Benevolent Sexism and Hostile Sexism Across the Ages

Abstract

Ambivalent sexism theory states that prejudice toward women comprises two interrelated ideologies. Endorsement of hostile sexism—aggressive and competitive attitudes toward women—is linked with endorsement of benevolent sexism—paternalistic and patronizing attitudes toward women. We conduct the first systematic tests of how endorsement of sexism differs across age and across time, using six waves of a nationally representative panel sample of New Zealand adults (N = 10,398). Results indicated U-shaped trajectories for men’s endorsement of hostile sexism, women’s hostile sexism, and women’s benevolent sexism across the life span. However, over time, endorsement of these sexist attitudes tended to decrease for most ages. In contrast, men’s benevolent sexism followed a positive linear trajectory across age and tended not to change over time. These results provide novel evidence of how ambivalent sexism differs across age and highlight that benevolent sexism is particularly tenacious.

Keywords

benevolent sexism hostile sexism ambivalent sexism development latent growth model

Ambivalent sexism theory (Glick & Fiske, 1996) states that prejudice toward women is distinctive because it comprises two related ideologies with ambivalent tones. Hostile sexism is an ideology which characterizes women as incompetent, overly emotional, and attempting to manipulate men to gain power. However, hostile sexism is not an effective way to maintain men’s power because it impedes the fulfilment of heterosexual men’s intimacy needs (Glick & Fiske, 1996). Benevolent sexism is an ideology that tempers these relational costs, while maintaining men’s power, by characterizing women as delicate, pure and in need of men’s protection and care. Existing research primarily focuses on the consequences of endorsing hostile sexism and benevolent sexism, particularly as a predictor of gender inequalities (for reviews, see Connor, Fiske, & Glick, 2016; Hammond & Overall, 2016), but only theorizes about how endorsement of sexism differs across the life span (see Glick & Fiske, 1996, 2001; Glick & Hilt, 2000). The current research utilizes a nationally representative longitudinal sample to model (1) how the endorsement and interrelation of sexist attitudes differ across age and (2) age-cohort changes in the endorsement of sexism over a 5-year span.

Ambivalent Sexism and the Structure of Sexist Attitudes

A fundamental principle of ambivalent sexism theory is that sexist attitudes encompass two positively associated ideologies that both function to maintain gender inequality (Glick & Fiske, 1996). Men and women who endorse hostile sexism, derogatory attitudes toward women who could challenge men’s power, also tend to endorse benevolent sexism, subjectively positive expressions toward women in traditional roles (Glick & Fiske, 1996). Sexist attitudes have three theorized sources that are rooted in the societal and biological differences between men and women (Glick & Fiske, 1996). First, sexist attitudes justify men’s power by expressing that men are dominant over women (hostile sexism) and simultaneously responsible to protect and provide for women (benevolent sexism). Second, sexist attitudes justify distinctiveness of gender identity by prescribing that women should be warm and docile, whereas men should be men competent and composed. Third, sexist attitudes encourage intimacy by idealizing women’s relational capacity to “complete” men (benevolent sexism) but temper women’s interpersonal power by vilifying women who might manipulate men’s relational needs (hostile sexism).

In tandem, sexist ideologies function to maintain gender inequality across societies (Brandt, 2011; Glick & Fiske, 1996; Glick et al., 2000), such as by discriminating against or intimidating women in career or political domains (Connor et al., 2016) and rewarding and praising women who adopt traditional gender roles (Hammond & Overall, 2016). Attention is often directed to the ways that benevolent sexism undermines women’s resistance to societal gender inequalities (e.g., Becker & Wright, 2011; Hammond & Sibley, 2011) and encourages women to invest in their relationships rather than pursue career or educational goals (e.g., Chen, Fiske, & Lee, 2009; Montañés, de Lemus, Moya, Bohner, & Megías, 2013). However, none of the evidence linking sexist attitudes to markers of gender inequality have investigated how sexist attitudes differ across the life span or over time. Thus, the literature on sexist attitudes lacks a contextual map of the ages for which endorsement of sexism (and its effects) are most prominent or the ages for which endorsement of sexism changes. In the current research, we begin to build this map by modeling differences in the endorsement of benevolent sexism and hostile sexism across age, and further, modeling the 5-year change in endorsement of sexist attitudes.

The Expected Trajectory of Sexist Attitudes Across Age

We established exploratory predictions how sexist attitudes might differ across the average life span based on the normative age-graded concerns that are relevant to the theorized sources of sexism we reviewed above—power, identity, and relational needs. Normative age-graded events are influences on development that occur in similar ways and at similar life stages for people in a given culture, including beginning a career or committing to a serious relationship (Baltes, Reese, & Lipsitt, 1980; also see Elder, 1995; Heckhausen, 2006, for similar paradigms). In younger adulthood, normative goals include the attainment of personal status and power, establishment of a social identity and career, and serious intimate relationships (Arnett, 2000; Baltes et al., 1980; Casey, Jones, & Somerville, 2011; Takahashi, 2005). By contrast, in middle adulthood, people’s traits and roles are less likely to follow societal norms (Bedford & Turner, 2006; O’Neil & Egan, 1992), and their romantic relationships tend to be stable but less central to the self (Chopik, Edelstien, & Fraley, 2013). However, in older ages, traditional and conservative values are relatively high (Gouveia, Vione, Milfont, & Fischer, 2015), and people experience high dependence on romantic partners (Wrzus, Hänel, Wagner, & Neyer, 2013). In sum, across the life span, concerns over power, identity, and dependence follow a high–low–high course. We therefore developed tentative expectations of U-shaped trajectories of sexism across age: relatively high endorsement of sexism in early adulthood, relative low endorsement in middle adulthood, and relatively high endorsement in late adulthood.

A U-shaped pattern of endorsement of hostile sexism and benevolent sexism across age is consistent with demographic data from two large Spanish samples. Endorsement of hostile sexism and benevolent sexism was relatively high in the youngest and oldest age groups relative to middle-aged age groups (Fernández, Castro, & Lorenzo, 2004, N = 1,003; Gariagordobil & Aliri, 2013, N = 5,110). We extend this prior research by testing whether this pattern is curvilinear, therefore providing the first formal test of whether endorsement of sexism follows a U-shaped trajectory across age. Furthermore, we employ novel methodologies to examine further characteristics of sexist attitudes: We (1) conduct age-cohort-based longitudinal tests over a 5-year span to examine how endorsement of sexism changes longitudinally, (2) model the relationship between hostile sexism and benevolent sexism across age (discussed next), and (3) estimate exploratory cohort effects by comparing the age-based trajectory of sexism with the model of longitudinal change in sexism.

The Expected Ambivalence of Sexist Attitudes Across Age

We examined age-related differences in the association between benevolent sexism and hostile sexism because this association is at the heart of ambivalent sexism theory; sexism is proposed to be ambivalent because of the interrelatedness of benevolent sexism and hostile sexism (Glick & Fiske, 1996; Glick et al., 2000). Benevolent sexism is a “crucial complement to hostile sexism” because subjectively positive praise for women in traditional roles is used to rationalize aggression toward nontraditional women (Glick & Fiske, 2001, p. 111). However, prior research has illustrated variability in this association. Benevolent sexism and hostile sexism appear to be more strongly related for women than for men and for younger samples than for older samples (Glick & Fiske, 1996; Glick et al., 2000). However, no research has specifically tested differences in the association between benevolent sexism and hostile sexism across age. Following the prediction that people’s sexist beliefs become more individuated and idiosyncratic across the life span (Glick & Fiske, 1996, p. 508), we expected that the association between benevolent sexism and hostile sexism will weaken across age.

Current Research

Our study tested a nationally representative, longitudinal panel sample in an egalitarian country to examine the age differences, relationships between, and changes in endorsement of benevolent sexism and hostile sexism. We conducted four sets of analyses. First, age-based latent growth models estimated the linear and curvilinear trajectories of benevolent sexism and hostile sexism across age for men and for women. Second, we estimated the standardized relationship between benevolent sexism and hostile sexism, adjusting for gender, across age. Third, we estimated the 5-year change in sexism in multigroup cohort-sequential latent growth models. Finally, we conducted tentative estimation of cohort effects by comparing effects between the age-based model and cohort-based model of sexism.

Method

Participants

Eligible participants were 10,398 people (62.5% women) who responded to at least three out of the first six waves of the New Zealand Attitudes and Values Study (NZAVS), an annual longitudinal panel survey. Average age at Wave 1 (2009) was 49.45 (SD = 14.98; see Table 1). At least three waves of completion were necessary for participants’ data to contribute information about the rate of change in sexism and meet the requirements to fit latent growth models. The majority of participants (85.7%) were New Zealand European. See the Online Supplemental Materials and Satherley et al. (2015) for detailed demographics of the sampling procedures, sample characteristics, and how these characteristics relate to attrition.

Table 1.
The Number of Participants in Each of the Sequential 5-Year Birth Cohorts Organized for the Multigroup Latent Growth Models.

Birth Cohort N Women N Men Ages Sampled (T1–T5)

1940 and older 198 197 69–74

1941–1945 363 330 64–69

1946–1950 637 509 59–64

1951–1955 744 492 54–59

1956–1960 803 497 49–54

1961–1965 794 488 44–49

1966–1970 721 366 39–44

1971–1975 644 284 34–39

1976–1980 491 206 29–34

1981–1985 356 104 24–29

1986–1990 381 165 19–24

1991 and younger 186 63 18–23

Measures

Benevolent sexism (e.g., “Women should be cherished and protected by men”) and hostile sexism (e.g., “Women are too easily offended”) were assessed at each wave using a 10-item short form of the Ambivalent Sexism Inventory (Glick & Fiske, 1996; 1 = strongly disagree; 7 = strongly agree).

Attrition

Over 60% of participants were retained from Wave 1, and approximately 80% of participants were retained wave to wave (also see Satherley et al. 2015). We examined attrition by comparing eligible participants who had completed Wave 1 with participants who did not meet the inclusion criterion. Eligible participants were more likely to be women (61.1% vs. 55.5%) and tended to be older (M = 49.51, SD = 14.97) than those not selected (M = 44.66, SD = 17.04, d = .30). Eligible participants’ endorsement of benevolent sexism (M = 4.07, SD = 1.17) and endorsement of hostile sexism at Wave 1 (M = 3.38, SD = 1.25) were both relatively lower that those not selected (benevolent sexism M = 4.40, SD = 1.16, d = .285; hostile sexism M = 3.69, SD = 1.31, d = .29).

Results

We first examined descriptive statistics for endorsement of sexism and tested measurement invariance. We then tested our predictions in four sets of multilevel models using MPlus 7.4 (Muthén & Muthén, 1998–2015), with full-information maximum-likelihood (robust standard error) estimation. These models also account for missing data by weighting model parameters so that slopes for participants who completed more waves contribute more to the overall parameter estimates.

Measurement

Descriptive statistics

Descriptive statistics, bivariate correlations, and reliabilities for the measures are presented in Table 2. The test–retest correlations for hostile sexism and benevolent sexism (presented in bold font) indicated relatively high stability for both hostile sexism and benevolent sexism, and measurements at each time point demonstrated consistently acceptable internal reliability.

Table 2.
Bivariate Correlations and Descriptive Statistics for Hostile Sexism and Benevolent Sexism Across the 6 Repeated Assessments From Wave 1 (W1) to Wave 6 (W6) and Age and Gender (Assessed at Wave 1).

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. HS_W1

2. HS_W2 .733

3. HS_W3 .700 .751

4. HS_W4 .682 .734 .765

5. HS_W5 .673 .721 .756 .774

6. HS_W6 .661 .699 .733 .745 .768

7. BS_W1 .430 .366 .346 .341 .355 .349

8. BS_W2 .366 .415 .353 .361 .364 .348 .746

9. BS_W3 .378 .371 .396 .368 .383 .370 .728 .751

10. BS_W4 .343 .352 .354 .417 .383 .371 .709 .752 .755

11. BS_W5 .351 .349 .373 .382 .431 .394 .703 .733 .752 .764

12. BS_W6 .342 .343 .371 .378 .392 .429 .681 .716 .728 .740 .770

13. Gender .252 .226 .262 .261 .236 .222 .118 .108 .134 .129 .141 .147

14. Age −.012 .012 .019 .022 .050 .052 .116 .104 .118 .136 .140 .148 .115**

Mean 3.317 3.212 3.021 3.014 3.023 3.057 4.073 4.030 3.863 3.826 3.791 3.811 – 49.450

SD 1.252 1.227 1.222 1.213 1.204 1.209 1.694 1.153 1.168 1.168 1.190 1.192 – 15.979

Reliability .807 .807 .821 .819 .819 .825 .722 .728 .742 .737 .746 .757 – –

Note. Bolded associations denote associations relevant to assessing the stability of sexism across time. Gender was coded −.5 = women, .5 = men. HS = hostile sexism; BS = benevolent sexism.

**p < .01.

Measurement invariance of sexism

We first tested the possibility that a particular 5-year age cohort (at Wave 1) answered the ambivalent sexism inventory in substantially different ways to other cohorts, or if men answered differently to women. We estimated three increasingly restrictive latent variable models (i.e., configural, metric, and scalar). Configural invariance was specified by restricting item loadings, so that manifest indicators for each sexist ideology loaded on the corresponding latent sexist ideology across all of the age cohorts or across gender. Metric invariance was specified by restricting the item loadings to equality across all of the age cohorts or across gender. Scalar invariance was specified by restricting both item loadings and item intercepts to equality across all of the age cohorts or across gender. As displayed in Table 3, the relative model fit indices demonstrated reasonably good fit for all measurement specifications across age groups and between genders, but minor violations when specifying scalar invariance (ΔCFI > .01; see Cheung & Rensvold, 2002). These results generally indicated that metric measurement invariance best fit the data, suggesting that people generally answered the ambivalent sexism inventory in similar ways, but that any mean differences in sexism across age or gender do not necessarily reflect group differences in the latent variables.

Table 3.
Tests of Strict Measurement Invariance.

Model χ² df RMSEA CFI TLI SRMR

Cross-sectional model

1. Between age groups configural 1,843.104 476 .079 .897 .863 .059

2. Between age groups metric 1,978.531 580 .072 .894 .885 .064

3. Between age groups scalar 2,268.621 684 .071 .880 .890 .068

4. Between gender configural 1,217.897 68 .072 .907 .877 .049

5. Between gender metric 1,253.184 76 .069 .905 .887 .050

6. Between gender scalar 1,594.752 84 .074 .878 .869 .058

Benevolent sexism over time

1. Configural over time 1,756.770 29 .076 .891 .831 .072

2. Metric over time 2,349.492 35 .080 .854 .813 .087

3. Scalar over time 2,467.839 39 .078 .847 .824 .089

4. Between age groups configural 2,578.507 341 .086 .854 .788 .090

5. Between age groups metric 2,732.488 425 .079 .849 .824 .096

6. Between age groups scalar 3,075.099 520 .075 .833 .841 .104

7. Between gender configural 2,136.083 62 .081 .867 .807 .082

8. Between gender metric 2,218.954 74 .075 .863 .833 .084

9. Between gender scalar 3,095.968 88 .082 .807 .803 .103

Hostile sexism over time

1. Configural over time 576.978 29 .043 .966 .948 .027

2. Metric over time 1,171.204 35 .056 .930 .911 .052

3. Scalar over time 1,251.499 39 .055 .926 .914 .053

4. Between age groups configural 1,408.857 341 .060 .934 .904 .056

5. Between age groups metric 1,601.291 425 .056 .927 .915 .064

6. Between age groups scalar 1,866.828 520 .054 .916 .920 .071

7. Between gender configural 1,054.736 62 .056 .936 .907 .048

8. Between gender metric 1,252.362 74 .056 .924 .907 .058

9. Between gender scalar 2,265.928 88 .070 .859 .856 .113

Note. Fit indices are displayed for benevolent sexism and hostile sexism, modeled simultaneously at Wave 1 (cross-sectional model), and separately to test temporal measurement invariance (over time models). All χ² values are significant at p < .05.

RMSEA = Root Mean Square Error of Approximation; SRMR = Standardized Root Mean Square Residual; CFI = Comparative Fit Index; TLI = Tucker Lewis Index.

Temporal measurement invariance of sexism

We next utilized all available waves of data to strict measurement invariance in latent growth models across time. We tested models in three parts, displayed in Table 3, imposing configural, metric, and scalar invariance for each ideology (1) over time, (2) across the 5-year age cohorts (see Table 1) over time, and (3) across gender over time. The measurement specifications identified minor violations of temporal measurement invariance for both benevolent sexism and hostile sexism across time (see Cheung & Rensvold, 2002). Assessing measurement invariance in these longitudinal models places high power demands on the data and, accordingly, should be considered conservative tests of measurement invariance. Nonetheless, these results again indicated metric measurement invariance best fit the data, again highlighting the need for caution in interpreting mean differences in sexism across time or between groups because these differences may not necessarily reflect effects in the latent variables.

Differences in Sexist Attitudes Across Age

Model estimation

We estimated an age-based cohort-sequential latent growth model (see Milojev & Sibley, 2016; Preacher, Wichman, MacCallum, & Briggs, 2008) based on participants’ age at each measurement wave, utilizing exact age at survey completion as an individually varying time indicator, allowing us to utilize data from all waves while accounting for any variation in time between survey completions.¹ We tested for gender differences by estimating a multigroup model that allowed the intercepts and trajectories for men and women to differ. The slopes reported here are a relatively conservative test of the age-related trajectories of sexism because they account for the cohort-by-cohort change across time that we report in the next section. We modeled the differences in benevolent sexism and hostile sexism as a polynomial growth function including linear, quadratic, and cubic components (nonsignificant higher-order interaction terms were retained in the models to most accurately model the trajectories across age).

Results

Parameter estimates for models estimating overall changes in the trajectories of benevolent sexism and hostile sexism from ages 19 to 74 years are presented in Table 4 and Figure 1 (dark line). For men, a significant linear effect indicated that the trajectory of benevolent sexism increased across age. For women, a significant quadratic effect of age suggested that endorsement of benevolent sexism followed a U-shaped trajectory across the life span. Women’s endorsement of benevolent sexism was high in early adulthood, low in middle adulthood, and relatively high again in older ages.

Table 4.
Estimates From Latent Growth Models Estimating the Age-Based Trajectories of Endorsement of Benevolent Sexism (Upper Model) and Hostile Sexism (Lower Model).

95% Confidence Interval

Model Estimate SE z Low High Variances

Benevolent sexism

Men

Intercept 3.924 .024 164.962* 3.878 3.971 1.020*

Age—linear 0.073 .017 4.371* 0.040 0.100 0.394*

Age—quadratic 0.016 .009 1.812 −0.001 0.030 0.038

Age—cubic 0.002 .003 0.890 −0.003 0.007 0.002

Women

Intercept 3.672 .018 199.213* 3.636 3.708 1.020*

Age—linear 0.010 .014 0.742 −0.017 0.038 0.394*

Age—quadratic 0.034 .007 5.111* 0.021 0.047 0.038

Age—cubic −0.003 .002 −1.076 −0.007 0.002 0.002*

Hostile sexism

Men

Intercept 3.437 .027 128.691* 3.385 3.490 1.099*

Age—linear −0.067 .019 −3.603* −0.103 −0.031 0.446*

Age—quadratic 0.033 .010 3.413* 0.014 0.052 0.051

Age—cubic −0.002 .003 −0.531 −0.008 0.004 0.003

Women

Intercept 2.766 .018 151.758* 2.731 2.802 1.099*

Age—linear −0.034 .014 −2.407* −0.062 −0.006 0.446*

Age—quadratic 0.040 .006 6.226* 0.028 0.053 0.051

Age—cubic 0.001 .002 0.212 −0.004 0.005 0.003

Note. N = 10,398 (men = 3,899; women = 6,499) for estimation of benevolent sexism. N = 10,397 (men = 3,899; women = 6,498) for estimation of hostile sexism. For ease of displaying parameters, age (grand-mean centered) was scaled so that each unit represented 10 years (See Online Supplemental Materials for more detail).

p < .05.

Figure 1.
Change trajectories for men’s and women’s endorsement of benevolent sexism (BS; displayed as a dark line in the upper panels) and hostile sexism (HS; displayed as a dark line in the lower panels) from ages 19 to 74. The lines within each 5-year cohort represent longitudinal change in endorsement of BS and HS estimating latent intercepts (i) and slopes (s). p < .05.

Endorsement of hostile sexism followed U-shaped trajectories across age for both men and women. The trajectory of hostile sexism was initially high in early adulthood but decreased into middle adulthood, at which point the trajectory recovered and increased into late adulthood (these results were similar when estimating benevolent sexism and hostile sexism simultaneously as latent variables at Wave 1; see Online Supplemental Materials).

The Association Between Benevolent Sexism and Hostile Sexism Across Age

Model estimation

We estimated whether the association between benevolent sexism and hostile sexism differed across age using data from the first wave, in which hostile sexism was predicted by benevolent sexism (modeled as latent scores), moderated by gender and age (modeled as latent interactions), estimated at every age from 18 to 80 years Parameters were fully standardized based on standard deviations of the predictor and the outcome, so that the standardized effect of benevolent sexism on hostile sexism represented the overall relationship between the two constructs, and any significant interactions indicated that this relationship varied by age or gender.

Results

The significant standardized effect of benevolent sexism displayed in Table 5 indicated that, holding age and gender constant, benevolent sexism and hostile sexism were strongly related. The effect of gender indicated that the relationship between benevolent sexism and hostile sexism was generally stronger for women (β = .528, 95% confidence interval (CI) [.482, .575], p < .001) than for men (β = .496, 95% CI [.434, .559], p < .001). However, for men and women equally, the significant negative effect of age indicated that the association between benevolent sexism and hostile sexism decreased from age 18 to 80 years (Figure 2). The average association between benevolent sexism and hostile sexism was the strongest at 18 years old (β = .600, 95% CI [.530, .670]), weakened linearly as age increased, and was the lowest at 80 years old (β = .452, 95% CI [.378, .526]). These results indicated that the typically strong link between benevolent sexism and hostile sexism is not uniform across age: Ambivalent attitudes were the strongest in young adulthood and relatively weakest, although still strongly related, in older ages.

Table 5.
Standardized Estimates of the Relationship between Latent Benevolent Sexism and Latent Hostile Sexism, Statistically Adjusting for the Random Effects of Gender, Age, and the Interaction Between Gender and Age.

95% Confidence Interval

Predictor β SE t p Low High

Benevolent sexism .528 .024 22.175 <.001 .482 .575

Gender −.032 .015 −2.123 .034 −.062 −.002

Age −.038 .014 −2.659 .008 −.065 −.010

Age × Gender −.002 .003 −0.521 .602 −.007 .004

Note. Gender was coded −.5 = women, .5 = men. Model fit: RMSEA = .072 [.069, .075]; CFI = .883; SRMR = .053. N = 6,466. Model fit estimates did not include the random-effect interaction term.

Figure 2.
The association between benevolent sexism (BS) and hostile sexism (HS) from age 18 to 80, modeled as the latent standardized coefficient at measurement Wave 1, statistically adjusting for gender. Error bars represent 95% confidence interval plotted around stability point estimates.

Longitudinal Change in Sexist Attitudes Within Cohorts

Model estimation

We conducted multigroup cohort-sequential latent growth models by organizing the sample into 12 sequential 5-year birth cohorts (see Table 1) to match the first five waves of annual assessment (see Lucas & Donnellan, 2011). Models utilized the time of survey completion as individually varying time indicators. We estimated a latent intercept and latent linear slope for each of the 5-year age cohorts. Variances and covariance of intercepts and slopes were constrained to equality across birth cohorts.

We organized birth cohorts into consecutive 5-year bands to match the consecutive 5-year spans of the multigroup models, meaning we could arrange the latent growth models in sequence (e.g., the 1986–1990 birth cohort represented change from 19 to 24 years old; the 1981–1985 cohort represented change from 24 to 29 years old, and so on). For each cohort, the youngest age was taken as an indicator of age. This approach simultaneously estimates cross-sectional cohort differences in the latent intercepts, change trajectories within each cohort, and cohort differences in the rate of change, as well as the overall pattern of change that may be observed across the adult life span. We approximated effect sizes by calculating the change as a percentage of the entire scale range (SR%; see Online Supplemental Materials for individual cohorts).

Results

Figure 1 displays the multigroup cohort-sequential latent growth models. Endorsement of benevolent sexism declined across time for men in the youngest cohort (SR% = 0.8%). As age increased, the steepness of the declines in benevolent sexism decelerated and benevolent sexism was relatively stable across all remaining cohorts in the life span, except for a decline in the 49- to 54-year-old cohort (SR% = 0.5%). Men’s endorsement of hostile sexism also decreased across time most sharply in the youngest cohorts (SR% = 1.5–2%), but longitudinal declines in endorsement of hostile sexism were present in most cohorts (SR% = 0.6–0.9%).

For women, endorsement of benevolent sexism declined over time in the age cohorts from 19 to 34, 39 to 44, and 49 to 69 (SR% = 0.3–1.6%) and was otherwise stable. Women’s endorsement of benevolent sexism was high in young adulthood but decreased over time, with the rate of decrease decelerating into middle adulthood. Women’s endorsement of hostile sexism also declined across time in the young adult cohorts (SR% = 0.8–1.2%). In older cohorts, these declines decelerated but the decreases across time remained significant up to the 54 to 59 cohort. In older cohorts, there was no significant change across time and a significant increase across time in the 69 to 74 cohort (SR% = 0.5%).

In sum, endorsement of sexism tended to decrease most prominently in early adulthood. However, unlike the age-based trajectory model, the longitudinal cohort model indicated that endorsement of sexist attitudes tended to decline for most of the age cohorts. This inconsistency indicated the presence of cohort effects which we tested next.

Estimation of Cohort Effects

Model estimation

We examined indicators for the presence of cohort effects by running models that compared estimates from the age-based trajectory model with the cohort-based longitudinal model (also see Milojev & Sibley, 2016). We reanalyzed the cohort-based longitudinal model by adding (1) constraints that equalized the latent intercepts and latent slopes across the cohorts and (2) constraints that further equalized these effects across gender. We then compared the fit of these different models. If the constrained models have substantially worse fit, then this would indicate (1) strong cohort effects and/or (2) between-gender differences in cohort effects. Second, we examined any cohort effects more closely by estimating the discrepancy between the mean levels of sexist attitudes based on the 5-year latent change trajectory with the mean levels of sexist attitudes based on participants’ age at first measurement. This discrepancy identifies ages at which there are differences between the model-implied values that are due to change over time and model-implied values that are due to between-cohort differences.

Results

First, fit estimates for the free versus constrained cohort-based models of sexist attitudes are displayed in Table 6. When effects were free to vary across gender but were constrained across age cohorts, there was very little difference in the fit of the model. This indicated that cohort effects were relatively weak overall. However, model fit worsened when effects were also constrained across gender, indicating relatively greater between-cohort variability for men than for women.

Table 6.
Model Fit Indices Comparing the Multi-Group Cohort Sequential Growth Models Allowed for Cohort Differences (Free Model), Constrained the Latent Intercepts and Latent Slopes to Equality Across Age Cohorts While Allowing Variation Across Gender (Constrained Model-Gender Free), or Imposed Equality Across Gender (Fully Constrained Model).

Model Free Model Constrained Model (Gender Free) Fully Constrained Model

Benevolent sexism

AIC 78,347.403 78,446.805 78,600.825

aBIC 78,557.196 78,479.081 78,625.032

Hostile sexism

AIC 80,461.231 80,491.153 81,281.206

aBIC 80,671.019 80,523.428 81,305.413

AIC = Akaike information criterion; aBIC = Sample-size adjusted Bayesian information criterion.

Second, the estimation of cohort effects is displayed in Figure 3. For men’s endorsement of benevolent sexism, significant differences (where the error bars do not overlap zero) emerged in ages 49, 59, and 64 years. These findings indicated only few cases of inconsistency between the age group trajectories and the linear trajectory across age. In contrast, for men’s hostile sexism, cohort effects were observed for ages 24, 34 to 49, 59, and 64, indicating relatively consistent cohort effects across age in which the within-cohort trends of hostile sexism decreased relative to the U-shape of the age trajectory model.

Figure 3.
Estimated cohort effects for benevolent sexism (BS) and hostile sexism (HS) obtained via comparing the model-implied estimated values from the age-based models of endorsement of sexist attitudes with values from the 5-year latent change trajectory models. Negative effects indicate that model-estimated values were lower at a given age for the younger cohort (i.e., born in a later year) relative to the subsequent cohort.

Figure 3 displays significant cohort effects for women’s endorsement of benevolent sexism at ages 24, 34, 39, and from 54 to 69. These effects indicated that women’s benevolent sexism decreased more in early ages (vs. a shallower decline in the age-trajectory model) and was relatively stable in late age (vs. an increase in benevolent sexism in the age-trajectory model). For women’s endorsement of hostile sexism, the discrepancies between the cohort-based model and age-based model were less pronounced and restricted to ages 24, 34, 39, and 54. Thus, the longitudinal changes in the cohorts of women’s hostile sexism largely matched the U-shaped trajectory across age, except for a relatively steeper decline in hostile sexism in younger age cohorts.

Discussion

We utilized a national panel sample in an egalitarian country to conduct four models of age-related differences in endorsement of benevolent sexism and hostile sexism. Endorsement of sexism was generally stable, but significant differences emerged across time and age. First, endorsement of sexism generally followed U-shaped curvilinear trajectories across age: Trajectories for men’s hostile sexism, women’s benevolent sexism, and women’s hostile sexism were relatively high in late adolescence, lower through middle adulthood, and high again in older ages. In contrast, men’s endorsement of benevolent sexism followed a positive linear trajectory across age. Second, the “ambivalence” of sexist attitudes was consistently present, but weakened slightly, across age: The association between hostile sexism and benevolent sexism was the strongest in late adolescence and decreased linearly as age increased. Third, we employed cutting-edge statistical analyses to model the changes in benevolent sexism and hostile sexism over 5 years. Endorsement of sexism declined over time in 26 of 44 age-based cohorts, declines that were pronounced in the youngest cohorts (ages 19–24 and 25–29 years), and did not differ over time in the 18 remaining cohorts. Finally, tentative estimation of possible cohort effects indicated the presence of weak differences that were relatively consistent across age, but were most prominent in the youngest cohort.

Two intriguing findings emerged that extend current understanding of ambivalent sexism theory. First, although sexism was high in the youngest cohorts, the youngest cohorts also reported pronounced decreases in endorsement of sexist attitudes over time. This is consistent with prior research across age indicating that attitudes toward political and racial issues also tend to be highly susceptible to change in early adulthood, particularly for subjectively relevant issues (Visser & Krosnick, 1998). Finding similar flexibility in sexism during early adulthood is consistent with the perspective that life experiences prompt individuated beliefs about gender (see Glick & Fiske, 1996). Indeed, endorsement of hostile sexism was less closely related to benevolent sexism as age increased; the ambivalence of attitudes exhibited a small but consistent decline across age, perhaps signaling the development of more idiosyncratic beliefs about gender. The sharp declines in men’s hostile sexism across young adulthood were consistent with this expectation; men’s hostile sexism is highly incompatible with attainment of relationship goals (Hammond & Overall, 2016) which are a priority in young adulthood (e.g., Baltes et al., 1980; Takahashi, 2005).

Our results also provide a contextual map for understanding benevolent sexism. First, the positive linear trajectory, and relative stability, for men’s benevolent sexism, was in contrast to the U-shaped trajectory (and longitudinal declines) found for men’s hostile sexism and women’s ambivalent sexism, as well as traditional values or prejudice more generally (e.g., Gouveia et al., 2015; Visser & Krosnick, 1998). In sum, men’s endorsement of benevolent sexism was unexpectedly persistent over time. These results indicate a consistent appeal of benevolent sexism to men at all ages, perhaps because of its romantic appearance (e.g., Barreto & Ellemers, 2005) or the ways that benevolent sexism subtly maintains men’s power (e.g., Hammond & Overall, 2015; Hammond, Overall, & Cross, 2016; Shnabel, Bar-Anan, Kende, Bareket, & Lazar, 2016).

We also provide context for effects linked with women’s endorsement of benevolent sexism. Women’s benevolent sexism is relatively higher in young adulthood than mid-adulthood. Accordingly, the orientation toward relationship-oriented goals (rather than career-oriented goals) that accompanies women’s benevolent sexism (e.g., Chen et al., 2009; Montañés et al., 2013) is relatively higher at a life stage when women are making decisions about higher education, careers and establishing serious relationships.

Caveats and Future Directions

Results from our models of the trajectories of sexism across age were consistent with age group differences found in Spanish samples (see Fernández et al., 2004; Gariagordobil & Aliri, 2013), and our models approximating cohort effects did not reveal large differences. Nonetheless, the current models cannot speak to the mechanisms underlying the links between sexism and age, completely distinguish age-related differences from cohort effects, or test potential moderators without exceeding computational limits. Our current findings do, however, provide a map for investigating the moderators of life span changes in sexism. For example, endorsement of sexism tended to be relatively low in middle-aged cohorts, perhaps reflecting transitioning gender role expectations as men take on parenting or alloparenting roles (e.g., Bedford & Turner, 2006). The results also point toward potential historical moderators. Analyses indicated the presence of small cohort effects for ages 59–64 and 65–69 (but not consistently for adjacent cohorts). These cohorts would have developed through adolescence to adulthood in a normative context of egalitarianism that occurred post-World War II (see Mason, Czajka, & Arber, 1976), a theorized critical age for people’s receptivity to sexist attitudes (see Glick & Hilt, 2000). A good direction for future research would be to test these potential personal-, historical-, and societal-level moderators.

Finally, data were collected in New Zealand, a country in which endorsement of sexism and objective indices of gender inequality are relatively low compared to other countries (Glick et al., 2000; United Nations Development Programme, 2016). In less egalitarian countries, individualistic motivations to endorse sexism (that would correspond to age-graded developmental changes) may be overshadowed by societal pressures to endorse sexism. For example, people may endorse benevolent sexism as a means of attaining security in response to threat (Fischer, 2006; Glick et al., 2000) or because of societal pressure to uphold traditional beliefs and customs (see Schwartz & Rubel-Lifschitz, 2009) rather than to justify personally held hostile beliefs toward women (Glick et al., 2000). Thus, in less egalitarian contexts, it is possible that the trajectories of individuals’ sexist attitudes are more closely related to shifts in societal and cultural norms.

Conclusion

We examined differences in sexism across age in a nationally representative sample. The trajectories of women’s benevolent sexism, women’s hostile sexism, and men’s hostile sexism were relatively high in early adulthood, lower in middle adulthood, and high in late adulthood. In addition, longitudinal models indicated that endorsement of hostile sexism and women’s benevolent sexism tended to decline in over half of the age cohorts and did not increase over time in any age cohort, suggesting that people are generally becoming more egalitarian over time. However, men’s benevolent sexism followed a positive linear trajectory across age and was also relatively stable over time, highlighting the resilience of men’s subjectively romantic sexist attitudes among the trend toward egalitarianism. These results provide the first map of differences in the trajectory of sexism over the life span and the development of sexism over time.

Supplemental Material

Supplemental Material, SPPS727588_suppl_mat - Benevolent Sexism and Hostile Sexism Across the Ages

Supplemental Material, SPPS727588_suppl_mat for Benevolent Sexism and Hostile Sexism Across the Ages by Matthew D. Hammond, Petar Milojev, Yanshu Huang, and Chris G. Sibley in Social Psychological and Personality Science

Birth Cohort	N Women	N Men	Ages Sampled (T1–T5)
1940 and older	198	197	69–74
1941–1945	363	330	64–69
1946–1950	637	509	59–64
1951–1955	744	492	54–59
1956–1960	803	497	49–54
1961–1965	794	488	44–49
1966–1970	721	366	39–44
1971–1975	644	284	34–39
1976–1980	491	206	29–34
1981–1985	356	104	24–29
1986–1990	381	165	19–24
1991 and younger	186	63	18–23

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. HS_W1
2. HS_W2	.733**
3. HS_W3	.700**	.751**
4. HS_W4	.682**	.734**	.765**
5. HS_W5	.673**	.721**	.756**	.774**
6. HS_W6	.661**	.699**	.733**	.745**	.768**
7. BS_W1	.430**	.366**	.346**	.341**	.355**	.349**
8. BS_W2	.366**	.415**	.353**	.361**	.364**	.348**	.746**
9. BS_W3	.378**	.371**	.396**	.368**	.383**	.370**	.728**	.751**
10. BS_W4	.343**	.352**	.354**	.417**	.383**	.371**	.709**	.752**	.755**
11. BS_W5	.351**	.349**	.373**	.382**	.431**	.394**	.703**	.733**	.752**	.764**
12. BS_W6	.342**	.343**	.371**	.378**	.392**	.429**	.681**	.716**	.728**	.740**	.770**
13. Gender	.252**	.226**	.262**	.261**	.236**	.222**	.118**	.108**	.134**	.129**	.141**	.147**
14. Age	−.012	.012	.019	.022	.050**	.052**	.116**	.104**	.118**	.136**	.140**	.148**	.115**
Mean	3.317	3.212	3.021	3.014	3.023	3.057	4.073	4.030	3.863	3.826	3.791	3.811	–	49.450
SD	1.252	1.227	1.222	1.213	1.204	1.209	1.694	1.153	1.168	1.168	1.190	1.192	–	15.979
Reliability	.807	.807	.821	.819	.819	.825	.722	.728	.742	.737	.746	.757	–	–

Model	χ²	df	RMSEA	CFI	TLI	SRMR
Cross-sectional model
1. Between age groups configural	1,843.104	476	.079	.897	.863	.059
2. Between age groups metric	1,978.531	580	.072	.894	.885	.064
3. Between age groups scalar	2,268.621	684	.071	.880	.890	.068
4. Between gender configural	1,217.897	68	.072	.907	.877	.049
5. Between gender metric	1,253.184	76	.069	.905	.887	.050
6. Between gender scalar	1,594.752	84	.074	.878	.869	.058
Benevolent sexism over time
1. Configural over time	1,756.770	29	.076	.891	.831	.072
2. Metric over time	2,349.492	35	.080	.854	.813	.087
3. Scalar over time	2,467.839	39	.078	.847	.824	.089
4. Between age groups configural	2,578.507	341	.086	.854	.788	.090
5. Between age groups metric	2,732.488	425	.079	.849	.824	.096
6. Between age groups scalar	3,075.099	520	.075	.833	.841	.104
7. Between gender configural	2,136.083	62	.081	.867	.807	.082
8. Between gender metric	2,218.954	74	.075	.863	.833	.084
9. Between gender scalar	3,095.968	88	.082	.807	.803	.103
Hostile sexism over time
1. Configural over time	576.978	29	.043	.966	.948	.027
2. Metric over time	1,171.204	35	.056	.930	.911	.052
3. Scalar over time	1,251.499	39	.055	.926	.914	.053
4. Between age groups configural	1,408.857	341	.060	.934	.904	.056
5. Between age groups metric	1,601.291	425	.056	.927	.915	.064
6. Between age groups scalar	1,866.828	520	.054	.916	.920	.071
7. Between gender configural	1,054.736	62	.056	.936	.907	.048
8. Between gender metric	1,252.362	74	.056	.924	.907	.058
9. Between gender scalar	2,265.928	88	.070	.859	.856	.113

				95% Confidence Interval
Benevolent sexism
Men
Intercept	3.924	.024	164.962*	3.878	3.971	1.020*
Age—linear	0.073	.017	4.371*	0.040	0.100	0.394*
Age—quadratic	0.016	.009	1.812	−0.001	0.030	0.038
Age—cubic	0.002	.003	0.890	−0.003	0.007	0.002
Women
Intercept	3.672	.018	199.213*	3.636	3.708	1.020*
Age—linear	0.010	.014	0.742	−0.017	0.038	0.394*
Age—quadratic	0.034	.007	5.111*	0.021	0.047	0.038
Age—cubic	−0.003	.002	−1.076	−0.007	0.002	0.002*
Hostile sexism
Men
Intercept	3.437	.027	128.691*	3.385	3.490	1.099*
Age—linear	−0.067	.019	−3.603*	−0.103	−0.031	0.446*
Age—quadratic	0.033	.010	3.413*	0.014	0.052	0.051
Age—cubic	−0.002	.003	−0.531	−0.008	0.004	0.003
Women
Intercept	2.766	.018	151.758*	2.731	2.802	1.099*
Age—linear	−0.034	.014	−2.407*	−0.062	−0.006	0.446*
Age—quadratic	0.040	.006	6.226*	0.028	0.053	0.051
Age—cubic	0.001	.002	0.212	−0.004	0.005	0.003

					95% Confidence Interval
Benevolent sexism	.528	.024	22.175	<.001	.482	.575
Gender	−.032	.015	−2.123	.034	−.062	−.002
Age	−.038	.014	−2.659	.008	−.065	−.010
Age × Gender	−.002	.003	−0.521	.602	−.007	.004

Model	Free Model	Constrained Model (Gender Free)	Fully Constrained Model
Benevolent sexism
AIC	78,347.403	78,446.805	78,600.825
aBIC	78,557.196	78,479.081	78,625.032
Hostile sexism
AIC	80,461.231	80,491.153	81,281.206
aBIC	80,671.019	80,523.428	81,305.413

Footnotes

Authors’ Note

MPlus syntax for the models reported here are available on the New Zealand Attitudes and Values Survey website ().

Declaration of Conflicting Interests

The author(s) declared no potential conflict of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a Templeton World Charity Foundation Grant (ID: 0077).

Supplemental Material

The supplemental material is available in the online version of the article.

Note

References

Arnett

J. J.

(2000). Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist, 55, 469–480. doi:10.1037/0003-066X.55.5.469

Baltes

E. B.

Reese

H. W.

Lipsitt

L. E.

(1980). Life-span developmental psychology. Annual Review of Psychology, 31, 65–110. doi:10.1146/annurev.ps.31.020180.000433

Barreto

Ellemers

(2005). The burden of benevolent sexism: How it contributes to the maintenance of gender inequalities. European Journal of Social Psychology, 35, 633–642. doi:10.1002/ejsp.270

Becker

J. C.

Wright

S. C.

(2011). Yet another dark side of chivalry: Benevolent sexism undermines and hostile sexism motivates collective action for social change. Journal of Personality and Social Psychology, 101, 62–77. doi:10.1037/a0022615

Bedford

V. H.

Turner

B. F.

(2006). Men in relationships: A new look from a life course perspective. New York, NY: Springer.

Brandt

M. J.

(2011). Sexism and gender inequality across 57 societies. Psychological Science, 22, 1413–1418. doi:10.1177/0956797611420445

Casey

B. J.

Jones

R. M.

Somerville

L. H.

(2011). Braking and accelerating of the adolescent brain. Journal of Research on Adolescence, 21, 21–33. doi:10.1111/j.1532-7795.2010.00712.x

Chen

Fiske

S. T.

Lee

T. L.

(2009). Ambivalent sexism and power-related gender-role ideology in marriage. Sex Roles, 60, 765–778. doi:10.1007/s11199-009-9585-9

Cheung

G. W.

Rensvold

R. B

. (2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modelling: A multidisciplinary Journal, 9, 233–255. doi:10.1207/S15328007SEM0902_5

10.

Chopik

W. J.

Edelstein

R. S.

Fraley

R. C.

(2013). From the cradle to the grave: Age differences in attachment from early adulthood to old age. Journal of Personality, 81, 171–183. doi:10.1111/j.1467-6494.2012.00793.x

11.

Connor

R. A.

Glick

Fiske

S. T.

(2016). Ambivalent sexism in the twenty-first century. In Sibley

C. G.

Barlow

F. K.

(Eds.), The Cambridge handbook of the psychology of prejudice (pp. 295–320). Cambridge, MA: Cambridge University Press.

12.

Elder

G. H.

Jr (1995). The life course paradigm. In Moen

Elder

G. H.

Jr Lüscher

(Eds.), Examining lives in context: Perspectives on the ecology of human development (pp. 101–139). Washington, DC: APA.

13.

Fernández

M. L.

Castro

Y. R.

Lorenzo

M. G.

(2004). Evolution of hostile sexism and benevolent sexism in a Spanish sample. Social Indicators Research, 66, 197–211. doi:10.1023/B: SOCI.0000003553.30419.f1

14.

Fischer

A. R.

(2006). Women’s benevolent sexism as reaction to hostility. Psychology of Women Quarterly, 30, 410–416. doi:10.1111/j.1471-6402.2006.00316.x

15.

Garaigordobil

Aliri

(2013). Ambivalent Sexism Inventory: Standardization and normative data in a sample of the Basque country. Behavioral Psychology-Psicologia Conductual, 21, 173–186.

16.

Glick

Fiske

S. T.

(1996). The ambivalent sexism inventory: Differentiating hostile and benevolent sexism. Journal of Personality and Social Psychology, 70, 491–512. doi:10.1037//0022-3514.70.3.491

17.

Glick

Fiske

S. T.

(2001). An ambivalent alliance: Hostile and benevolent sexism as complementary justifications for gender inequality. American Psychologist, 56, 109–118. doi:10.1037/0003-066X.56.2.109

18.

Glick

Fiske

S. T.

Mladinic

Saiz

Abrams

Masser

… López

W. L

. (2000). Beyond prejudice as simple antipathy: Hostile and benevolent sexism across cultures. Journal of Personality and Social Psychology, 79, 763–775. doi:10.1037//0022-3514.79.5.763

19.

Glick

Hilt

(2000). From combative children to ambivalent adults: The development of gender prejudice. In Eckes

Trautner

(Eds.), Developmental social psychology of gender (pp. 243–272). Mahwah, NJ: Erlbaum.

20.

Gouveia

V. V.

Vione

K. C.

Milfont

T. L.

Fischer

(2015). Patterns of value change during the life span: Some evidence from a functional approach of values. Personality and Social Psychology Bulletin, 41, 1276–1290. doi:10.1177/0146167215594189

21.

Hammond

M. D.

Overall

N. C.

(2015). Benevolent sexism prompts dependence-oriented support in men and autonomy-oriented support in women. Personality and Social Psychology Bulletin, 41, 1180–1194. doi:10.1177/0146167215593492

22.

Hammond

M. D.

Overall

N. C.

(2016). Sexism in intimate contexts: How romantic relationships help explain the origins, functions, and consequences of sexist attitudes. In Sibley

C. G.

Barlow

F. K.

(Eds.), The Cambridge handbook of the psychology of prejudice (pp. 321–343). Cambridge, MA: Cambridge University Press.

23.

Hammond

M. D.

Overall

N. C.

Cross

E. J.

(2016). Internalizing sexism in close relationships: The perceived sexism of partners maintains women’s endorsement of benevolent sexism. Journal of Personality and Social Psychology, 110, 214–238. doi:10.1037/pspi0000043

24.

Hammond

M. D.

Sibley

C. G.

(2011). Why are benevolent sexists happier? Sex Roles, 65, 332–343. doi:10.1007/s11199-011-0017-2

25.

Heckhausen

(2006). Developmental regulation in adulthood: Age-normative and sociostructural constraints as adaptive challenges. Cambridge, MA: Cambridge University Press.

26.

Lucas

R. E.

& Donnellan

M. B.

(2011). Personality development across the life span: Longitudinal analyses with a national sample from Germany. Journal of Personality and Social Psychology, 101, 847–861. doi:10.1037/a0024298

27.

Mason

Czajka

Arber

(1976). Change in U.S. women’s sex-role attitudes, 1964-1974. American Sociological Review, 41, 573–596.

28.

Milojev

Sibley

C. G

. (2016). Normative personality trait development in adulthood: A 6-year cohort-sequential growth model. Journal of Personality and Social Psychology. Published online before print. doi:10.1037/pspp0000121

29.

Montañés

de Lemus

Moya

Bohner

Megías

J. L.

(2013). How attractive are sexist intimates to adolescents? The influence of sexist beliefs and relationship experience. Psychology of Women Quarterly, 37, 494–506. doi:10.1177/0361684313475998

30.

Muthén

L. K.

Muthén

B. O

. (1998–2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Muthén & Muthén.

31.

O’Neil

J. M.

Egan

(1992). Men’s gender role transitions over the life span: Transformations and fears of femininity. Journal of Mental Health Counseling, 14, 305–324.

32.

Preacher

K. J.

Wichman

A. L.

MacCallum

R. C.

Briggs

N. E.

(2008). Latent growth curve modeling. Thousand Oaks, CA: Sage.

33.

Satherley

Milojev

Greaves

L. M.

Huang

Osborne

Bulbulia

Sibley

C. G.

(2015). Demographic and psychological predictors of panel attrition: Evidence from the New Zealand Attitudes and Values Study. PLoS ONE, 10, e0121950.

34.

Schwartz

S. H.

Rubel-Lifschitz

(2009). Cross-national variation in the size of sex differences in values: Effects of gender equality. Journal of Personality and Social Psychology, 97, 171–185. doi:10.1037/a0015546

35.

Shnabel

Bar-Anan

Kende

Bareket

Lazar

(2016). Help to perpetuate traditional gender roles: Benevolent sexism increases engagement in dependency-oriented cross-gender helping. Journal of Personality and Social Psychology, 110, 55–75. doi:10.1037/pspi0000037

36.

Takahashi

(2005). Toward a life span theory of close relationships: The affective relationships model. Human Development, 48, 48–66. doi:10.1159/000083215

37.

United Nations Development Programme. (2016). Human Development Report 2016. Retrieved May 1, 2017, from hdr.undp.org/en/2015-report/download

38.

Visser

P. S.

Krosnick

J. A.

(1998). Development of attitude strength over the life cycle: Surge and decline. Journal of Personality and Social Psychology, 75, 1389–1410. doi:10.1037/0022-3514.75.6.1389

39.

Wrzus

Hänel

Wagner

Neyer

F. J.

(2013). Social network changes and life events across the life span: A meta-analysis. Psychological Bulletin, 139, 53–80. doi:10.1037/a0028601

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB

				95% Confidence Interval
Model	Estimate	SE	z	Low	High	Variances
Benevolent sexism
Men
Intercept	3.924	.024	164.962*	3.878	3.971	1.020*
Age—linear	0.073	.017	4.371*	0.040	0.100	0.394*
Age—quadratic	0.016	.009	1.812	−0.001	0.030	0.038
Age—cubic	0.002	.003	0.890	−0.003	0.007	0.002
Women
Intercept	3.672	.018	199.213*	3.636	3.708	1.020*
Age—linear	0.010	.014	0.742	−0.017	0.038	0.394*
Age—quadratic	0.034	.007	5.111*	0.021	0.047	0.038
Age—cubic	−0.003	.002	−1.076	−0.007	0.002	0.002*
Hostile sexism
Men
Intercept	3.437	.027	128.691*	3.385	3.490	1.099*
Age—linear	−0.067	.019	−3.603*	−0.103	−0.031	0.446*
Age—quadratic	0.033	.010	3.413*	0.014	0.052	0.051
Age—cubic	−0.002	.003	−0.531	−0.008	0.004	0.003
Women
Intercept	2.766	.018	151.758*	2.731	2.802	1.099*
Age—linear	−0.034	.014	−2.407*	−0.062	−0.006	0.446*
Age—quadratic	0.040	.006	6.226*	0.028	0.053	0.051
Age—cubic	0.001	.002	0.212	−0.004	0.005	0.003

					95% Confidence Interval
Predictor	β	SE	t	p	Low	High
Benevolent sexism	.528	.024	22.175	<.001	.482	.575
Gender	−.032	.015	−2.123	.034	−.062	−.002
Age	−.038	.014	−2.659	.008	−.065	−.010
Age × Gender	−.002	.003	−0.521	.602	−.007	.004