The Impact of Social Roles on Trait Judgments

Abstract

Consistent with social role theory’s assumption that the role behavior of men and women shapes gender stereotypes, earlier experiments have found that men’s and women’s occupancy of the same role eliminated gender-stereotypical judgments of greater agency and lower communion in men than women. The shifting standards model raises the question of whether a shift to within-sex standards in judgments of men and women in roles could have masked underlying gender stereotypes. To examine this possibility, two experiments obtained judgments of men and women using measures that do or do not restrain shifts to within-sex standards. This measure variation did not affect the social role pattern of smaller perceived sex differences in the presence of role information. These findings thus support the social role theory claim that designations of identical roles for subgroups of men and women eliminate or reduce perceived sex differences.

Keywords

social perception social roles shifting standards gender stereotypes judgment

Studies have shown that men and women are judged to be similar when they occupy the same role, despite the fact that the sexes are in general judged to have different traits (Eagly & Steffen, 1984). Yet the question of whether gender stereotypes are operating in the presence of role information is controversial (Bridges, Etaugh, & Barnes-Farrell, 2002). According to social role theory, perceivers’ inferences of traits from roles allow them to view the sexes similarly in the same role and thus eliminate gender stereotypes. However, according to the shifting standards model, this apparent elimination of stereotypes might actually be the result of gender stereotypes causing perceivers to judge male or female role occupants relative to their own sex (Biernat, 2003, 2005). This research tests the validity of these competing explanations of the effects of role information on gender stereotypes, and the findings have implications for understanding whether observations of men and women in the same roles can change gender stereotypes.

Social role theory proposes that stereotypical beliefs about male agency and female communion stem from observations of the two sexes in societal roles that are thought to require different types of personality traits and competencies (Eagly, 1987; Eagly & Wood, in press; Eagly, Wood, & Diekman, 2000). Men’s greater occupancy of the breadwinner role and authority roles in society causes social perceivers to infer that men are especially agentic. Similarly, women’s greater occupancy of the domestic role and communally demanding employment roles causes social perceivers to infer that women are especially communal (Cejka & Eagly, 1999).

Driving these beliefs is an important principle of human inference, known as correspondent inference or correspondence bias, by which people’s behavior is thought to reflect their psychological dispositions (Gawronski, 2003; Gilbert, 1998; Gilbert & Malone, 1995). Therefore, people believe that men and women possess different attributes corresponding to the activities they typically perform in their society. This logic further implies that gender-stereotypical beliefs should erode for men and women who occupy the same role. In support of this principle, experiments by Eagly and Steffen (e.g., 1984, 1986, 1988) and other researchers (e.g., Harrison, 2005; Kite, 1996; Sczesny & Kühnen, 2004) have found that men and women were rated similarly in the same role (e.g., homemaker role, employee role) and gender stereotypically only in the absence of role information.

The shifting standards model raises questions about the interpretation of these experiments that manipulated the social roles of women and men (e.g., Eagly & Steffen, 1984). From the perspective of this model, social perceivers’ shifts of judgment standards may account for the apparent erosion of perceived sex differences in the presence of role information, as argued by Bridges and colleagues (2002). Specifically, people may judge individuals in relation to their own sex when they believe that women and men differ in general—that is, they judge a woman relative to other women and a man relative to other men (Biernat, 2003, 2005). For example, because men are thought to have more managerial competence than women (e.g., Schein, 2001), people would judge an individual male job applicant for a managerial position relative to a higher, male standard of job-related competence and an individual female job applicant relative to a lower, female standard of competence. The result would be that, even if a particular woman had lesser qualifications than her male counterpart, she could nonetheless be judged as equally competent or even more competent than him. In research, these standard shifts are enabled by the use of subjective rating scales (e.g., 1 = not at all competent to 7 = very competent), which allow participants to implicitly recalibrate their judgments to within-group standards.

This shifting standards logic does not apply to judgments of all women and all men as social categories because each sex can be compared only to the other sex, even on subjective rating scales. Judgments of the female sex imply a comparison to men, and judgments of the male sex imply a comparison to women. Gender stereotypes thus prevail in ratings of women and men as social groups. In contrast, according to the shifting standards model, judgments of individual role occupants identified by sex (e.g., Susan, the firefighter) might instead elicit a comparison to the role occupant’s own sex group (e.g., Susan, the firefighter, is more agentic than most women). The apparent disappearance of gender stereotypes in the presence of role information might thus be the result of perceivers’ shift to within-sex standards under these circumstances. Specifically, although gender-stereotypical judgments seemingly disappeared in experiments that portrayed men and women in the same role (e.g., Eagly & Steffen, 1984, 1986), stereotypes might still have operated by facilitating shifts to different, sex-specific standards for rating women and men in roles.

To restrain these standard shifts and thereby reveal the stereotypes that underlie them, Biernat (1995) recommended the use of objective, or common rule measures. The response units of these scales have a constant meaning and thus require the application of a common standard, regardless of the sex of the person being judged. Examples of common rule measures include (a) behavioral frequency scales that require judges to estimate how often men and women engage in particular trait-relevant behaviors (e.g., Kobrynowicz & Biernat, 1997), (b) standardized personality scales that require judges to estimate how men and women would score on personality tests (e.g., Biernat & Manis, 1994; Biernat, Manis, & Nelson, 1991), and (c) comparative scales that require judges to use a common standard by directly comparing men and women (e.g., Bosak, Sczesny, & Eagly, 2007; Diekman & Eagly, 2000, Experiment 4). Many studies that compared judgments on common rule and subjective scales have supported the assumptions of the shifting standards model by finding greater evidence of stereotyping individuals on common rule scales than subjective measures (see reviews by Biernat, 2003, 2005). Therefore, to compare the shifting standards model and the social role theory, our research used both types of measures.

Research Question and Overview of Present Research

The present research compared two explanations for the impact of social roles on trait judgments in previous social role experiments (e.g., Eagly & Steffen, 1984), which used only subjective rating scales. The social role interpretation implies that role information overrode the influence of sex as shown by the similar trait ratings of male and female role occupants. In contrast, the shifting standards interpretation implies that this convergence in trait judgments of male and female role occupants reflected participants’ use of within-sex standards that mask gender stereotypes. For example, a male and a female homemaker might have been rated similarly communal because the male homemaker was judged relative to men in general (i.e., a within-sex comparison). In the absence of role information, an average man would have been judged relative to the communion of the average woman and therefore as less communal than her (i.e., a between-sex comparison).

The present research was designed to overcome the limitations of the prior experiments that have compared the validity of the social role and shifting standards interpretations of the effects of roles on judgments. Only three prior experiments have addressed this issue by varying whether participants made judgments on subjective or common rule measures. Two of these produced shifting standards effects (Bridges et al., 2002; Park, Smith, & Correll, 2008), and one experiment produced mixed findings (Bosak, Sczesny, & Eagly, 2008). In addition, an experiment that used only a common rule measure produced social role findings (Bosak et al., 2007).

These disparate findings likely reflect the shortcomings of these experiments. In particular, the Bridges et al. (2002) and the Park et al. (2008) experiments lacked comparison conditions presenting men and women in general, which are necessary to establish the presence of gender stereotypes in the context of the experiment. Also, these researchers’ common rule measures, which consisted of ratings of the frequency of various role-relevant behaviors, were not constructed to ensure that they conveyed the same meaning as their subjective measures, which consisted of trait rating scales. Bosak et al.’s (2008) manipulation of social roles (homemaker, full-time employee) was compromised by the full-time employee role having become typical of women as well as men. Therefore, this employee role did not convey specific information about a culturally masculine subtype of women as it had in earlier experiments (e.g., Eagly & Steffen, 1984). Finally, the Bosak et al. (2007) experiment was limited by its lack of a subjective measure for comparison to its common rule measure. These four articles thus fail to provide a convincing answer to the question of how social role effects on judgments of women and men should be interpreted.

The two experiments presented in this article provide much stronger tests of the validity of social role and standard shift explanations of trait judgments of male and female role occupants by using designs that overcome the limitations of these earlier studies. Specifically, both experiments compared judgments on subjective rating scales to judgments on two other types of common rule measures, namely, behavioral frequency scales in Experiment 1 and standardized personality scales in Experiment 2. These common rule measures were carefully matched to the content of the subjective measures to ensure that the two types of measures conveyed similar meaning. Also, these two experiments enlarged the selection of social roles compared with earlier experiments. In addition to a no-role condition referring to men and women in general, Experiment 1 presented male- and female-dominated occupational roles, and Experiment 2 included these occupational roles as well as homemaker and employee roles.

Experiment 1

Participants judged women and men presented as general social categories and as portrayed in specific female- or male-dominated occupational roles on agentic and communal traits using both a subjective measure and a common rule measure. The inclusion of both types of measures allowed for a comparative test of the validity of social role and standard shift explanations of trait judgments of men and women. The subjective measure consisted of trait rating scales similar to those used by Eagly and Steffen (1984). The common rule measure consisted of behavioral frequency scales similar to those used by Kobrynowicz and Biernat (1997).

From the social role perspective, the social role effect of larger sex differences in the absence than the presence of specific role information will be obtained regardless of type of measure. From the shifting standards perspective, this social role effect will be more pronounced on a subjective measure than a common rule measure. Inherent in this shifting standards prediction is the assumption that subjective measures facilitate shifts to within-sex comparisons for male and female role occupants.

Method

Participants and procedure

Among the 111 male, 120 female, and 5 unknown sex participants, 55.5% were European American, 15.3% were African American, 7.2% were Hispanic American, 8.9% were Asian American, and 13.1% were of other or unknown ethnicity, and they ranged in age from 18 to 79 with a mean of 33.0 years. The participants were recruited in public places in Chicago (e.g., public library, museums, coffee shops) by asking every fifth person encountered to participate in a study on “impressions of other people.” Of those approached, 84.3% participated. The two surveyors handed each participant a questionnaire and collected it approximately 10 minutes later. Finally, the surveyor thanked the participant and provided a written debriefing statement.

Independent variables

The no-role target persons were described as an average man or an average woman. To add role information, the target persons were portrayed as occupying the role of employee in either a female-dominated occupation (nurse or social worker) or a male-dominated occupation (firefighter or law enforcement officer). In both experiments, the instructions indicated that all employees held full-time positions, and the designation of average preceded the stimulus person designation (e.g., average female social worker).

The selection of the female-dominated and male-dominated occupations was based on their sex ratios and beliefs about gender-stereotypical attributes important to success in these occupations as assessed by Cejka and Eagly (1999). In that study of 80 occupations, high agency and low communion were ascribed to firefighters and law enforcement officers, and low agency and high communion to nurses and social workers. Also, women were rare as firefighters (4.8%) and police and sheriff’s patrol officers (14.7%) but common as registered nurses (91.7%) and social workers (79.4%; U.S. Department of Labor, 2010). Because the two female-dominated occupations produced similar findings, as did the two male-dominated occupations, they were combined in the results. Also, because in both experiments sex of participant produced no effects, the data were combined across the two sexes.

Dependent variables

Participants judged a target person on subjective or common rule scales that represented four agentic traits (dominant, aggressive, competitive, adventurous) and four communal traits (affectionate, supportive, sympathetic, gentle).

Subjective measures

Participants responded on 7-point scales ranging from not at all to extremely (e.g., “How dominant do you think the [target person] is?”). The ratings were averaged across the agentic items (α = .72) and the communal items (α = .81).

Common rule measures

Participants estimated the frequency with which the target person would engage in six prototypical behaviors indicative of each gender-stereotypical trait on 11-point scales ranging from 0 to 10 times out of 10 opportunities (e.g., “How often does the [target person] get the last word in an argument?”). The ratings were averaged across the agentic behaviors (α = .85) and the communal behaviors (α = .90).

To ensure that the meaning of the common rule scales corresponded to that of the subjective scales, these behaviors were selected by a two-stage pretest that was similar to a method used by Buss and Craik (1980). In the first stage, 75 male and 83 female participants listed behaviors that came to mind when thinking of an acquaintance with the respective personality trait (e.g., dominant). These behaviors were reduced to 20 for each trait (i.e., by eliminating redundancies and overly general statements) and phrased to apply to men or women in a wide range of situations. In the second stage, using a 7-point rating scale ranging from 1 = not at all to 7 = extremely, 73 male and 87 female participants rated the 20 behaviors (presented in two different orders) on the extent to which each behavior was prototypical of the corresponding trait. The 6 most prototypical behaviors for each trait constituted the common rule measure (for item examples, see the appendix).

Because the subjective and common rule measures used different scales (7-point vs. 11-point), the data were Z standardized within each measure and transformed to a mean of 50 and standard deviation of 10.

Given this manipulation of measure type, the resulting mixed four-factor design was Target Sex (male vs. female) × Target Role (none vs. employee in female-dominated occupation vs. employee in male-dominated occupation) × Measure Type (subjective vs. common rule) × Stereotype Dimension (communion vs. agency), with the last factor being within-subjects.

Results

Throughout this article, the results of the overall mixed four-factor ANOVA are presented, followed by separate, more detailed reports of the findings on agency and communion. Also, throughout this article, p values of .05 or less are considered significant, and all contrasts are two-tailed.

The means and standard deviations corresponding to the four-factor mixed ANOVA appear in Table 1. From the social role perspective, judgments of target men and women should display women’s greater communion and men’s greater agency, moderated by the convergence of women and men portrayed in the same social role. Consistent with this assumption, the following interaction effects were significant: Stereotype Dimension × Target Sex, F(1, 224) = 20.75, p < .001, η² = .09; Stereotype Dimension × Target Role, F(2, 224) = 20.99, p < .001, η² = .16; and Stereotype Dimension × Target Sex × Target Role, F(2, 224) = 7.02, p = .001, η² = .06. These interactions are presented as decomposed within each stereotype dimension. From the shifting standards perspective, this three-way interaction would be moderated by measure type, but this four-way interaction was not significant, F(2, 224) = 0.61, p = .543, η² = .01.

Table 1.

Experiment 1: Means and Standard Deviations on Communion and Agency by Measure, Role, and Target Sex

		Role
		None		Male dominated		Female dominated		Total
Measure	Target sex	M	SD	M	SD	M	SD	M	SD
Communion
Subjective	Male	48.37	8.91	45.67	12.55	50.27	8.57	48.09	10.06
	Female	54.07	8.86	47.85	7.71	50.90	9.24	51.39	8.90
	Total	51.22	9.25	46.69	10.46	50.58	8.77	49.70	9.61
Common rule	Male	42.08	7.27	45.88	7.72	52.02	9.70	46.52	9.14
	Female	55.81	5.22	49.77	7.91	54.16	11.50	53.43	8.51
	Total	49.24	9.31	47.83	7.96	53.00	10.47	49.95	9.45
Total	Male	45.36	8.68	45.78	10.21	51.21	9.12	47.29	9.60
	Female	54.94	7.25	48.89	7.77	52.58	10.43	52.44	8.72
	Total	50.25	9.28	47.29	9.17	51.86	9.71	49.83	9.51
Agency
Subjective	Male	52.99	7.05	56.03	11.08	45.57	10.16	51.78	10.09
	Female	46.66	7.31	55.51	10.59	43.71	9.26	48.35	9.94
	Total	49.83	7.79	55.79	10.69	44.67	9.62	50.11	10.12
Common rule	Male	56.60	8.56	52.90	9.59	44.07	9.68	51.34	10.56
	Female	50.94	8.18	49.43	8.36	45.05	10.35	48.79	9.08
	Total	53.64	8.75	51.17	9.05	44.52	9.87	50.08	9.90
Total	Male	54.72	7.94	54.43	10.32	44.76	9.79	51.56	10.29
	Female	48.80	7.98	52.21	9.80	44.40	9.71	48.58	9.47
	Total	51.70	8.45	53.35	10.06	44.59	9.68	50.09	9.98

Both measures were standardized to produce an average of 50 and a standard deviation of 10, with higher numbers indicating greater communion or agency. Cell ns ranged from 16 to 24 participants.

Communion

In the 3 (target role) × 2 (target sex) × 2 (measure type) design, the main effect for target sex was significant, F(1, 224) = 16.41, p < .001, η² = .07, whereby women were judged as more communal than men. In addition, the main effect for target role was significant, F(2, 224) = 4.78, p = .009, η² = .04. Participants thus judged targets in female-dominated occupations and without role information as equally communal, p = .279, and they judged both groups as more communal than targets in male-dominated occupations, p = .004 and p = .045, respectively. As predicted by social role theory, these main effects were qualified by a Target Role × Target Sex interaction, F(2, 224) = 5.21, p = .006, η² = .04. Without role information, participants thus judged women as more communal than men, p < .001, whereas with role information, they judged men and women in female-dominated occupations and in male-dominated occupations as equally communal, p = .520 and p = .136, respectively. The critical Target Role × Target Sex × Measure Type interaction predicted by the shifting standards model was nonsignificant, F(2, 224) = 0.03, p = .395, η² = .01.

Agency

In the 3 (Target Role) × 2 (Target Sex) × 2 (Measure Type) design, the main effect for target sex was significant, F(1, 224) = 5.50, p = .02, η² = .02, whereby men were judged as more agentic than women. In addition, the main effect for target role was significant, F(2, 224) = 19.21, p < .001, η² = .15. Participants judged targets in male-dominated occupations and without role information as equally agentic, p = .259, and they judged both groups as more agentic than targets in female-dominated occupations, ps < .001. Even though the Target Role × Target Sex interaction predicted by social role theory was not significant, F(2, 224) = 2.06, p = .130, the critical contrasts within each role were as predicted. Without role information, participants judged men as more agentic than women, p = .002, whereas with role information, they judged men and women in female-dominated occupations and in male-dominated occupations as equally agentic, p = .870 and p = .303, respectively. The critical Target Role × Target Sex × Measure Type interaction predicted by the shifting standards model was nonsignificant, F(2, 224) = 0.46, p = .633, η² = .00.

Effects of measure type

Although none of the effects of measure type predicted by the shifting standards model were obtained, two nonpredicted effects were significant. In the overall four-factor ANOVA, the only significant effect was the Stereotype Dimension × Target Role × Measure Type interaction, F(2, 224) = 5.08, p = .007, η² = .04. Decomposing this interaction within stereotype dimension yielded a significant simple two-way interaction only on agency: Target Role × Measure Type, F(2, 224) = 4.48, p = .012, η² = .04. Consistent with Table 1, measure type contrasts yielded the following results: (a) In the no-role condition, agency was lower with the subjective than the common rule measure, p = .037; (b) in the male-dominated role condition, agency was higher with the subjective than the common rule measure, p = .028; and (c) in the female-dominated role condition, agency did not differ between the subjective and the common rule measure, p = .945. These effects are irrelevant to both social role theory and the shifting standards model.

Discussion

Experiment 1 provided a comparative test of social role theory and the shifting standards model by examining judgments of men and women with and without role information on subjective scales that allow shifts to within-sex standards and on common rule scales that restrain such standard shifts. Consistent with social role theory, the presence of specific role information restrained gender-stereotypical trait judgments of men and women, regardless of measure type. Although the expected Target Role × Target Sex interaction was significant on communion only, the contrasts were as predicted on communion and agency. In the absence of role information, men were judged as more agentic and less communal than women, whereas in the presence of role information, men and women occupying a female-dominated role or a male-dominated role were judged similarly agentic and communal. These findings are in line with previous research that used subjective measures only (e.g., Eagly & Steffen, 1984). Role information did not cause participants to shift to a within-sex standard with the subjective scales as shown by the equivalence of the findings with the common rule and subjective measures.

These findings show that the specific occupational roles presented in the experiment provided relatively clear-cut information about the traits of the role occupants. Regardless of the role occupant’s sex, targets in female-dominated occupations were judged similar to stereotypical women (high in communion and low in agency), and targets in male-dominated occupations were judged similar to stereotypical men (low in communion and high in agency).

Experiment 2

In this experiment, we tested the generalizability of the findings of Experiment 1 by designing a different common rule measure, which consisted of standardized personality scales rather than behavioral frequency scales. In addition, we included not only the male- and female-dominated occupational roles of Experiment 1 but also the homemaker and the generic employee roles used in previous social role experiments (e.g., Eagly & Steffen, 1984). According to social role theory, the gender-stereotypical judgments of the social categories of men and women should be restrained by information about specific roles. The exception to this prediction is the generic employee role, which should produce findings similar to the no-role condition because full-time employment has become the usual situation for both sexes in the United States (U.S. Department of Labor, 2010; see Bosak et al., 2008). Despite these high levels of labor force participation, employment is considerably sex segregated, with women overrepresented in occupations demanding more communal behaviors and men overrepresented in occupations demanding more agentic behaviors (Cejka & Eagly, 1999; Tomaskovic-Devey et al., 2006). Therefore, participants should make gender-stereotypical inferences about the traits of women and men in the generic employee condition just as in the no-role condition. With respect to measure type, social role theory predicts that the manipulation of subjective versus common rule measures will not affect responding, and the shifting standards model predicts that the role effects predicted by social role theory will be stronger on the subjective than the common rule measure.

Method

Participants and procedure

Among the 177 male, 194 female, and 3 unknown sex participants, 61% were European American, 12% were African American, 8% were Hispanic American, 10% were Asian American, and 9% were of other or unknown ethnicity, and they ranged in age from 16 to 76, with a mean of 34.2 years. Six surveyors recruited these participants following the same procedure described in Experiment 1. Of those approached, 70% participated.

Independent variables

The no-role target persons were described as an average man or an average woman. To add role information, the target persons were portrayed as occupying the role of employee, homemaker, or employee in a female-dominated occupation (nurse or social worker) or a male-dominated occupation (firefighter or law enforcement officer). As in Experiment 1, the two female-dominated and the two male-dominated occupations were combined because their findings were similar.

Dependent variables

As in Experiment 1, participants judged a target person on four agentic traits (dominant, aggressive, competitive, adventurous) and four communal traits (affectionate, supportive, sympathetic, gentle) using either the same subjective rating scales as in Experiment 1 or a common rule scale. This experiment’s common rule measure asked participants to estimate how the target person would score on personality tests assessing each of these agentic and communal traits. Participants responded on an 11-point scale ranging from 0 to 100 in decadal steps (e.g., score of 70 out of 100 for dominance; see Biernat, 2003). As in Experiment 1, the data were Z standardized within each measure and transformed to a mean of 50 and standard deviation of 10. The subjective measures produced alphas of .86 for agency and .87 for communion, and the common rule measures produced alphas of .85 for agency and .93 for communion.

Including the manipulation of type of measure, the resulting mixed four-factor design was Target Sex (male vs. female) × Target Role (none vs. homemaker vs. employee vs. employee in male-dominated occupation vs. employee in female dominated occupation) × Measure Type (subjective vs. common rule) × Stereotype Dimension (communion vs. agency), with the last factor being within-subjects.

Results

The means and standard deviations corresponding to the four-factor mixed design appear in Table 2. As predicted by social role theory, the following interactions were significant: Stereotype Dimension × Target Sex, F(1, 349) = 63.87, p < .001, η² = .16; Stereotype Dimension × Target Role, F(4, 349) = 89.88, p < .001, η² = .51; and Stereotype Dimension × Target Sex × Target Role, F(4, 349) = 5.77, p < .001, η² = .06. These interactions are presented as decomposed within each stereotype dimension. From the shifting standards perspective, this three-way interaction would be moderated by measure type, but this four-way interaction was not significant, F(4, 349) = 0.68, p = .608, η² = .01.

Table 2.

Experiment 2: Means and Standard Deviations on Communion and Agency by Measure, Role, and Target Sex

		Role
		None		Homemaker		Employee		Male dominated		Female dominated		Total
Measure	Target sex	M	SD	M	SD	M	SD	M	SD	M	SD	M	SD
Communion
Subjective	Male	45.37	6.09	52.44	7.37	41.07	8.77	42.56	9.73	51.16	8.49	46.37	9.45
	Female	55.45	7.76	59.49	5.50	50.70	5.54	45.44	7.26	59.03	8.77	53.48	9.29
	Total	50.41	8.56	56.10	7.29	46.05	8.66	43.98	8.65	55.32	9.43	50.00	10.00
Common rule	Male	43.87	7.51	51.04	10.10	43.78	7.54	40.35	6.88	55.15	8.84	46.62	10.01
	Female	57.86	4.49	58.02	6.96	51.12	6.50	46.96	8.24	58.92	5.99	53.72	8.60
	Total	48.96	9.44	54.72	9.08	47.45	7.83	43.37	8.17	57.00	7.74	49.95	10.00
Total	Male	44.62	6.75	51.87	8.39	42.26	8.20	41.40	8.36	53.23	8.82	46.49	9.70
	Female	56.33	6.74	58.88	6.05	50.88	5.84	46.19	7.72	58.98	7.49	53.59	8.96
	Total	49.77	8.89	55.53	8.01	46.65	8.26	43.67	8.38	56.16	8.63	49.97	9.99
Agency
Subjective	Male	55.90	4.72	42.30	10.51	52.69	8.39	58.05	9.12	44.50	5.98	51.22	10.09
	Female	44.37	7.25	46.41	8.63	47.11	6.96	58.45	7.18	43.56	8.52	48.83	9.82
	Total	50.13	8.40	44.43	9.62	49.80	8.06	58.25	8.15	44.00	7.38	50.00	10.00
Common rule	Male	53.94	7.50	45.04	8.67	55.60	6.69	57.55	6.69	42.02	8.08	51.06	9.94
	Female	47.49	8.75	44.58	7.70	48.50	8.61	57.26	6.98	42.48	8.79	48.82	10.05
	Total	51.60	8.39	44.80	7.95	52.05	8.36	57.42	6.77	42.25	8.36	50.00	10.03
Total	Male	54.92	6.23	43.42	9.68	53.97	7.68	57.79	7.87	43.21	7.19	51.14	9.99
	Female	45.50	7.77	45.65	8.13	47.70	7.57	57.87	7.04	43.03	8.59	48.83	9.90
	Total	50.78	8.34	44.58	8.88	50.77	8.18	57.83	7.46	43.12	7.90	50.00	10.00

Both measures were standardized to produce an average of 50 and a standard deviation of 10, with higher numbers indicating greater communion or agency. Cell ns ranged from 20 to 25 participants.

Communion

In the 2 (target sex) × 5 (target role) × 2 (measure type) design, the main effect for target sex was significant, F(1, 349) = 74.13, p < .001, η² = .18, whereby women were judged as more communal than men. In addition, the main effect for target role was significant, F(4, 349) = 42.13, p < .001, η² = .33. Participants judged targets in the homemaker role and female-dominated occupations as more communal than targets without role information, ps ≤ .001, targets in the employee role, ps < .001, and targets in male-dominated occupations, ps < .001. Participants also judged targets without role information as marginally more communal than targets in the employee role, p = .065, and both groups as more communal than targets in male-dominated occupations, p < .001 and p = .037, respectively. As predicted by social role theory, these main effects were qualified by a (marginal) Target Sex × Target Role interaction, F(4, 349) = 2.12, p = .078, η² = .03. Although within-role comparisons revealed that women were judged as more communal than men, ps < .002, regardless of their role, these differences were larger in the no-role and employee conditions than in the other, specific-role conditions, as shown by an interaction contrast that compared the judged sex difference in the no-role and employee conditions with the difference in the other three conditions, F(1, 349) = 4.68, p = .032. The critical Target Role × Target Sex × Measure Type interaction predicted by the shifting standards model was nonsignificant, F(4, 349) = 1.14, p = .339, η² = .01, and no other effects of measure type approached significance, ps ≥ .339.

Agency

In the 2 (target sex) × 5 (target role) × 2 (measure type) design, the main effect for target sex was significant, F(1, 350) = 9.39, p = .002, η² = .03, whereby men were judged as more agentic than women. In addition, the main effect for target role was significant, F(4, 350) = 55.20, p < .001, η² = .39. Participants judged targets in male-dominated occupations as more agentic than targets without role information, p < .001, and targets in the employee role as similarly agentic as targets without role information, p = .998. Participants further judged targets in male-dominated occupations, in the employee role, and without role information as more agentic than homemakers, ps < .001, and targets in female-dominated occupations, ps < .001. As predicted by social role theory, these main effects were qualified by a Target Sex × Target Role interaction, F(4, 350) = 4.83, p = .001, η² = .05. Without role information, participants judged men as more agentic than women, p < .001, whereas with role information, they judged male and female targets as equally agentic, ps ≥ .335, with the predicted exception of judging the male employee as more agentic than the female employee, p = .004. The critical Target Role × Target Sex × Measure Type interaction predicted by the shifting standards model was nonsignificant, F(4, 350) = 0.65, p = .629, η² = .01, and no other effects of measure type approached significance, ps ≥ .629.

Discussion

The social role pattern of information about specific roles restraining perceived sex differences emerged on communion and agency in this experiment. Participants judged men as more agentic than women in the absence of role information and with the generic employee role but judged men and women more similarly in the other conditions. Although they judged women as more communal than men in all of the conditions, these differences were larger in the absence of role information and with the generic employee role than in the other, specific role conditions.

Like Experiment 1, this experiment did not yield the predicted effects of measure type. Therefore, we conclude that participants did not shift to within-sex standards in judging male and female role occupants on the subjective ratings scales. The finding that role effects were robust to variation of measure type thus appears to be generalizable, given that this experiment’s standardized personality common rule measure was different from Experiment 1’s behavioral frequency common rule measure.

General Discussion

Our two experiments examined the influence of information about social roles on judgments of the gender-stereotypic attributes of women and men. The central question that we addressed is how researchers should interpret the weakening or disappearance of judged differences between women and men, given information about their role occupancy. Are such effects genuine perceptual effects or merely by-products of switches to within-sex judgment standards in the presence of role information? For example, although women would be judged as more supportive and sympathetic than men, female and male nurses might be rated as equally supportive and sympathetic, perhaps merely because a male nurse is rated relative to a low male standard for these qualities. In contrast, the underlying process accounting for typical role effects in past research (e.g., Eagly & Steffen, 1984; Kite, 1996) could be consistent with the social role theory assumption that perceivers believe that occupational roles demand certain personality traits, regardless of role occupants’ sex. Alternatively, the underlying process could be consistent with the shifting standards assumption that perceivers judge role occupants in relation to their own sex—that is, an average male nurse in relation to other men. Perceivers might thus believe that such a man would be more communal than men in general but not more communal than female nurses.

Our method of resolving this interpretational ambiguity consisted of having participants use rating scales on which they could not shift their standards, to observe whether the role effects then disappeared. In both experiments, participants’ judgments were similar regardless of whether the experiment provided them with a subjective measure that allowed shifts of judgment standards or a common rule measure that did not allow such shifts. The use of two different common rule scales, both carefully constructed to assess judged agency and communion, strengthens our research.

These studies addressed reservations (e.g., Bridges et al., 2002) regarding social role interpretations of earlier findings (e.g., Eagly & Steffen, 1984, 1986). They also provided a much stronger test of social role theory and the shifting standards model than previous research (e.g., Bosak et al., 2007, 2008) by using two different types of common rule measures and additional social roles. Both experiments provide evidence in favor of the assumptions of social role theory in that specific role information restrained perceived sex differences regardless of measure type.

In review, the two experiments provide evidence in support of social role theory, given that the manipulation of measure type (subjective vs. common rule) did not influence the effects of roles on communion or agency. Experiment 1 showed that in the absence of role information, men were judged to be more agentic and less communal than women, whereas in the presence of role information men and women were judged to be equally agentic and communal. Experiment 2 produced similar findings except that specific role information weakened but did not eliminate the judged sex difference on communion. It is perhaps not surprising that gender stereotypes appeared more robust on communion than agency (albeit only in this experiment), given that other researchers have sometimes found that communion is more resistant to role effects than agency (e.g., Kite, 1996; Moskowitz, Suh, & Desaulniers, 1994).

This research helps to clarify the informational bases of gender stereotypes. According to the early formulation of social role theory (Eagly, 1987), the division of labor between the sexes is the fundamental cause of gender stereotypes because the communal behaviors associated with the homemaker role and the agentic behaviors associated with the breadwinner role have become associated with women and men, respectively. Therefore, men and women presented in homemaker and employee roles were judged as similarly communal and agentic (e.g., Eagly & Steffen, 1984). However, given the increase in women’s participation in the U.S. labor force and the continuing sex segregation of employment, the role of employee appears to have become so generic that it does not restrain gender-stereotypical judgments. These findings suggest that the information underlying gender stereotyping emerges now mainly from observations of women in communally demanding occupations, men in agentically demanding occupations, and women in the communally demanding homemaker role.

Adequacy of Common Rule Measures

One consideration in relation to interpreting the lack of shifting standards effects is whether our common rule measures were adequate. Following previous work by Biernat and colleagues (e.g., Biernat et al., 1991; Biernat & Manis, 1994; Kobrynowicz & Biernat, 1997), we used behavioral frequency scales and standardized personality scales as common rule measures. These common rule measures were designed to hold the meaning of the scales constant despite change in the rating targets and thus to restrain the possibility of participants shifting to within-sex standards. Despite the successful and frequent use of such common rule measures in shifting standards research, designing measures that are perfectly “common rule” remains challenging. In particular, research by Kunda and colleagues (e.g., Kunda & Sherman-Williams, 1993; Kunda, Sinclair, & Griffin, 1997) showed that the meaning of a trait or behavior can be qualitatively different for members of different social groups. This finding suggests that participants might have differently construed the behavioral opportunities described by the items of Experiment 1’s behavioral frequency scales, depending on the target person’s sex. However, in defense of this common rule measure, the specificity of its behavioral items should have restrained such reconstruals (e.g., “say ‘I love you’ to his [her] partner? _____times per week [max 10 times]”). In support of this claim about specificity, Kunda and Sherman-Williams (1993; Experiment 1) found that specific, unambiguous behaviors of target persons are not open to reconstruals and further interpretation. Thus, a general, ambiguous description of an aggressive act (“hit someone who annoyed him/her”) led to higher ratings of aggressiveness for the housewife than the construction worker because the act was construed as more aggressive for the housewife (e.g., “spanked her six year old when he trudged mud all over the carpet”) than for the construction worker (e.g., “decked a coworker who had been taunting him”). However, the housewife and the construction worker were judged similarly aggressive when they had performed either of these more specifically constructed disambiguated behaviors. Also, the framing of Experiment 2’s common rule measure as personality tests implied standardized testing that does not differ for women and men. In summary, our measures, which were designed to eliminate shifts to within-sex standards, probably also largely eliminated the types of qualitative meaning shifts demonstrated by Kunda and her colleagues (e.g., Kunda et al., 1997). Further investigation of these issues is advisable, possibly by comparing the implications of different types of common rule measures for shifts in judgment standards and meanings.

Possible Moderators of Shifting Standards and Social Role Effects

Both the shifting standards model of social judgment (Biernat, 2003, 2005) and social role theory (Eagly et al., 2000) remain helpful and valid for explaining stereotyping. Biernat (2003) has repeatedly demonstrated shifting standards effects in judgments of individual women and men on subjective scales. She thus stated that “when one judges individual members of stereotyped groups on stereotyped dimensions, one does so with reference to within-category standards” (p. 1019). Studies that have provided key evidence for the shifting standards model pertain to judgments of individual men and women, for example, by showing photographs of individuals (e.g., Biernat et al., 1991). Such research has found standard shifts that take the form of stronger stereotyping for individual men and women on common rule measures than subjective measures. Examples include experiments reporting standard shifts in judgments of individual fathers and mothers described as a very good or alright parent (see Kobrynowicz & Biernat, 1997) or in judgments of individual male and female captains in the U.S. Army (see Biernat, Crandall, Young, Kobrynowicz, & Halpin, 1998).

In contrast to these shifting standards experiments, the studies that have provided key evidence for social role theory pertain to judgments of social categories: These experiments presented role-identified stimulus persons as average role occupants (e.g., average female homemaker). The designation of role occupants as average implies a group of such individuals. Given this designation, we found no evidence that participants shifted their standards. These results suggest that subjective scales may foster shifts to within-sex standards for judgments of individual men and women but not for groups of men and women identified by their roles. Apparently social perceivers do engage in correspondent inferences from groups of role occupants to their role-consistent traits, whereas they decline to do so in relation to individual role occupants. These individual role occupants remain more characterized by the traits of their gender group, a phenomenon that is revealed on common rule scales.

Evidence for standard shift effects in judgments of individual male and female role occupants and the absence of such effects in judgments of average men and women identified by their roles makes sense in view of the processes of subtyping and subgrouping (see Richards & Hewstone, 2001). Studies that present individual role occupants such as Kevin, the very good parent (see Kobrynowicz & Biernat, 1997), or Susan, the firefighter, probably encourage perceivers to subtype the individual in relation to the general stereotype of men or women. For example, Susan, the firefighter, likely would be subtyped as a puzzling and normatively deviant woman and implicitly compared with women in general, whereas Peter, the firefighter, would be perceived as fitting the superordinate group of men. Compared with other women, Susan would gain high ratings on agency even though she would likely not be regarded as more agentic than male firefighters. Consistent with the shifting standards model, such subtyping and the associated use of within-sex standards would mask the influence of gender stereotypes on subjective scales, revealing instead null effects or even contrast effects (e.g., Susan as apparently more agentic than Peter), although gender-stereotypical judgments could still be observed on common rule measures.

In contrast to this subtyping of deviant individuals, groups of people are more likely to be subgrouped, which is a process that entails “organizing information into multiple clusters of individuals who are similar to one another in some way and different from other group members” (Maurer, Park, & Rothbart, 1995, p. 813). Therefore, studies that present stimulus persons as average role occupants such as “average female firefighter” likely encourage perceivers to construe these stimulus persons as a subgroup of women consisting of “female firefighters.” Research has identified other subgroups of women such as “woman athletes” or “housewives” and found them to be associated with distinctive, culturally shared stereotypes (see Clifton, McGrath, & Wick, 1976; Vonk & Olde-Monnikhof, 1998). This consideration of stimulus persons as a subgroup (e.g., female firefighters) would foster correspondent inference in relation to the occupational role (e.g., firefighting), resulting in the ascription of high agency and low communion to female firefighters—beliefs that are quite different from those typically held for the superordinate group of women.

Conclusion

The main purpose of this research was to determine whether the judgmental role effects that provide one of the foundations of social role theory were artifacts of participants’ shifting standards in the presence of role information. The answer to this question appears to be no. By comparing participants’ judgments on subjective and common rule scales that restrain standard shifts, our experiments confirmed the adequacy of the social role interpretation of earlier findings using similar designs.

Our findings have important implications. Social role theory conveys optimism for the capacity of social groups to change their stereotypes. The theory suggests that the key to stereotype change is changing a group’s social roles (Wood & Eagly, 2010). If it were true, as suggested by the shifting standards model, that men or women who do move into new roles are still perceived as generic men and women, this type of stereotype change would not occur. Instead, our research suggests that social perceivers do engage in correspondent inference in relation to groups of men and women portrayed in specific social roles. Therefore, as more members of each sex occupy unconventional social roles, the general stereotypes of the sexes should change accordingly. As Richards and Hewstone (2001) argued, subgroups are included in the superordinate group, which becomes broader as they encompass subgroups.

The implications of our analysis pertain to possibilities for changing groups’ disadvantageous stereotypes. From a practical standpoint, social programs could ameliorate such stereotypes by working to enhance the opportunities for members of disadvantaged groups to occupy desirable social roles. Although an individual’s role change would not affect group stereotypes, role changes affecting more group members should produce the subgrouping that would broaden the superordinate stereotype.

With respect to gender stereotypes, policies that could broaden them would foster men’s opportunities to enter communally demanding roles and women’s opportunities to enter agentically demanding roles. Given the 20th century role changes by which many women have entered agentically demanding occupational roles in the U.S. (see England, 2010), the female stereotype should now be broader than the male stereotype. In fact, there is evidence that this is so. For example, social perceivers ascribe masculine characteristics to women to a greater extent than they ascribe feminine characteristics to men (see present time conditions in Diekman & Eagly, 2000), and they are more approving of desirable counterstereotypical qualities in women than men (Diekman & Goodfriend, 2006; Prentice & Carranza, 2002). In conclusion, this evidence as well as the results of our own research suggest that the entry of groups of men and women into nontraditional roles will broaden the overarching stereotypes of both sexes and thereby foster greater behavioral flexibility.

Footnotes

Appendix

Acknowledgements

The authors thank Anne Koenig and Anna Kirchner for research assistance.

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

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported in part by a research fellowship from the Federal State of Baden-Württemberg, Germany.

References

Biernat

(1995). The shifting standards model: Implications of stereotype accuracy for social judgment. In Yueh-Ting

Jussim

L. J.

(Eds.), Stereotype accuracy: Toward appreciating group differences (pp. 87-114). Washington, DC: American Psychological Association. doi:10.1037/10495-00410.1037/10495-004

Biernat

(2003). Toward a broader view of social stereotyping. American Psychologist, 58, 1019-1027. doi:10.1037/0003-066X.58.12.101910.1037/0003-066X.58.12.1019

Biernat

(2005). Standards and expectancies: Contrast and assimilation in judgments of self and others. New York, NY: Psychology Press.

Biernat

Crandall

C. S.

Young

L. V.

Kobrynowicz

Halpin

S. M.

(1998). All that you can be: Stereotyping of self and others in a military context. Journal of Personality and Social Psychology, 75, 301-317. doi:10.1037/0022-3514.75.2.30110.1037/0022-3514.75.2.301

Biernat

Manis

(1994). Shifting standards and stereotype-based judgments. Journal of Personality and Social Psychology, 66, 5-20. doi:10.1037/0022-3514.66.1.510.1037/0022-3514.66.1.5

Biernat

Manis

Nelson

T. E.

(1991). Stereotypes and standards of judgment. Journal of Personality and Social Psychology, 60, 485-499. doi:10.1037/0022-3514.60.4.48510.1037/0022-3514.60.4.485

Bosak

Sczesny

Eagly

A. H.

(2007). Die Bedeutung von Informationen zur sozialen Rolle für die Reduktion geschlechtsstereotypen Urteilens: Ein methodisches Artefakt? [The impact of social role information on reducing gender-stereotypical judgments: A methodological artifact?]. Zeitschrift für Sozialpsychologie, 38, 277-284. doi:10.1024/0044-3514.38.4.27710.1024/0044-3514.38.4.277

Bosak

Sczesny

Eagly

A. H.

(2008). Communion and agency judgments of women and men as function of role information and response format. European Journal of Social Psychology, 38, 1148-1155. doi:10.1002/ejsp.53810.1002/ejsp.538

Bridges

Etaugh

Barnes-Farrell

(2002). Trait judgments of stay-at-home and employed parents: A function of social role and/or shifting standards? Psychology of Women Quarterly, 26, 140-150. doi:10.1111/1471-6402.0005210.1111/1471-6402.00052

10.

Buss

D. M.

Craik

K. H.

(1980). The frequency concept of disposition: Dominance and prototypically dominant acts. Journal of Personality, 48, 379-392. doi:10.1111/j.1467-6494.1980.tb00840.x10.1111/j.1467-6494.1980.tb00840.x

11.

Cejka

M. A.

Eagly

A. H.

(1999). Gender-stereotypic images of occupations correspond to the sex segregation of employment. Personality and Social Psychology Bulletin, 25, 413-423. doi:10.1177/014616729902500400210.1177/0146167299025004002

12.

Clifton

A. K.

McGrath

Wick

(1976). Stereotypes of women: A single category? Sex Roles, 2, 135-148. doi:10.1007/BF0028724610.1007/BF00287246

13.

Diekman

A. B.

Eagly

A. H.

(2000). Stereotypes as dynamic constructs: Women and men of the past, present, and future. Personality and Social Psychology Bulletin, 26, 1171-1188. doi:10.1177/014616720026200110.1177/0146167200262001

14.

Diekman

A. B.

Goodfriend

(2006). Rolling with the changes: A role congruity perspective on gender norms. Psychology of Women Quarterly, 30, 369-383. doi:10.1111/j.1471-6402.2006.00312.x10.1111/j.1471-6402.2006.00312.x

15.

Eagly

A. H.

(1987). Sex differences in social behavior: A social-role interpretation. Hillsdale, NJ: Lawrence Erlbaum.

16.

Eagly

A. H.

Steffen

V. J.

(1984). Gender stereotypes stem from the distribution of women and men into social roles. Journal of Personality and Social Psychology, 46, 735-754. doi:10.1037/0022-3514.46.4.73510.1037/0022-3514.46.4.735

17.

Eagly

A. H.

Steffen

V. J.

(1986). Gender stereotypes, occupational roles, and beliefs about part-time employees. Psychology of Women Quarterly, 10, 252-262. doi:10.1111/j.1471-6402.1986.tb00751.x10.1111/j.1471-6402.1986.tb00751.x

18.

Eagly

A. H.

Steffen

V. J.

(1988). A note on assessing stereotypes. Personality and Social Psychology Bulletin, 14, 676-680. doi:10.1177/014616728814400310.1177/0146167288144003

19.

Eagly

A. H.

Wood

(in press). Social role theory. In van Lange

Kruglanski

Higgins

E. T.

(Eds.), Handbook of theories in social psychology. London, UK: Sage.

20.

Eagly

A. H.

Wood

Diekman

A. B.

(2000). Social role theory of sex differences and similarities: A current appraisal. In Eckes

Trautner

H. M.

(Eds.), The developmental social psychology of gender (pp. 123-174). Mahwah, NJ: Lawrence Erlbaum.

21.

England

(2010). The gender revolution: Uneven and stalled. Gender & Society, 24, 149-166. doi:10.1177/089124320936147510.1177/0891243209361475

22.

Gawronski

(2003). Implicational schemata and the correspondence bias: On the diagnostic value of situationally constrained behavior. Journal of Personality and Social Psychology, 84, 1154-1171. doi:10.1037/0022-3514.84.6.115410.1037/0022-3514.84.6.1154

23.

Gilbert

D. T.

(1998). Ordinary personology. In Gilbert

D. T.

Fiske

S. T.

Lindzey

(Eds.), The handbook of social psychology (pp. 788-827). New York, NY: McGraw-Hill.

24.

Gilbert

D. T.

Malone

P. S.

(1995). The correspondence bias. Psychological Bulletin, 117, 21-38. doi:10.1037/0033-2909.117.1.2110.1037/0033-2909.117.1.21

25.

Harrison

L. A.

(2005). Social role theory and the perceived gender role orientation of athletes. Sex Roles, 52, 227-236. doi:10.1007/s11199-005-1297-110.1007/s11199-005-1297-1

26.

Kite

M. E.

(1996). Age, gender, and occupational label: A test of social role theory. Psychology of Women Quarterly, 20, 361-374. doi.org/10.1111/j.1471-6402.1996.tb00305.x

27.

Kobrynowicz

Biernat

(1997). Decoding subjective evaluations. How stereotypes provide shifting standards. Journal of Experimental Social Psychology, 33, 579-601. doi:10.1006/jesp.1997.133810.1006/jesp.1997.1338

28.

Kunda

Sherman-Williams

(1993). Stereotypes and the construal of individuating information. Personality and Social Psychology Bulletin, 19, 90-99. doi:10.1177/014616729319101010.1177/0146167293191010

29.

Kunda

Sinclair

Griffin

(1997). Equal ratings, but separate meanings: Stereotypes and the construal of traits. Journal of Personality and Social Psychology, 72, 720-734. doi:10.1037/0022-3514.72.4.72010.1037/0022-3514.72.4.720

30.

Maurer

K. L.

Park

Rothbart

(1995). Subtyping versus subgrouping processes in stereotype representation. Journal of Personality and Social Psychology, 69, 812-824. doi:10.1037/0022-3514.69.5.81210.1037/0022-3514.69.5.812

31.

Moskowitz

D. S.

Suh

E. J.

Desaulniers

(1994). Situational influences on gender differences in agency and communion. Journal of Personality and Social Psychology, 66, 753-761. doi:10.1037//0022-3514.66.4.75310.1037//0022-3514.66.4.753

32.

Park

Smith

J. A.

Correll

(2008). “Having it all” or “doing it all”? Perceived trait attributes and behavioral obligations as a function of work role and gender. European Journal of Social Psychology, 38, 1156-1164. doi:10.1002/ejsp.53510.1002/ejsp.535

33.

Prentice

D. A.

Carranza

(2002). What women and men should be, shouldn’t be, are allowed to be, and don’t have to be: The contents of prescriptive gender stereotypes. Psychology of Women Quarterly, 26, 269-281. doi:10.1111/1471-6402.t01-1-0006610.1111/1471-6402.t01-1-00066

34.

Richards

Hewstone

(2001). Subtyping and subgrouping: Processes for the prevention and promotion of stereotype change. Personality and Social Psychology Review, 5, 52-73. doi:10.1207/S15327957PSPR0501_410.1207/S15327957PSPR0501_4

35.

Schein

V. E.

(2001). A global look at psychological barriers to women’s progress in management. Journal of Social Issues, 57, 675-688. doi:10.1111/0022-4537.0023510.1111/0022-4537.00235

36.

Sczesny

Kühnen

(2004). Meta-cognition about biological sex and gender-stereotypic physical appearance: Consequences for the assessment of leadership competence. Personality and Social Psychology Bulletin, 30, 13-21. doi:10.1177/014616720325883110.1177/0146167203258831

37.

Tomaskovic-Devey

Zimmer

Stainback

Robinson

Taylor

McTague

(2006). Documenting desegregation: Segregation in American workplaces by race, ethnicity, and sex, 1966-2003. American Sociological Review, 71, 565-588. doi:10.1177/00031224060710040310.1177/000312240607100403

38.

U.S. Department of Labor, U.S. Bureau of Labor Statistics. (2010). Women in the labor force: A databook. Retrieved from http://www.bls.gov/cps/wlftable11.htm

39.

Vonk

Olde-Monnikhof

(1998). Gender subgroups: Intergroup bias within the sexes. European Journal of Social Psychology, 28, 37-47. doi:10.1002/(SICI)1099-0992(199801/02)10.1002/(SICI)1099-0992(199801/02)

40.

Wood

Eagly

A. H.

(2010). Gender. In Fiske

Gilbert

Lindzey

(Eds.), Handbook of social psychology (Vol. 1, pp. 629-667). New York, NY: John Wiley.