predictions about nonlinear strengthening and weakening across cohorts of the effects of socio-demographic determinants.

Methods

Testing the hypotheses requires data on gender egalitarianism and its individual determinants for a period of decades and for cohorts born over an even longer period. The General Social Surveys, a set of consecutive cross-sectional surveys based on full probability samples of the non-institutional, English-speaking adult population in the United States (Davis, Smith and Marsden 2007; NORC 2008), meet these criteria. There are four identical gender egalitarian items used in 15 surveys, the first in 1977 and the last in 2006. Pooling the data for the surveys and cohorts yields 20,985 cases with data on the key variables.

The 30-year time span of data means the surveys contain older cohorts born since 1900 and younger cohorts born through 1985. The cohorts thus entered adulthood during periods of widely varying degrees of economic prosperity, gender inequality and acceptance of post-materialist values. Historical time also plays a role in that all cohorts are affected by events and general economic, social and cultural trends. Controlling for year removes these influences in the analysis of differences across cohorts.

Measures

The gender equality items available for 15 surveys from 1977 to 2006 ask about four statements: (1. it is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family (strongly agree, agree, disagree, strongly disagree); (2. most men are better suited emotionally for politics than are women (agree, disagree); (3. a preschool child is likely to suffer if his or her mother works (strongly agree, agree, disagree, strongly disagree); and (4. a working mother can establish just as warm and secure a relationship with her children as a mother who does not work (strongly disagree, disagree, agree, strongly agree).1 Pooled across all available years, these four items define a single dimension of gender egalitarianism in an exploratory factor analysis and form a scale with an alpha reliability of .712 (with high values indicating support for gender equality). The reliability differs little across men and women and remains largely stable over time.

The lack of nuance in these items means they miss important components of views about gender equality (Brewster and Padavic 2000; Mason and Lu 1988), and attitudes about public equality relating to work and politics in items 1 and 2 may change at different rates than components relating to private or family equality in items 3 and 4 (Goldscheider, Oláh and Puur 2010). Still, dozens of studies based on the GSS or cross-national survey data use these or similar items. The reliability of the scale used here plus the meaningful relationships observed in previous studies between these sorts of gender equality items and socio-economic characteristics and period trends suggest that, despite clear weaknesses, the items have value. To the extent that the items lack validity, miss key elements of gender relations, and mix components of public and private equality, it makes it harder to find support for the hypotheses. The scale thus provides conservative tests of the theoretical arguments.

Cohort measures single years of birth, excepting the first category of 1900 or earlier and the last category of 1985 or later. As such, it can reflect time-dependent value changes and diffusion processes. The models treat the effects of cohort as a quadratic. Based on BIC statistics, two quadratic cohort terms do nearly as well as a set of 9 dummy variables for 10-year cohorts and better than 16 dummy variables for 5-year cohorts. Although they give results that are similar for the dummy variables, the quadratic terms require fewer interaction terms to test the hypotheses. Thirteen dummy variables for year supplement the cohort variables and reflect historical changes that affect all cohorts similarly. Following Brooks and Bolzendahl (2004), measures of life course statuses are used to represent the effects of age while avoiding redundancy with cohort and year.2

For the socio-demographic variables, race takes the form of two dummy variables for blacks and others (with whites as the referent). Region of residence takes the form of eight dummy variables created from categories of similarly located states, and size of city of residence ranges from open country (1) to city with more than 250,000 (10). A dummy variable measures married versus others, and a continuous variable measures the number of children (up to eight or more) the respondent ever had. A measure of church attendance ranges from never (0) to more than once a week (8).

For the SES variables, education equals the respondent's completed years of schooling. A dummy variable measures those working, unemployed or going to school (coded 1) relative to those who are keeping house, retired or in a residual, other category. As an alternative to this measure of work status, another dummy variable distinguishes self-identified homemakers (coded 1) relative to all others. The homemaker variable may have special importance for gender egalitarianism among women (but includes only 219 men). Prestige of current or former occupation is coded on the basis of a scale constructed from ratings of the general social standing of occupations (Davis, Smith and Marsden 2007). Those with no information on current or former occupation (5.9 percent of the sample) are assigned the mean for the year of survey. Current personal income in dollars comes from categories ranging from under $1,000 to over $75,000, with the values recoded to the midpoint of the category and adjusted for inflation.3 The midpoints used for the top open-ended categories come from Hout (2004), who computes values that downwardly adjust the usual Pareto-formula estimates. As evidence that the coding does not bias the measures, a dummy variable for belonging to a top category proves insignificant when added to the models. For the 10.5 percent of the sample lacking income data, the mean for the year of the survey is assigned. Dummy variables for those not reporting occupational prestige and income remain insignificant when included in the models.

Of the socio-demographic determinants of gender egalitarianism, education seems best suited for the study of changes across cohorts. Determined in early adulthood, it best reflects the position of persons when attitudes toward gender equality develop. The other work, job, income, family and religion variables refer to current characteristics that, for older cohorts in particular, may differ from when younger. The lack of retrospective data on work and family life limits the measures and weakens the results. Still, as a supplement to education, the other measures can provide additional if less ideal tests of the hypotheses. Descriptive statistics for the variables appear in the appendix. For all the variables with effects that differ across cohorts, however, I transform the original units into standardized units so comparisons of the extent of cohort-based changes are meaningful.

Appendix.

Descriptive Statistics by Cohort Groups

Variable All Cohorts Means by Cohort Groups
Mean Min Max 1900-1929 1930-1949 1950-1969 1970-1985
Gender egalitarianism scale .00 -1.95 1.60 -.46 -.06 .15 .22
Cohort 49.49 0 85 19.28 40.92 59.15 75.56
Cohort2 (*.01) 27.88 0 72.25 4.27 17.07 35.30 57.29
Education 12.99 0 20 11.31 13.11 13.50 13.24
Labor force or school .72 0 1 .25 .72 .86 .90
Homemaker .14 0 1 .24 .14 .12 .09
Occupational prestige 42.29 12 86 40.86 43.59 42.64 40.31
Individual income 31.31 .5 187.5 15.57 38.44 35.37 23.47
Married .60 0 1 .62 .73 .59 .29
Number of children 1.92 0 8 2.76 2.66 1.50 .65
Church attendance 3.83 0 8 4.43 4.03 3.63 3.25
N 20,985 4,043 5,820 8,673 2,449
Open in a new tab

Models

The hypotheses specify possible changes across cohorts that affect not only the level of gender egalitarianism but also its determinants. Multilevel or hierarchical models (Raudenbush and Bryk 2002), which nest one level of data (in this case, individuals) within another level of data (in this case, cohorts), are well suited for testing such hypotheses. Gender egalitarianism for individual i in cohort j (GEij) is a function of k individual-level socio-demographic variables (Zkij) measured in standard units, m individual-level control variables (Xmij), and an error term (rij). The level-one model specifies separate equations for i individuals within each cohort j:

GEij=β0j+Σβkj∗Zkij+Σβmj∗Xmij+rij (1)

The β0j and βkj coefficients, treated as random rather than fixed effects, then serve as outcomes in a second set of equations that use cohort (Cj) and cohort squared (C2j) as determinants:

β0j=γ00+γ01Cj∗+γ02∗C2j+u0j, (2a)
βkj=γk0+γk1Cj∗+γk2∗C2j+ukj, (2b)
βmj=γm0 (2c)

Cohort can non-linearly influence the adjusted level of gender egalitarianism in equation 2a and the slopes of the individual-level determinants in equations 2b (the slopes of the control variables remain fixed). The coefficient estimates come from restricted maximum likelihood techniques in HLM 6.08 that adjust for probability weights used by the GSS (Raudenbush et al. 2004). With statistical tests confirming that the models differ by gender, separate results are presented for males and females.4

The three hypotheses translate into predictions about the multilevel coefficients. The structural position hypothesis (H1) predicts that, because changing population composition accounts for increases in gender egalitarianism, cohort will have little influence net of the individual determinants (i.e., γ01 = γ02 = γk1 = γk2 = 0). The value shift hypothesis (H2) predicts that, because of the widespread increase in gender egalitarianism among all groups, cohort will non-linearly increase the intercept or levels of gender egalitarianism (i.e., γ01 > 0 and γ02 < 0). The diffusion hypothesis (H3) predicts that the slopes of the individual determinants, not just the level of gender egalitarianism, will change across cohorts as favorable views first emerge among innovative groups with direct interests in equality and then diffuse to other groups. This change in relationships across cohorts implies that the key determinants of GE interact with cohort and cohort squared. Determinants with a positive relationship to GE, such as education and work variables, should have a positive interaction term with cohort and a negative interaction term with cohort squared (i.e., γk1 > 0 and γk2 < 0). Determinants with a negative relationship to GE, such as family and church variables, should have a negative interaction term with cohort and positive interaction term with cohort squared (i.e., γk1 < 0 and γk2 > 0). However, the interactions should be stronger for females than males.

Results

Table 1 shows effects of the level-1 additive SES and family determinants of GE (net of controls for race, city size, region and year). In general, higher SES men and women with lower family and religious involvement hold more egalitarian gender attitudes. Married women who attend church tend to have low GE, while educated women in the labor force or school and women with prestigious jobs and high personal income tend to have high GE. The effects for males are similar in pattern but different in specific determinants. Being married has little influence, but number of children reduces GE. Much as for women, education, prestige and personal income raise and church attendance lowers GE.

Table 1.

Coefficients and t Ratios from Multilevel Models of the Gender Egalitarianism Scale

Female Male
Predictors b t b t
Level 1
    Education .112*** 12.50 .104*** 12.28
    Labor force or school .096*** 11.18 -.033** -3.52
    Occupational prestige .037*** 4.70 .026** 2.92
    Individual income .026** 3.34 .028** 2.91
    Married -.040*** -6.01 .003 .40
    Number of children -.002 -.22 -.021* -2.18
    Church attendance -.113*** -14.59 -.107*** -12.22
    Intercept -.554*** -17.69 -.764*** -15.01
Level 2
    Cohort .020*** 13.46 .017*** 7.94
    Cohort2 (x .01) -.013*** -7.52 -.007** -3.06
Variance Components
Level 1 .42812 .39772
Level 2 intercept .00154** .00138**
Level 1 df 11,833 9,084
Level 2 df 83 83
Open in a new tab

Notes: Models control for year, race, region and city size.

+p < .10

*

p < .05

**

p < .01

***

p < .001

Table 1 also includes the effects of cohort on the intercept or adjusted GE mean. For both men and women, the positive coefficient for cohort and the negative coefficient for cohort squared reflect a rise and slight leveling off of GE across birth cohorts. However, the cohort-based curve rises more quickly and levels off more for women than men. These effects indicate increasing GE when controlling for compositional changes and thus favor the value shift argument over the structural position argument. In addition, unlisted coefficients for the level-1 year dummy variables, which represent within cohort change (Firebaugh 1997), reveal a pattern of increase and decline when controlling for composition differences across time.

Table 2 presents the cross-level interaction coefficients that show how the slopes of key determinants change across cohorts. The results support the diffusion hypothesis with respect to education, labor force status and income for women but not for men. Family variables show less consistent change in their influence on GE and less consistent support for the diffusion hypothesis, but religion attendance proves more important. Consider each in turn.

Table 2.

Cross-Level Interaction Coefficients and t-Ratios from Multilevel Models of the Gender Egalitarianism Scale

Level-2 Predictors Variance Component
Cohort Cohort2 Without Cohort With Cohort Proportion Reduction
b t b t
Female Slopes
Education .003+ 1.94 -.004* -2.38 .00126** .00074** .412
Labor force or school .007** 3.30 -.006** -2.96 .00099** .00061 .384
Homemaker -.008*** -5.47 .008*** 5.05 .00157** .00070 .544
Occupational prestige .001 .34 -.001 -.79 .00108* .00110* -.019
Individual income .004+ 1.88 -.004+ -1.87 .00083 .00073 .120
Married -.001 -.58 .001 .65 .00087+ .00090* -.034
Number of children -.002 -1.40 .001 1.01 .00052 .00045 .135
Church attendance -.003* -2.22 .002 1.36 .00118** .00092* .220
Male Slopes
Education .002 1.47 -.002 -1.31 .00009 .00010 -.111
Labor force or school -.001 -.60 .001 .49 .00004 .00004 .000
Homemaker -.003+ -1.80 .003* 1.99 .00047 .00039 .170
Occupational prestige .002 1.37 -.003 -1.56 .00061 .00075 -.230
Individual income .002 .88 -.002 -.72 .00121** .00126** -.041
Married .001 .45 -.001 -.27 .00029 .00034 -.172
Number of children -.003+ -1.74 .003 1.57 .00065* .00064* .015
Church attendance -.005** -2.71 .004* 2.16 .00189** .00117* .381
Open in a new tab

Notes: Each row of coefficients comes from a separate model.

Models control for variables in Table 1 plus year, race, region and city size.

+

p<.1

*

p<.05

**

p < .01

***

p < .001

Education

Changes in the effects of education, as represented by the cohort and cohort squared interaction terms in Table 2, fit the predicted curvilinear pattern specified by the diffusion hypothesis for females but less clearly for males. For females, the cohort interaction term is positive (significant at .10) and the cohort-squared term is negative (significant at .05). For males, the coefficients have the expected signs but do not reach statistical significance. Figure 1a depicts the predicted female and male slope of education for each cohort as implied by the interaction model. For females, education has a small positive effect for the oldest cohorts (the coefficient equals about .085). The effect rises to a peak of .136 for the cohort born in 1934, thus indicating greatest divergence in views across education groups among Depression-era cohorts that reached adulthood in the 1950s. The effect of education declines afterward to a low of about .028, reflecting the greater adoption of gender egalitarianism across all education levels among the baby-boom generations. Among cohorts born in the 1970s and 1980s, the near zero effects of education indicate that more and less educated women differ little their in views.

Figure 1.

Figure 1

Figure 1

Open in a new tab

Predicted Effects by Cohort

For males, a curvilinear pattern of cohort change in the education slope also appears, but it does not reveal as much convergence as for females (i.e., the effects remain more positive among recent cohorts). The effects of education are smaller for males than females across most cohorts, show a more modest reversal across cohorts, and are not significantly influenced by cohort and cohort squared – all evidence that diffusion has proceeded less far among men than women. Whereas the educational gap in GE declines substantially for the youngest cohorts of women, it declines only modestly for men.

To gauge the impact of cohort differences on the education slopes, Table 2 presents another statistic. It lists the level-2 variance in the education slopes without the cohort determinants, the variance in the slopes after accounting for the influence of the cohort determinants, and the reduction in the variance due to the cohort determinants. The variance for females drops by 41 percent when including the cohort measures but does not improve for males.5 Again, diffusion of gender equality across cohorts appears most clearly for women.

Labor Force/School

With labor force liberally defined to include those unemployed and going to school as well as with a full- or part-time job, Table 2 compares changes in the effects of the dummy variable across cohorts. For women but not men, cohort and cohort squared significantly influence the slope. Consistent with the results for education, the cohort interaction term has a positive sign and the cohort-squared interaction term has a negative sign. Including the cohort variables reduces the variance of the slopes by 38 percent for women. In the curve implied by the interaction coefficients (Figure 1b), the effect of labor force/school for women begins below zero,6 rises to a peak of .115 for the cohort born in 1956, and then declines for younger cohorts to near zero again (.063). For males, however, work has little influence, and its effect fails to change significantly across cohorts.

As a check on the results, a dummy variable measuring keeping home relative to all other work categories replaces the labor force/school dummy variable in the model. Table 2 lists these results for women and men (though the category has little importance for men). Consistent with the effects for labor force/school, the effect of keeping home on GE first becomes more negative, thus reflecting divergence in views, perhaps even a backlash to changes, before reversing and becoming less negative. Figure 1c plots the curve for the effect among females. It begins near zero but becomes increasingly negative as less egalitarian views of women homemakers contrast more greatly with those of other women. After its low point of -.126 in the 1947 cohort, the effect becomes increasingly less negative as homemakers adopt the more egalitarian views of others and for the youngest cohorts equals only -.008. The male curve also changes but by a small amount. Given the ambiguous meaning of homemaking for men and the small number of cases in the category, the results have limited value.

Prestige

The interaction of occupational prestige with cohort and cohort squared fails to reach statistical significance for either men or women. Figure 1d accordingly reveals only a weak curvilinear pattern of effects.

Income

The coefficients for personal income indicate significant curvilinear patterns of change for women but not for men. As illustrated in Figure 1e, the effect of income for women is negative for the 1900 cohort, becomes positive for the 1923 cohort, and peaks for the 1952 cohort. Afterward it declines and falls to levels slightly below zero. Although not large, the differences in support of gender egalitarianism by income levels among females rise and fall across cohorts. Cohort changes in the effects of income prove weaker for males.

Marriage and Number of Children

Although being married and having many children might be expected to define a dimension of cohort change in GE, neither interacts significantly with cohort and cohort squared in Table 2. Recall in Table 1 that marriage reduces GE among women and the number of children reduces GE among men. For both these determinants, however, the effects remain largely constant across cohorts (see figures 1f and 1g).

Church Attendance

The interaction coefficients reach significance for both the cohort and cohort squared terms among men but only for the cohort term among women. As illustrated in Figure 1h, however, both genders show similar changes in the slopes for church attendance. The effect begins near zero for the oldest cohorts and becomes more negative for new cohorts but do not show the convergence found for SES determinants. More so than for other cleavages, church attendance remains a source of divergent views on GE among women and men.

Sensitivity Checks

Tests for interactions can prove sensitive to model specification, but the results appear sufficiently consistent across multiple variables as to lend confidence to the findings. Further, a check of the scatterplots of cohort by the level-2 empirical Bayes residuals for the education slope illustrates the substantially greater variation across cohorts in education effects among women than men but otherwise reveals little in the way of non-randomness.

Other checks on the sensitivity of the results to influential cases strengthen the findings. In relation to the female education interaction models, for example, deleting the five cohorts with the largest average dfbeta values does little to change the effects on the education slopes of the level-2 cohort and cohort squared determinants. Similarly, deleting the five earliest and five latest cohorts changes the level-2 estimates only trivially. Still further, the results appear similar when using dummy variables rather than quadratic terms to represent cohort differences.

Alternative Contextual Measure

Although the level-2 contextual results depend on a simple measure of the year of birth, the measure is well suited for capturing the time-dependent process of diffusion. To extend this logic, a measure of aggregate gender equality for each cohort should have similar curvilinear effects on the cohort-specific slopes for gender egalitarianism. As support for gender equality increases within cohorts, the influence of individual determinants should rise and decline. To check, I calculated the mean gender equalitarianism from the GSS for each cohort and used it as a quadratic in the level-2 model for the education slopes (available on request). The mean has a significant positive effect on the education slope and the mean squared has a significant negative effect, again supporting the diffusion hypothesis.7 For cohorts with low levels of gender egalitarianism, education has little influence, as few accept the tenets of equal treatment of men and women. As levels of gender egalitarianism increase initially and some groups adopt gender egalitarianism, the divisions in views and the effects of education increase. At the highest levels of gender egalitarianism, social divisions in views largely disappear and the effect of education declines to near zero.

Conclusion

Based on the analysis of the GSS, the adoption of gender egalitarian views occurs steadily and persistently across cohorts over the 20th century. Less obviously, however, the results also show that adoption reflects non-linear changes in the influence of the determinants and the strength of social cleavages. The effects of predictors become stronger across cohorts as attitudes shift from largely unfavorable toward gender equality to favorable among women with greater education and higher commitment to work. Among the most recent cohorts, the effects of the predictors become weaker as favorable attitudes spread widely through the population and group differences decline in importance. The results support and extend the diffusion arguments and over-time comparisons made by Fischer and Hout (2006).

The evidence proves strong and robust for women but less supportive for men. Men not only hold less egalitarian views than women but also show less responsiveness to social change. For example, the effects of education on gender egalitarianism increase and decrease for men as for women but by small and statistically insignificant amounts. Gender egalitarianism among recent cohorts thus depends more on level of education for men than women. This difference in patterns of change suggests that diffusion has moved faster and farther among women than men and qualifies theoretical claims about diffusion. The diffusion process best fits groups with stronger interests in equality.

Theories of both structural position and value change, though incomplete on their own, play a role in explaining the pattern of results. Structural changes that increase education, work, and career opportunities for women provide the impetus to adopt egalitarianism by increasing direct interests in equality among certain segments of the population. Value changes among other groups of women with less commitment to work and career (but still some interests in gender equality) occur later. According to the diffusion theory, both types of change contribute to the rising support for gender equality over the last century.

The approach used here to specify and test hypotheses about sources of change in gender egalitarianism has several advantages. Examining differences across cohorts rather than years, modeling the varying effects of the determinants of gender egalitarianism rather than just the level of gender egalitarianism, and testing non-linear predictions of an integrative diffusion theory all help extend the literature in new directions. Previous studies based on cross-period comparisons of the GSS offer valuable insights but find general stability in determinants of gender egalitarianism (e.g., Bolzendahl and Myers 2004). Making comparisons across cohorts builds on other studies (Brewster and Padovic 2000; Brooks and Bolzendahl 2004; Ciabattari 2001; Schnittker, Freese and Powell 2003; Wilkie 1993) but also more fully exploits the potential for change to occur across groups that are born and socialized in different historical periods. This empirical approach is well suited to testing a theory based on diffusion of ideas.

However, the available GSS data used to evaluate the approach face some limitations and suggest needs for additional research. First, the data allow only for the indirect study of diffusion. The patterns of change in determinants are consistent with initial adoption of gender egalitarianism by innovative groups and the later diffusion to other groups. Indeed, the predictions of non-linear interactions of cohort and the determinants of gender egalitarianism are highly falsifiable. Yet, other types of data and forms of analysis are needed to more directly validate claims about how beliefs and values change.

Second, the inability to measure SES and family characteristics of older cohorts at the time of young adulthood rather than at older ages compromises the tests. Current education reflects past education better than current employment, job characteristics, income, marital status, children and church attendance reflect past work and family status, especially for older cohorts. That the results prove consistent despite this measurement error is reassuring. Still, the ability to measure only current characteristics rather than past characteristics remains a limitation. Further, selective mortality may bias estimates of the determinants. To minimize this problem, additional research focusing on cross-sectional differences across nations at varied stages of movement toward gender equality can supplement over-time comparisons within a single nation.

Third, using a contextual measure of birth year (and secondarily the mean of gender egalitarianism) represents only a first step in understanding the sources of change in gender egalitarianism. The first step proves valuable in offering supportive evidence, but better measures of historical changes in the structural positions of women over the past century can extend the approach. Gathering cohort-based measures of education, work, and earning opportunities of women from historical sources and matching them to cohorts present a daunting task. Even so, future research can use such data to provide additional tests of predictions that the context of gender equality relates nonlinearly to the effects of the determinants of gender egalitarianism. With more direct measures of gender equality and access of women to school, work and good jobs, the effects of the determinants of gender egalitarianism should increase and decrease in much the same way they do for the cohort measure.

Acknowledgments

This research received support from the National Science Foundation, grant SES 0848623. I thank Isaac Reed, anonymous reviewers and the editor for helpful comments.

Footnotes

1

Four other measures of gender egalitarianism available from 1973 to 1998 prove less suitable for the analysis. The questions ask about leaving running the country to men, approving of married woman earning money in business or industry, voting for a woman for President, and men being better suited emotionally for politics than women. While overlapping with items used in the analysis, the four measures truncate the range of cohorts, fail to capture change among more recent cohorts, and do less well to test convergence predictions.

2

Net of the cohort and year variables, age has little variation and little influence on gender egalitarianism.

3

A measure of family income does less well to predict gender egalitarianism and its changes across cohorts.

4

I also replicated key models with a technique developed by Yang and Land (2006) that allows for age, period and cohort effects and is based on a cross-classification multilevel design. This method gives much the same results as reported in the tables.

5

Chi-square tests of significance of the variance components included in the table provide only approximate probabilities (Raudenbush and Bryk 2002).

6

The negative impact of work in the oldest cohorts likely reflects the divergence between current and past work among older persons. Although more accurate relationships would appear with measures of work history, the pattern of results is similar to that for education.

7

The mean of gender egalitarianism has little importance as a predictor of the intercept, merely showing that the mean for a cohort predicts individual levels within the cohort. However, the mean has more value in predicting the effect of education on gender egalitarianism. It shows that the context of gender equality helps define the nature of group divisions in views on gender egalitarianism.

References

  1. Alwin Duane F., Braun Michael, Scott Jacqueline. The Separation of Work and the Family: Attitudes towards Women's Labour-Force Participation in Germany, Great Britain, and the United States. European Sociological Review. 1992;8(1):13–37. [Google Scholar]
  2. Amato Paul R., Booth Alan. Changes in Gender Role Attitudes and Perceived Marital Quality. American Sociological Review. 1995;60(1):58–66. [Google Scholar]
  3. Banaszak Lee Ann, Plutzer Eric. The Social Bases of Feminism in the European Community. Public Opinion Quarterly. 1993a;57(1):29–53. [Google Scholar]
  4. Banaszak Lee Ann, Plutzer Eric. Contextual Determinants of Feminist Attitudes: National and Subnational Influences in Western Europe. American Political Science Review. 1993b;87(1):145–57. [Google Scholar]
  5. Baxter Janeen, Kane Emily W. Dependence and Independence: A Cross-National Analysis of Gender Inequality and Gender Attitudes. Gender and Society. 1995;9(2):193–215. [Google Scholar]
  6. Bolzendahl Catherine I., Myers Daniel J. Feminist Attitudes and Support for Gender Equality: Opinion Change in Women and Men 1974-1998. Social Forces. 2004;83(2):759–89. [Google Scholar]
  7. Bolzendahl Catherine, Brooks Clem. Women's Political Resources and Welfare State Spending in 12 Capitalist Democracies. Social Forces. 2007;85(4):1509–34. [Google Scholar]
  8. Bourdieu Pierre, Nice Richard. Distinction: A Social Critique of the Judgment of Taste. Harvard University Press; 1994. [Google Scholar]
  9. Brewster Karin, Padavic Irene. Change in Gender Ideology, 1977-1996: The Contributions of Intracohort Change and Population Turnover. Journal of Marriage and the Family. 2000;62(2):477–87. [Google Scholar]
  10. Brooks Clem. Civil Rights Liberalism and the Suppression of a Republican Political Realignment in the United States, 1972-1996. American Sociological Review. 2000;65(4):482–505. [Google Scholar]
  11. Brooks Clem, Bolzendahl Catherine. The Transformation of U.S. Gender Role Attitudes: Cohort Replacement, Social-Structural Change, and Ideological Learning. Social Science Research. 2004;33(1):106–33. [Google Scholar]
  12. Carter J. Scott, Corra Mamadi, Carter Shannon K. The Interaction of Race and Gender: Changing Gender-Role Attitudes, 1974-2006. Social Science Quarterly. 2009;90(1):196–211. [Google Scholar]
  13. Cha Youngjoo, Thébaud Sarah. Labor Markets, Breadwinning, and Beliefs: How Economic Context: Shapes Men's Gender Ideology. Gender and Society. 2009;23(2):215–43. [Google Scholar]
  14. Charles Maria, Bradley Karen. Equal but Separate? A Cross-National Study of Sex Segregation in Higher Education. American Sociological Review. 2002;67(4):573–99. [Google Scholar]
  15. Cherlin Andrew, Barnhouse Walters Pamela. Trends in United States Men's and Women's Sex-Role Attitudes: 1972 to 1978. American Sociological Review. 1981;46(4):453–60. [Google Scholar]
  16. Ciabattari Teresa. Changes in Men's Conservative Gender Ideologies: Cohort and Period Influences. Gender Society. 2001;15(4):574–91. [Google Scholar]
  17. Clark Roger, Ramsbey Thomas W., Stier Adler Emily. Culture, Gender, and Labor Force Participation: A Cross-National Study. Gender and Society. 1991;5(1):47–66. [Google Scholar]
  18. Crompton Rosemary, Harris Fiona. Women's Employment and Gender Attitudes: A Comparative Analysis of Britain, Norway and the Czech Republic. Acta Sociologica. 1997;40(2):183–202. [Google Scholar]
  19. Davis James A., Smith Tom W., Marsden Peter V. General Social Surveys, 1972-2006: Cumulative Codebook. CD Rom. NORC; Chicago: 2007. [Google Scholar]
  20. Davis Nancy J., Robinson Robert V. Men's and Women's Consciousness of Gender Inequality: Austria, West Germany, Great Britain, and the United States. American Sociological Review. 1991;56(1):72–84. [Google Scholar]
  21. Firebaugh Glenn. Where Does Social Change Come From? Population Research and Policy Review. 1992;11(1):1–20. [Google Scholar]
  22. Firebaugh Glenn. Analyzing Repeated Surveys. Sage Publications; 1997. [Google Scholar]
  23. Fischer Claude S. Urban-to-Rural Diffusion of Opinions in Contemporary America. American Journal of Sociology. 1978;84(1):151–59. [Google Scholar]
  24. Fischer Claude S., Hout Michael. Century of Differnce: How American Changed in the Last One Hundred Years. Russell Sage Foundation; 2006. [Google Scholar]
  25. Glass Jennifer. Housewives and Employed Wives: Demographic and Attitudinal Change 1972-1986. Journal of Marriage and the Family. 1992;54(3):559–69. [Google Scholar]
  26. Goldscheider Frances, Oláh Livia Sz., Puur Allan. Reconciling Studies of Men's Gender Attitudes and Fertility: Response to Westoff and Higgins. Demographic Research. 2010;22(8):189–98. [Google Scholar]
  27. Hout Michael. Getting the Most Out of the GSS Income Measures. GSS Methodological Report101. 2004 Available at: http://publicdata.norc.org:41000/gss/DOCUMENTS/REPORTS/Methodological_Reports/MR101.pdf.
  28. Huber Joan. Macro-Micro Links in Gender Stratification: 1989 Presidential Address. American Sociological Review. 1990;55(1):1–10. [Google Scholar]
  29. Huber Joan, Spitze Glenna. Wives’ Employment, Household Behaviors, and Sex-Role Attitudes. Social Forces. 1981;60(1):150–69. [Google Scholar]
  30. Inglehart Ronald. Culture Shift in Advanced Industrial Society. Princeton University Press; 1989. [Google Scholar]
  31. Inglehart Ronald, Norris Pippa. Rising Tide: Gender Equality and Cultural Change. Cambridge University Press; 2003. [Google Scholar]
  32. Inglehart Ronald, Baker Wayne E. Modernization, Cultural Change, and the Persistence of Traditional Values. American Sociological Review. 2000;65(1):19–51. [Google Scholar]
  33. Jackson Robert Max. Destined for Equality: The Inevitable Rise of Women's Status. Harvard University Press; 1998. [Google Scholar]
  34. Kane Emily, Sanchez Laura. Family Status and Criticism of Gender Inequality at Home and at Work. Social Forces. 1994;72(4):1079–102. [Google Scholar]
  35. Kaufman Gayle. Do Gender Role Attitudes Matter? Journal of Family Issues. 2000;21(1):128–34. [Google Scholar]
  36. Mason Karen Oppenheim, Czajka John L., Arber Sara. Change in U.S. Women's Sex-Role Attitudes 1964-1974. American Sociological Review. 1976;41(4):573–96. [PubMed] [Google Scholar]
  37. Mason Karen Oppenheim, Lu Yu-Hsia. Attitudes toward Women's Familial Roles: Changes in the United States 1977-1985. Gender and Society. 1988;2(1):39–57. [Google Scholar]
  38. Montgomery Mark R., Casterline John B. The Diffusion of Fertility Control in Taiwan: Evidence from Pooled Cross- Section Time-Series Models. Population Studies. 1993;47(3):457–79. doi: 10.1080/0032472031000147246. [DOI] [PubMed] [Google Scholar]
  39. Morgan Carolyn Stout, Walker Alexis J. Predicting Sex Role Attitudes. Social Psychology Quarterly. 1983;46(2):148–51. [Google Scholar]
  40. NORC General Social Survey. 2008 Available at: http://www.norc.org/projects/General+Social+Survey.htm.
  41. Norris Pippa. Politics and Sexual Equality: The Comparative Position of Women in Western Democracies. Rienner; 1987. [Google Scholar]
  42. Oppenheimer Valerie Kincade. The Female Labor Force in the United States: Demographic and Economic Factors Governing Its Growth and Changing Composition. Greenwood Publishing Group; 1976. [Google Scholar]
  43. Petersen Larry R., Donnenwerth Gregory V. Religion and Declining Support for Traditional Beliefs about Gender Roles and Homosexual Rights. Sociology of Religion. 1998;59(4):353–71. [Google Scholar]
  44. Plutzer Eric. Work Life, Family Life, and Women's Support of Feminism. American Sociological Review. 1988;53(4):640–49. [Google Scholar]
  45. Poole Keith T., Zeigler L. Harmon. The Diffusion of Feminist Ideology. Political Behavior. 1981;3(3):229–56. [Google Scholar]
  46. Raudenbush Stephen W., Bryk Anthony S. Hierarchical Linear Models. Second Edition Sage Publications; 2002. [Google Scholar]
  47. Raudenbush Stephen W., Bryk Anthony S., Fai Cheong Yuk, Congdon Richard. HLM 6: Hierarchical Linear and Nonlinear Modeling. Scientific Software International; Chicago: 2004. [Google Scholar]
  48. Rindfuss Ronald R., Brewster Karen L., Kavee Andrew L. Women, Work and Children: Behavioral and Attitudinal Change in the United States. Population and Development Review. 1996;22(3):457–82. [Google Scholar]
  49. Rogers Everett M. Diffusion of Innovations. Fifth Edition Free Press; 2003. [Google Scholar]
  50. Schnittker Jason, Freese Jeremy, Powell Brian. Who Are Feminists and What Do They Believe? The Role of Generations. American Sociological Review. 2003;68(4):607–22. [Google Scholar]
  51. Scott Jacqueline, Alwin Duane F., Braun Michael. Generational Changes in Gender-Role Attitudes: Britain in a Cross-National Perspective. Sociology. 1996;30(3):471–92. [Google Scholar]
  52. Simmel Georg. Fashion. In: Levine Donald., editor. Georg Simmel: On Individuality and Social Forms. University of Chicago Press; 1971[1904]. pp. 294–323. [Google Scholar]
  53. Smith Tom W. Working Wives and Women's Rights: The Connection between the Employment Status of Wives and the Feminist Attitudes of Husbands. Sex Roles. 1985;12(5/6):501–08. [Google Scholar]
  54. Spitze Glenna, Huber Joan. Changing Attitudes toward Women's Nonfamily Roles: 1938 to 1978. Work and Occupations. 1980;7(3):317–35. [Google Scholar]
  55. Strang David, Meyer John W. Institutional Conditions for Diffusion. Theory and Society. 1993;22(4):487–511. [Google Scholar]
  56. Strang David, Soule Sarah A. Diffusion in Organizations and Social Movements: From Hybrid Corn to Poison Pills. Annual Review of Sociology. 1998;24:265–90. [Google Scholar]
  57. Thornton Arland, Young-DeMarco Linda. Four Decades of Trends in Attitudes toward Family Issues in the United States: The 1960s through the 1990s. Journal of Marriage and the Family. 2001;63(4):1009–37. [Google Scholar]
  58. Treas Judith, Widmer Eric D. Married Women's Employment over the Life Course: Attitudes in Cross-National Perspective. Social Forces. 2000;78(4):1409–36. [Google Scholar]
  59. Veblen Thorstein. Theory of the Leisure Class. Transaction; 1992[1899]. [Google Scholar]
  60. Warner Rebecca L. Does the Sex of Your Children Matter? Support for Feminism among Women and Men in the United States and Canada. Journal of Marriage and the Family. 1991;53(4):1051–56. [Google Scholar]
  61. Weber Max, Gerth HH, Wright Mills C. From Max Weber: Essays in Sociology. Oxford University Press; 1958. [Google Scholar]
  62. Wejnert Barbara. Integrating Models of Diffusion of Innovations: A Conceptual Framework. Annual Review of Sociology. 2002;28:297–326. [Google Scholar]
  63. Wilkie Jane Riblett. Changes in U.S. Men's Attitudes toward the Family Provider Role, 1972-1989. Gender and Society. 1993;7(2):261–79. [Google Scholar]
  64. Yang Yang, Land Kenneth C. A Mixed Models Approach to the Age-Period-Cohort Analysis of Repeated Cross-Section Surveys, With an Application to Data on Trends in Verbal Test Scores. Sociological Methodology. 2006;36(1):75–97. [Google Scholar]
  65. Zuo Jiping, Tang Shengming. Breadwinner Status and Gender Ideologies of Men and Women Regarding Family Roles. Sociological Perspectives. 2000;43(1):29–43. [Google Scholar]

RESOURCES