Gender Stereotypes Stem From the Distribution of …gribouts.free.fr/psycho/menace%20du%20st%e9r%e9o/... · to hold positions of lower status and authority, ... Experiments 1 and

Journal or Personality and Social Psychology1984, Vol. 46, No. 4, 735-754

Copyright 1084 bv thei Psychological Association, Inc.

Gender Stereotypes Stem From the Distribution ofWomen and Men Into Social Roles

Alice H. Eagly and Valerie J. SteffenPurdue University

According to stereotypic beliefs about the sexes, women are more communal (selflessand concerned with others) and less agentic (self-assertive and motivated to master)than men. These beliefs were hypothesized to stem from perceivers' observationsof women and men in differing social roles: (a) Women are more likely than mento hold positions of lower status and authority, and (b) women are more likelythan men to be homemakers and are less likely to be employed in the paid workforce. Experiments 1 and 2 failed to support the hypothesis that observed sexdifferences in status underlie belief in female communal qualities and male agenticqualities. Experiment 3 supported the hypothesis that observed sex differences indistribution into homemaker and employee occupational roles account for thesebeliefs. In this experiment, subjects perceived the average woman and man stereo-typically. Female and male homemakers were perceived as high in communion andlow in agency. Female and male employees were perceived as low in communionand high in agency, although female employees were perceived as even more agenticthan their male counterparts. Experiments 4 and 5 examined perceptions thatmight account for the belief that employed women are especially agentic: (a) Adouble burden of employment plus family responsibilities did not account for thisbelief, and (b) freedom of choice about being employed accounted for it reason-ably well.

Gender stereotypes, like other social stereo-types, reflect perceivers' observations of whatpeople do in daily life. If perceivers often ob-serve a particular group of people engaging ina particular activity, they are likely to believethat the abilities and personality attributes re-quired to carry out that activity are typical ofthat group of people. For example, if perceiversconsistently observe women caring for chil-

The research reported in this article was supported byNational Science Foundation Grants BNS-7711671, BNS-7924471, and BNS-8023311. A preliminary report waspresented at the meetings of the Midwestern PsychologicalAssociation, May 1982, and the American PsychologicalAssociation, August 1982.

The authors thank Wendy Wood for help with the designand statistical analyses of Experiment 1; Vathsala Venu-gopalan for help with the statistical analyses of Experiments1 and 2; and Barry Brumer, Deborah Rugs, Jan Schafer,Erik Bronner, and Gail Winbury for assistance in admin-istering materials to subjects. The authors also thank ShellyChaiken, Kay Deaux, Chester Insko, Dafna Izraeli, LaurieLewis, Myron Rothbart, Wendy Wood, and two anony-mous reviewers for their comments on a draft of the article.

Requests for reprints should be sent to Alice H. Eagly,Department of Psychological Sciences, Purdue University,West Lafayette, Indiana 47907.

dren, they are likely to believe that charac-teristics thought to be necessary for child care,such as nurturance and warmth, are typicalof women. Because most of people's activitiesare determined by their various social roles,stereotypes about groups of people should re-flect the distribution of these groups into socialroles in a society. Furthermore, certain ste-reotypes may reflect the distribution of groupsinto broader aspects of social structure suchas social class. For example, beliefs about racialdifferences may be based at least in part onobservations that racial groups differ in socialclass (Feldman, 1972; Smedley & Bayton,1978; Triandis, 1977).

In applying this social structural analysis topeople's beliefs about gender, we faced twoissues: (a) What is the content of stereotypesabout women and men? (b) What are the majordifferences in the ways that women and menare distributed into social roles? Concerningcontent, we decided to restrict our focus tothe beliefs about gender that, by virtue of thefrequency with which they have been docu-mented by research and amplified in theoret-ical discussions, appear to be most important.

735

736 ALICE H. EAGLY AND VALERIE J. STEFFEN

These beliefs concern communal and agenticpersonal qualities: Perceivers generally assumethat men are oriented toward agentic goalsand women toward communal goals (e.g.,Bern, 1974; Block, 1973; Broverman, Vogel,Broverman, Clarkson, & Rosenkrantz, 1972;Spence & Helmreich, 1978). Following Bak-an's (1966) discussion of this distinction,agentic qualities are manifested by self-asser-tion, self-expansion, and the urge to master,whereas communal qualities are manifestedby selflessness, concern with others, and a de-sire to be at one with others. This distinctionhas been accorded considerable importance intheoretical discussions of gender (Bakan, 1966;Parsons, 1955) and in the development ofmeasures of sex-typed and androgynous per-sonalities (e.g., Bern, 1974; Spence & Helm-reich, 1978). Furthermore, these beliefs aboutgender appear to be cross-culturally general(Williams & Best, 1982).

To examine why women are perceived ascommunal and men as agentic, we consideredtwo major differences in the distribution offemales and males into social roles. The firstof these differences is that women are morelikely than men to hold positions at low levelsin hierarchies of status and authority and areless likely to hold higher level positions. Thesecond difference is that women are more likelythan men to be homemakers and are less likelyto be employed in the paid work force.

Given the pervasiveness in natural settingsof sex differences in status, it seems plausiblethat gender stereotypes stem from the tendencyof perceivers to observe women in lower statusroles than men. Such observations would bemade in organizational settings in which thepositions held by men tend to be higher instatus and authority than the positions heldby women (e.g., Brown, 1979; England, 1979;Kanter, 1977; Mennerick, 1975). Also, infamily settings husbands tend to have an over-all power and status advantage over wives(Blood & Wolfe, 1960; Gillespie, 1971; Scan-zoni, 1982). Such differences in the status ofmen's and women's roles may be determiningfactors in beliefs about gender.

Another reason for examining status dif-ferences is that recent research by Eagly andWood (1982) demonstrated that the stereotypicbeliefs that women are more easily influencedthan are men and that men exert influence

more easily than do women (e.g., Brovermanet al., 1972; Spence & Helmreich, 1978; Taylor,Fiske, Etcoff, & Ruderman, 1978) stem fromperceivers' inferences that (a) women occupylower status positions than men and (b) thelower an individual's status relative to otherpersons, the more that individual yields to theirinfluence. To extend this analysis to the com-munal and agentic aspects of gender stereo-types, we hypothesized that people who arehigher in status and authority have been ob-served to behave with less communion (self-lessness and concern with others) and moreagency (self-assertion and urge to master) thanthose who have lower status positions. There-fore, perceivers' observations that women oc-cupy lower status positions than men may leadthem to believe that women are more com-munal and less agentic than men.

By a similar logic, the differing distributionsof women and men into the roles of home-maker and employee may account for the ste-reotypic beliefs that women are communal andmen are agentic. Because the labor-force par-ticipation rates of women (51.2%) and men(77.2%) still differ considerably (U.S. Depart-ment of Labor, 1980), perceivers are likely tohave observed fewer women than men in em-ployee roles and almost exclusively women inthe homemaker role. The perception of womenas less agentic and more communal than menwould follow if employees have been observedto behave more assertively and masterfullythan homemakers as well as less selflessly andsupportively toward others. According to thisanalysis, the stereotypic differences betweenwomen and men should parallel the differencesthat people perceive between homemakers andemployees. Some empirical support for thishypothesis is provided by Clifton, McGrath,and Wick's (1976) finding that the communalattributes ordinarily ascribed to women wereassigned only to housewives and not to fourother categories of women (female athlete, ca-reer woman, club woman, and "bunny").

The experiments that we have carried outto test these ideas share several features ofdesign. To minimize demand characteristicsstemming from subjects' knowledge of our hy-potheses, each subject read a description ofonly one woman or man. In addition, the as-pect of social roles presumed to account forgender stereotypes (hierarchical status, or oc-

GENDER STEREOTYPES 737

cupation as a homemaker or employee) wasvaried: (a) In Experiments 1 and 2, which ex-amined hierarchical status, some stimuluspersons had high-status job titles and somehad low-status job titles, and (b) in Experiment3, which examined the homemaker-employeedistinction, some stimulus persons werehomemakers and others were employees. Forother stimulus persons,, this stereotype-relevantaspect of the social role was omitted: (a) Inthe status experiments, the job title was omit-ted, and (b) in the homemaker-employee ex-periment, designation as a homemaker or em-ployed person was omitted. It was in theseexperimental conditions, in which the role de-scriptions were omitted, that subjects shouldhave manifested gender stereotyping. Accord-ing to our analysis, perceivers view men andwomen stereotypicaUy in the absence of roleinformation, because under such conditionsthe attributes ascribed to women and men re-flect the differing social roles that underlie thestereotypes. In contrast, the addition of roledescriptions to female and male stimulus per-sons prevents gender-stereotypic judgments ifsuch descriptions (e.g., job titles) provide clear-cut information about the aspect of social rolesthat ordinarily covaries with sex (e.g., hier-archical status). In the presence of such roleinformation, the covariation of sex and rolethat is the implicit basis of gender stereotypesis removed, and role would determine per-ceivers' beliefs about people's attributes.Women and men who have the same rolewould be perceived equivalently.

Experiments 1 and 2Method

SubjectsIn Experiment 1, 276 females and 208 males partici-

pated. Of these subjects, 256 were University of Massa-chusetts psychology students who participated in a lab-oratory setting to obtain extra-credit course points. Anexperimenter randomly selected the remaining 228 subjectsby choosing on repeated occasions every fourth personseated in a University of Massachusetts coffee shop. InExperiment 2, 237 females and 243 males participated.One female and one male experimenter each randomlyselected half of the subjects by choosing every fourth personseated in a coffee shop or general library at Purdue Uni-versity. Especially in Experiment 1, the subjects sampledfrom the public campus locations included university staffas well as students. la the experiments reported in thisarticle, 80% or more of the persons selected from such

locations agreed to participate. Subjects' mean age was22.30 years in Experiment 1 and 21.81 years in Experi-ment 2.

ProcedureEach subject read a brief description of an employee

(e.g., "Phil Moore is about 35 years old and has beenemployed for a number of years by a supermarket. He isone of the managers") and rated this stimulus person.The descriptions in Experiment 1 varied according to a2 X 3 X 2 (female vs. male X high-status job title vs. low-status job title vs. no job title X bank vs. supermarketsetting) factorial design. The design of Experiment 2 dif-fered with respect to the setting variable, which had fourlevels because a medical clinic and a university departmentof biology were added.

In the laboratory sessions of Experiment 1, a femaleexperimenter administered materials to subjects in groupsof about 25. Subjects first indicated their age and sex. Toensure that subjects thought carefully about the stimulusperson, the experimenter had them "spend a momentthinking about" the stimulus person (after reading thedescription) and then write a few sentences about the per-son. Subjects then responded to the measures describedbelow.

At the public campus locations in both experiments,an experimenter approached each subject by asking heror him to participate in a study on "impressions of otherpeople." After the subject had completed the questionnaire,the experimenter asked her or his-age and recorded thisinformation along with the subject's sex.

Manipulation of Independent VariablesSex of stimulus person. The stimulus persons were

either female or male. Sex was identified by sex-typednames (e.g., Sue Fisher, Phil Moore).

Status of job title and setting of jab. The stimuluspersons had either a high-status job title, a low-status jobtitle, or no job title. To provide an internal replication ofthe design, the stimulus persons were described as employedby a bank or a supermarket and, in Experiment 2, alsoby a medical clinic or a university department of biology.The high- and low-status job titles for these settings werevice-president and teller, manager and cashier, physicianand x-ray technician, and professor and lab technician,respectively.

Measuring Instruments1

Beliefs about stereotypic attributes. Using 5-pointscales, subjects rated the stimulus persons on 18 attributes,presented either as personality characteristics (Experiment1) or as.attributes of on-the-job behavior (Experiment 2).Each on-the-job rating scale was preceded by a questionasking how much of the attribute the stimulus personexhibited on the job (e.g., "How competitive do you thinkthis person is on the job?").

' For all experiments, measures are listed in the orderin which they were administered.

738 ALICE H. EAGLY AND VALERIE J. STEPHEN

Attributes were selected primarily from the PersonalAttributes Questionnaire (Spence & Helmreich, 1978) toensure that they (a) represented gender stereotypes and(b) included both communal and agentic qualities. A factoranalysis (varimax orthogonal rotation) of subjects' ratingswas performed for each of the five experiments in thisseries. All analyses yielded similar two-factor solutions.One factor, labeled communal, accounted for an averageof 17.4% of the variance in the five experiments, and an-other factor, labeled agentic, accounted for an average of23.9% of the variance. Although various other labels havebeen used by researchers to characterize these dimensions(e.g., expressiveness vs. instrumentality, social orientationvs. task orientation, and femininity vs. masculinity), noneprovided as close a match to the actual content of thefactors as did Bakan's (1966) terms, communion andagency.

The measure of perceived communion was the meanof each subject's ratings on the attributes that (in the fiveexperiments) consistently loaded highly on the communalfactor: kind, helpful, understanding, warm, aware of others'feelings, and (in all but Experiment 1, which omitted thisattribute) able to devote self to others. The measure ofperceived agency was the mean of each subject's ratingson the attributes that consistently loaded highly on theagentic factor: active, not easily influenced, aggressive, in-dependent, dominant, self-confident, competitive, makesdecisions easily, never gives up easily, and (in all but Ex-periment 1, which omitted this scale) stands up well underpressure. These measures had satisfactory internal con-sistency: The mean values of coefficient alpha (Cronbach,1951) in the five experiments were .84 for communionand .86 for agency. The findings of all five experimentsare presented in terms of these measures. Because fewsignificant effects were obtained on several rating scalesnot included in these measures, these findings will not bereported.

Inferred job status. In the laboratory portion of Ex-periment 1 and in Experiment 2, subjects estimated (indollars) the stimulus person's annual salary. In Experiment2, for stimulus persons who had no job title, subjects alsogave their "best guess" concerning the individual's jobtitle. Coders blind to the experimental conditions dividedthe job titles into (a) a high-status category consisting ofjobs that either included an administrative or managerialcomponent or required high-level technical skills and (b)a low-status category of other jobs. Coders agreed on ap-proximately 95% of these relatively objective judgments,and disagreements were resolved by discussion. In the lab-oratory portion of Experiment 1, subjects also rated thestatus of the stimulus person's job on a 15-point scaleranging from low status to high status.2

Results

The principal data analyses were Sex ofStimulus Person X Status of Job Title X Settinganalyses of variance (ANOVAS). Because no sig-nificant effects occurred in Experiment 1 forlocale in which stimulus materials were ad-ministered (coffee shop vs. laboratory), thisvariable was dropped from the analyses re-

ported here. Also, analyses including subjectsex as an additional variable yielded very fewdifferences between female and male subjectsfor any of the experiments in this series.3

Therefore, this variable was also dropped fromall analyses reported in this article. Finally,because analyses including experimenter (fe-male vs. male) as an additional factor yieldedvery few effects in Experiment 2 (or in Ex-periments 3 and 5, which also used one ex-perimenter of each sex), this variable wasdropped from all reported analyses.

Inferred Job Status

Salary estimates. On subjects' estimates ofthe stimulus persons' salaries, the main effectsof sex of stimulus person and status of jobtitle were highly significant for both Experi-ments 1 and 2.4 Women were judged to havelower salaries than men: For Experiment 1,Ms = $12,154 versus $15,425, respectively,F(\, 244) = 27.24, p < .001, and for Exper-iment 2, Ms = $19,749 versus $23,535, re-spectively, F(\, 449) = 15.29, p < .001. Con-sistent with the significant main effects of sta-tus—for Experiment 1, F(2, 244) = 50.58,and for Experiment 2, F(2, 449) = 137.28,ps < .001—persons with high-status job titleswere judged to earn considerably more thanpersons whose job titles were not given, who

2 Because this rating-scale measure proved insensitiveto differences between conditions, it was not included inadditional studies and will not be discussed further.

3 The absence of Sex of Subject X Sex of Stimulus Personinteractions is noteworthy in view of research suggestingthat in-group members perceive in-groups more favorablyand less stereotypically and homogeneously than they per-ceive out-groups (e.g., Brewer, 1979; Tajfel, 1981; see alsoPark & Rothbart, 1982, regarding perceptions of womenand men). Perhaps this in-group-out-group bias is a man-ifestation of the self-enhancing tendency in person per-ception (e.g., Zuckerman, 1979) and occurs when re-spondents rate in-group members as they would ratethemselves. It is likely that in the present experiments,subjects merely retrieved their concepts of various groupsof people (e.g., male bank tellers, employed women, averagemen) and did not treat themselves as exemplars of thesame-sex categories.

4 Because the variance of subjects' salary estimates wasextremely heterogeneous, with larger means associated withlarger variances, analyses were performed on the logarithmof the salaries in all experiments in this series. There wasno serious heterogeneity on the other dependent variablesin these experiments.


in turn were judged to earn more than personswith low-status job titles: For Experiment 1,Ms = $19,236 versus $11,773, p < .001, and$11,773 versus $10,120, p < .025; for Exper-iment 2, Ms = $32,057 versus $18,548, p<.001, and $18,548 versus $14,359, p < .001.The Sex X Status interactions were nonsignif-icant, F(2, 244) = 1.64 and P(2, 449) = 1.61.

The findings that women were judged tohave lower salaries than men, regardless ofwhether job titles were indicated, suggest thatthe salary estimates may have functionedlargely as a measure of perceived wage dis-crimination. To provide clear-cut evidence thatlower status was associated with women morethan with men, the tendency to ascribe lowersalaries to women should have been especiallystrong in conditions omitting job titles, inwhich the typical status difference was notcountered by information that equated wom-en's and men's jobs. Thus the absence of theexpected Sex X Status interaction in Experi-ment 1 led us to include a purer indicator ofstatus, job title guesses, in Experiment 2.

Job title guesses. For stimulus personswithout job titles in Experiment 2, a greaterproportion of subjects' job-title guesses werecategorized as high status (vs. low status) formale than for female stimulus persons. High-status job titles were ascribed to 48 men and30 women, and low-status job titles were as-cribed to 30 men and 39 women. A loglinearanalysis of the cell frequencies (Bishop, Fien-berg, & Holland, 1975; Davis, 1974) revealeda significant likelihood ratio chi-square (G2 =4.82, p < .05) for the interaction between sexof stimulus person and judged job-title status.Thus the job-title measure of inferred statusyielded clear evidence that without knowingjob titles, subjects inferred that women heldlower status positions than men.

Beliefs About Stereotypic Attributes

Subjects' mean ratings of the stimulus per-sons' communion and agency appear in Tables1 and 2. On communion, the main effect ofsex of stimulus person was marginally signif-icant in Experiment 1 and significant in Ex-periment 2. Women were perceived as morecommunal than men: For Experiment 1, Ms =3.61 versus 3.48, respectively, F(l, 472) = 3.71,p < .06; for Experiment 2, Ms = 3.53 versus

Table 1Mean Ratings of Stereotypic Attributes of Femaleand Male Employees Who Varied in Statusof Job Title: Experiment 1

Stimulusperson

FemaleCommunalAgentic

MaleCommunalAgentic

Highstatus

3.553.74

3.413.52

Lowstatus

3.592.63

3.582.43

Notitle

3.692.85

3.482.78

Note. Means are on a 5-point scale; larger numbers indicategreater communion or agency. Cell ns ranged from 78 to85. For communal, MS, = 0.54; for agentic, MS, = 0.44.

3.34, F(l, 456) = 9.95, p < .002. Consistentwith a significant Sex X Status interaction inExperiment 2 only, F(2,456) = 3.46, p < .05,this stronger communal tendency was ascribedto women (vs. men) with low-status or no jobtitles (ps < .01 or smaller) but was nonsig-nificant with high-status job titles.

On agency, the main effects of sex and statuswere highly significant in both experiments.Women were perceived as more agentic thanmen: For Experiment 1, Ms = 3.08 versus2.92, respectively, F(\, 472) = 7.80, p < .005;for .Experiment 2, Ms = 3.53 versus 3.37, F[l,456) = 9,99, p < .002. Consistent with thesignificant main effects of status—for Exper-iment 1, F(2, 472) = 121.01, and for Exper-iment 2, F(2,456) = 21.33.ps < .001—personswith high-status job titles were perceived as

Table 2Mean Ratings qfOn-the-Job Behavior of Femaleand Male Employees Who Varied in Statusof Job Title: Experiment 2

Stimulusperson


MaleCommunalAgentic

Highstatus

3.483:77

3.513.61

Lowstatus

3.533.37

3.213.23

Notitle

3.573.46

3.303.25

Note. Means are on a 5-point scale; larger numbers indicategreater communion or agency. All cell «s = 80. For com-munal, MS, = 0.41; for agentic, MS, = 0.34.


more agentic than persons whose job titles werenot given or persons with low-status job titles:For Experiment 1, Ms - 3.63 versus 2.81 forhigh-status versus no job title and 3.63 versus2.53 for high- versus low-status job title, re-spectively, ps < .001; for Experiment 2, Ms =3.69 versus 3.36, and 3.69 versus 3.30, re-spectively, ps < .001. In Experiment 1, thecontrast between the no job title and low-status job title conditions was also significant(p < .001).5

DiscussionSeveral aspects of these findings are unfa-

vorable to the hypothesis that the stereotypesof female communion and male agency stemfrom having observed women in lower statusroles than men. First, our status-difference ex-planation of gender stereotypes implies thatperceptions of lower status persons resemblethose of average women and that perceptionsof higher status persons resemble those of av-erage men. This expectation was only partlyconfirmed: Persons with low-status job titleswere perceived as considerably less agentic thanpersons with high-status job titles, but therewas no difference in communion.

The second and most surprising aspect ofthe findings that discount the status hypothesisis the counterstereotypic effect that the sex ofthe stimulus persons had on the ascription ofagentic traits. Women were perceived as moreagentic than men despite the perceptions that(a) women earn less than men and hold lowerstatus positions and (b) persons in lower statuspositions are less agentic than persons in higherstatus positions. Furthermore, women wereperceived as more communal than men, eventhough persons in lower and higher status po-sitions did not differ in perceived communion.

These perceived sex differences were rela-tively small, despite their statistical signifi-cance. The differences were no larger than .19on a 5-point scale, with an effect size (d) of.30 (Cohen, 1977). Nevertheless, several as-pects of the experiments promote confidencein the reliability of the findings. First, the gen-eralizability of the findings across subjects re-cruited at two universities and by two differentmethods (subject pool and random samplingat public campus locations) reduces the like-lihood that artifacts arose from particular sub-ject populations. Second, the generalizability

of the findings across four organizational set-tings reduces the likelihood that artifacts arosefrom beliefs about particular work environ-ments. Third, the generalizability of the find-ings across two different types of ratings (per-sonality traits and on-the-job behavior) reducesthe likelihood that artifacts arose from insuf-ficient sensitivity of global trait ratings. Be-cause inferences from job titles to attributesof on-the-job behavior should be easier andmore direct than inferences from job titles topersonality traits, the job behavior ratings ofExperiment 2 should have maximized thepossibility of obtaining any effects of explicitand implicit variation of job status.

Our findings may be more consistent withthe hypothesis that distributions of women andmen into homemaker and employee roles un-derlie gender stereotypes. Because all of thestimulus persons presented in Experiments 1and 2 were described as employed, the relativeabsence of gender-stereotypic perceptions mayreflect the inclusion of this information aboutoccupational role. Therefore, our third ex-periment tested the hypothesis that gender ste-reotypes stem from perceivers' observations ofthe distribution of women and men into theroles of homemaker and employed person. Inthis experiment, some female and male stim-ulus persons were described as employees andsome as homemakers, and the occupation ofothers was not indicated.

One implication of the homemaker versusemployee explanation of gender stereotypes isthat the differences perceived between personsin these two occupational roles parallel thestereotypic differences between women andmen. Therefore, homemakers were expectedto be perceived as more communal and lessagentic than employed persons. It also followsthat women and men are perceived stereotyp-ically when their role assignment as home-

5 In Experiments 1 and 2, several effects of employmentsetting were also obtained on beliefs and on inferred status.Only one of these effects involved the sex of the stimuluspersons. Consistent with this Sex X Setting interactionobtained in Experiment 2 on perceived agency (p < .01),subjects rated women's behavior as more agentic than men'sin the bank and the supermarket (ps < .01), as marginallymore agentic than men's in the university department ofbiology (p < .09), but not different from men's in themedical clinic.


maker or employee is unknown, because per-ceivers have observed that more women thanmen are homemakers and fewer women areemployees. Furthermore, the homemaker-employee analysis implies that women andmen are perceived similarly if they have thesame occupational role, that is, if both arehomemakers or both are employees. Yet, con-sistent with the agency findings of the first twoexperiments and inconsistent with our socialrole analysis, we expected that female em-ployees would be perceived as somewhat moreagentic than male employees. We were lessconfident that female employees would beperceived as more communal because of themarginal significance of this finding in Ex-periment 1.

The design of our third experiment also al-lowed us to examine two possible explanationsof the relatively high agency ratings of em-ployed women. One explanation is that re-pondents are no longer willing to derogatewomen on stereotype questionnaires becauseof changes in attitudes toward women, greaterwariness about revealing one's stereotyping,or possibly other causes. This explanation im-plies that it would be impossible to replicatethe stereotypes of women and men obtainedby other investigators (e.g., Broverman et al.,1972; Spence & Helmreich, 1978). The inclu-sion of average woman and average man cuesin our third experiment allowed us to examinethis issue.

According to another explanation for thehigh level of agency ascribed to employedwomen, subjects believed that women are lesslikely to be employed than men and thereforeinferred that higher standards are applied towomen than to men by employers (or bywomen themselves) when women are selectedfor jobs. This selection hypothesis implies thatpeople who are thus highly selected (or self-selected) for a role are believed to be moreextreme in role-relevant characteristics. Ifagentic qualities are believed to be relevant tojob success, employed women would be per-ceived as being more agentic than employedmen. It also follows that male homemakerswould be perceived as being more communalthan female homemakers (provided that com-munal characteristics are believed to be rel-evant to the homemaker role). The inclusionof homemaker stimulus persons (as well as

employees) allowed us to examine these se-lection considerations.

The greater agency ascribed to employedwomen (vs. men) is not plausibly explainedin terms of a belief that discrimination makesit necessary for female employees to be morequalified than their male counterparts. Peopleare likely to believe that discrimination existsin relation to high-status positions or othermale-dominated jobs, for which traditionallythere were barriers excluding or discouragingwomen. However, in Experiments 1 and 2,subjects were found to believe in women's su-periority in agentic qualities when low-statusas well as high-status job titles were given. Itseems unlikely that subjects believed thatwomen face discrimination in obtaining thelow-status positions used in our research (e.g.,bank teller, supermarket cashier).

Experiment 3Method

SubjectsA total of 108 females and 132 males participated. One

female and one male experimenter each randomly selectedhalf of the subjects by choosing persons seated in a coffeeshop or general library at Purdue University. The subjects'mean age was 21.61 years.Procedure

Each subject read a brief description (e.g., "an averageman" or "an average woman who is employed full-time")and rated this stimulus person. The descriptions variedaccording to a 2 X 3 (female vs. male X employee vs.homemaker vs. no occupational description) factorial de-sign.


described as an average woman or as an average man.Occupation of stimulus person. The stimulus persons

were described as employed full-time, or as caring for ahome and children and not employed outside of the home,or no occupational description was provided.

Measuring InstrumentsBeliefs about stereotypic attributes. The measures de-

scribed for Experiment 1 were used.Inferred likelihood of employment. For the stimulus

persons for whom no occupational description was pro-vided, subjects indicated on an 11-point scale, rangingfrom 0% chance to 100% chance, the likelihood that theperson was employed full-time.

Inferred job status. For stimulus persons described asemployed full-time, the salary measure described in Ex-periment 1 was used.


Results

The principal data analyses were Sex ofStimulus Person X Occupation of StimulusPerson ANOVAS.

Inferred Likelihood of Employment andInferred Job Status

Subjects who received no occupational in-formation about the stimulus person inferredthat the woman was less likely than the manto be employed full-time (Ms = 56.50% vs.79.75%, respectively), F(l, 78) = 30.39, p <.001. Subjects who rated employees ascribedlower salaries to the woman (M = $15,615)than the man (M= $21,193), F(\, 78) = 13.06,p< .001.


Subjects' mean ratings of communion andagency appear in Table 3. On communion,the significant main effects of sex, F(\, 234) =13.32, p < .001, and occupation, F(2, 234) =49.08, p < .001, should be interpreted in thecontext of a significant Sex X Occupation in-teraction, F(2, 234) = 12.34,p < .001. Theseeffects are best described in terms of theplanned contrasts implied by the hypotheses.As expected, homemakers, regardless of theirsex, were perceived as more communal thanemployees (p < .001 for female stimulus per-sons; p < .005 for male stimulus persons). Inaddition, for stimulus persons without occu-pational descriptions, the traditional genderstereotype of the woman as being more com-munal than the man was obtained (p < .001).The female and male employees were not per-ceived to differ in communion, nor were thefemale and male homemakers. The communaltendency of the woman whose occupation wasnot given was less than that of the femalehomemaker (p < .001) but greater than thatof the employed woman (p < .001). The com-munal tendency of the man whose occupationwas not given was less than that of the malehomemaker (p < .001) or of the employedman (p < .005).

On agency, the significant main effect ofoccupation, F(2, 234) = 21.88, /?<.001,should be interpreted in the context of thesignificant Sex X Occupation interaction, F(2,234) =10.14, p<m\. According to the

Table 3Mean Ratings of Stereotypic Attributes ofFemales and Males Who Varied inOccupation: Experiment 3

Stimulusperson


MaleCommunalAgentic

Employee

3.313.69

3.393.40

Homemaker

4.223.02

4.112.90

Nooccupationaldescription

3.813.00

3.033.46

Note. Means are on a 5-point scale; larger numbers indicategreater communion or agency. All cell ns = 40. For com-munal, MS, = 0.33; for agentic, MSe = 0.31.

planned contrasts, employees (regardless oftheir sex) were perceived as more agentic thanhomemakers (ps < .001 for female and malestimulus persons). In addition, for stimuluspersons without an occupational description,the traditional gender stereotype of the manas more agentic than the woman was obtained(p < .001). The female employee was perceivedas more agentic than the male employee (p <.025), whereas the female and the male home-maker were not perceived to differ. The agentictendency of the woman without an occupa-tional description did not differ from that ofthe female homemaker but was less than thatof the female employee (p < .001). The agentictendency of the man without an occupationaldescription was greater than that of the malehomemaker (p < .001) but not different fromthat of the male employee.

These ANOVA findings are generally consis-tent with the theory that gender stereotypesstem from the observed distribution of womenand men into homemaker and employee roles.Therefore, it is worthwhile to examine thecorrelations between (a) inferred role distri-butions of the stimulus persons who lackedoccupational descriptions and (b) the ascrip-tion of gender-stereotypic attributes to them.Overall, the higher the inferred likelihood thatthe average woman or man was employed, thelower was her or his communion, r(78) = —.34,p < .01, and the higher was her or his agency,r(78) = .41, p < .001. Yet inferred likelihoodof employment should have been a strongerpredictor of Stereotypic perceptions for the


woman than for the man. This differentialpredictability would be expected because like-lihood of employment for the average manwas relatively invariant (SD = 15.61 for av-erage man vs. 21.31 for average woman), F(39,39) = 1.86, p < .05, no doubt because thislikelihood was generally quite high (see theprevious subsection of results). For the woman,greater inferred likelihood of employment wasassociated with slightly less communion,r(38) = -.19, p < .25, and greater agency,r(38) = .43, p < .01. For the man, the cor-relations between inferred likelihood of em-ployment and perceived communion, r(38) =.13, and agency, r(38) = .01, were very smalland nonsignificant.

Disctission

The findings of Experiment 3 were generallyfavorable to the hypothesis that a sex differencein the distribution of women and men intohomemaker and employee roles underlies thestereotype that women are communal and menare agentic. Two findings confirmed prereq-uisite conditions for testing this hypothesis: (a)The average woman and man without occu-pational descriptions were perceived stereo-typically (i.e., the woman was perceived asmore communal and the man as more agentic),and (b) the average man was judged as con-siderably more likely than the average womanto be employed.

The correlational analyses relating likeli-hood of employment to the attributes of theaverage woman and man provided one test ofwhether observations of women's and men'sdiffering occupational roles account for ste-reotypes. The likelihood that the averagewoman was employed related positively to heragency and negatively (albeit weakly) to hercommunion. In other words, subjects who re-ported that women are often employed tendedto view women counterstereotypically, andthose who reported that women are less oftenemployed (and presumably more often home-makers) tended to view women stereotypically.Comparable findings were not obtained formen, probably because they are generally em-ployees and rarely homemakers in the natu-rally occurring situations in which subjectshad made their previous observations.

The ANOVA findings concerning the attri-

butes of the stimulus persons whose occupa-tions were described reveal even more clearlythat social roles underlie gender stereotypes.Subjects believed that, regardless of sex, per-sons described as homemakers differed frompersons described as employees. Female andmale homemakers were perceived to be likestereotypic women (high in communion andlow in agency), whereas female and male em-ployees were perceived to be like stereotypicmen (low in communion and high in agency).Another important aspect of subjects' beliefsabout the stimulus persons with occupationaldescriptions is that the females and males wereperceived relatively equivalently once their so-cial role as employee or homemaker was spec-ified. The only exception to this pattern, andan exception that demands explanation, is thesignificantly greater perceived agency of em-ployed women compared with employed men.

The comparisons between same-sex personswho differed in occupation were also moder-ately consistent with our role-distribution the-ory of gender stereotypes. Because subjectsjudged that there was roughly a 50-50 chancethat the average woman was employed, per-ceptions of her personal attributes should havefallen between those of the employed womanand the female homemaker. Although this ex-pectation was fulfilled (see Table 3), the averagewoman's perceived agency did not differ sig-nificantly from that of the homemaker. Be-cause subjects considered it quite likely thatthe average man was employed, their beliefsabout his personal attributes should have beensimilar to their beliefs about the employedman. This expectation was fulfilled foragency, but the average man was perceived assignificantly less communal than the em-ployed man.6

6 In the interpretation of these findings, subjects' judg-ments of the probability that the average woman and manare employed should not be treated as exact estimates ofthe observed distributions of people into employee andhomemaker roles. One reason for our caution is that maleswho are not employed have probably been observed inlargely different roles than females who are not employed.Whereas such females are usually homemakers, probablysuch males are retired or seeking employment. Therefore,the subjects' judgment that there is an 80% chance thatthe average man is employed does not imply that 20% ofmen have been observed as homemakers. It also does not


Although the findings of this experimentare generally supportive of the idea that per-ceived sex differences stem from perceivers'observations of women and men in differingoccupational roles, they do not explain thetendency for employed women to be perceivedas more agentic than employed men.7 It isnoteworthy that this finding was obtained inExperiments 1 and 2 for the stimulus personswithout job titles as well as in Experiment 3for the average employed persons because thesestimulus persons were probably thought tohave lower status positions if they were female.As demonstrated in Experiments 1 and 2,lower status employees were believed to be lessagentic than higher status employees; there-fore, any tendency to ascribe lower status towomen would tend to counteract any factorsmaking them appear more agentic than men.

Both of the explanations that we introducedearlier for this sex difference in agentic qualitieswere discounted by the findings from Exper-iment 3. The idea that subjects would be un-willing to derogate women on stereotype ques-tionnaires was discounted by their perceptionof the average woman as significantly lessagentic than the average man. The other ex-planation of female employees' greater agencywas that categories of people (e.g., women)who are relatively uncommon in a social role(e.g., employee) are believed to be more strin-gently selected in terms of the requirementsof the role than people who commonly occupythe role. This explanation was discounted bythe finding that the male homemaker, who isconsiderably rarer than the female employee,was not perceived to differ from the femalehomemaker in communion, the quality thatsubjects believed typical of homemakers.8

Yet another explanation of employed wom-en's higher agency is that this perceived genderdifference reflects a semantic or response-lan-guage difference in the way males and femalesare judged (e.g., Manis, 1971;Upshaw, 1969)

imply that the average man's personality attributes wouldbe perceived as between those ascribed to employees andthose ascribed to homemakers. Perhaps the low level ofcommunion ascribed to the average man reflects the un-favorably evaluated roles ascribed to the 20% of men notthought to be employed (e.g., being retired, seeking em-ployment).

rather than a difference in how they are per-ceived. If stimulus persons are implicitly com-pared with same-sex reference groups, this se-mantic interpretation suggests that femaleemployees are implicitly compared with otherfemales and therefore are judged as very agen-tic, whereas male employees are implicitlycompared with other males and therefore arejudged as not especially agentic. This inter-pretation also suggests that male homemakersare implicitly compared with other males andtherefore are judged as very communal,whereas female homemakers are implicitlycompared with other females and thereforeare not judged as especially communal. Al-though female employees were perceived asmore agentic than male employees, malehomemakers were not perceived as more com-munal than female homemakers. The fact thatfemale and male homemakers were perceivedequivalently, then, does not support the ex-planation that the perceived sex differencesare an artifact of differences in the referencegroup implicitly used by subjects when ratingfemales and males. Yet so little is known abouthow people use implicit comparison standardsthat we cannot be completely certain that sucha process has no relevance to our findings.

Because our experiments so consistently re-vealed that female employees were perceivedas more agentic than their male counterpartsand because this finding is counterstereotypicand perhaps important, we considered anotherexplanation of the finding. Accordingly, thefact that employed women often balance

'Other investigators have reported higher perceivedagency and competence in females than males, yet in thesestudies the situation faced by the stimulus person wasjudged to be especially difficult or demanding for women(Abramson, Goldberg, Greenberg, & Abramson, 1977;Taynor & Deaux, 1973), or the stimulus materials mayhave implied employment or special competence (Deaux& Lewis, in press). Also, in Experiment 3 the communalratings of employed women and men did not differ (al-though they did differ significantly in Experiment 2 andmarginally in Experiment 1). Therefore, only the sex dif-ference in employees' agency appears reliable and con-sequently requires explanation.

8 Yet it could be that this selection hypothesis pertainsto the employee role (because perceivers know that peopleare screened for employment) but not to the homemakerrole (because perceivers are uncertain about whether peopleare screened for the occupation of homemaker).


two demanding roles—homemaker and em-ployee—may account for the relatively highagency ratings of such women. Perceivers mayhave observed these agentic qualities amongthe women who experience this potential roleoverload and role conflict.

To enable us to test this "double-burden"explanation, subjects in a fourth experimentrated the personality attributes of an employedwoman or man described as either married orsingle and as either having or not having chil-dren at home. Other subjects rated an em-ployed woman or man whose marital and pa-rental statuses were hot described. Should thedouble-burden explanation of women's greateragency be correct, employed women withfamily responsibilities, especially involvingchildren, would be regarded as particularlyagentic. Family responsibilities would increasethe agency ascribed to employed men only ifsuch men did not have wives who typicallywould carry out the household duties. Thesingle father would fulfill this criterion and,as a consequence, might also be perceived asespecially agentic.

Experiment 4

Method

Subjects

A total of 108 females and 135 males participated ingroups of about 15 in a laboratory setting to fulfill a psy-chology course requirement at Purdue University. Onefemale experimenter administered all stimulus materials.Subjects' mean age was 19.33 years.

Procedure

Each subject read a brief description of a full-time em-ployee (e.g., "an average man who is employed full-time,is married, and has no children") and rated this stimulusperson. The descriptions varied according to a 2 X 2 X2 (employed female vs. employed male X married vs. notmarried X children at home vs. no children at home)factorial design. Two control conditions were provided: afemale for whom neither marital nor parental status wasprovided and a male for whom neither status was provided.

Manipulation of Independent Variables

Sex of stimulus person. The stimulus persons weredescribed as an average employed woman or an averageemployed man.

Marital status of stimulus person. The stimulus personswhose marital status was provided were described as mar-ried or as not married. For the remaining stimulus persons,no information about marital status was provided.

Parental status of stimulus person. The stimulus per-sons whose parental status was provided were describedas having children at home or not having children. (Theunmarried woman or man with children at home wasfurther described as a "single parent," to aid subjects ininterpreting this potentially ambiguous description.) Forthe remaining stimulus persons, no information aboutchildren was provided.

Measuring Instruments

Subjects' beliefs about stereotypic attributes and jobstatus were assessed using measures described in Exper-iment 1. The subjects also estimated the number of hoursthat the stimulus person worked per day doing householdand family-related work. For stimulus persons whose mar-ital and parental statuses were not provided, subjects Usedan 11-point rating scale, ranging from 0% to 100% chance,to rate the likelihood that the person was married. On asecond, similar scale, the subjects rated the likelihood thatthe person had children at home.

Results and Discussion

The principal data analyses were Sex ofStimulus Person X Marital Status of StimulusPerson X Parental Status of Stimulus PersonANOVAS, with the error term including the twocontrol conditions, which omitted the infor-mation about marital and parental statuses.

Inferences About Marriage, Children,Household Work, and Job Status

For employees whose marital and parentalstatuses were omitted, the woman was judgedless likely to be married (M = 54.17%) thanthe man (M = 76.00%), F(\, 47) = 20.50, p <.001. She was also judged less likely to havechildren at home than her male counterpart(Ms = 43.75% versus 66.00%, respectively),F(l, 47) = 16.22, p< .001. Thus subjects didnot consider it highly likely that the averageemployed woman faced a double burden offamily and employment responsibilities.

Analyses of subjects' estimates of the num-ber of hours per day the employed personsspent doing household and family-related workrevealed only one significant effect: Personswith children were judged to do more workthan persons without children (Ms = 4.91 ver-sus 3.34, respectively), F(l, 233) = 6.96, p <.01. Subjects failed to acknowledge that mar-ried female employees spend more time onhousehold work than married male employees,even though this sex difference has been welldocumented in sociological literature onhousework (e.g., Hartmann, 1981).


Analyses of subjects' salary estimates rep-licated the effect of sex obtained in our pre-vious experiments: Women were judged tohave lower salaries than men (Ms = $ 18,074versus $25,055, respectively), F(l, 233) =38.29, p<. 001.


Subjects' mean ratings of communion andagency appear in Table 4. On communion,the main effects of marital and parental sta-tuses were both significant. Married employeeswere judged as being more communal thansingle employees (Ms = 3.54 versus 3.26, re-spectively), F(l, 233) = 9.60, p < .01. Em-ployees with children at home were judged asmore communal than employees without chil-dren at home (Ms = 3.68 versus 3.11, respec-tively), F(\, 233) = 41.75, p < .001.9

On agency, only the sex effect proved sig-nificant: Replicating the three previous ex-periments, female employees were perceivedas more agentic than male employees (Ms =3.75 versus 3.59, respectively), F(\, 233) =7.27, p < .01, calculated on the basis of all10 conditions.10 Thus, contrary to the double-burden hypothesis, neither marital nor paren-tal status influenced the agency ascribed toemployed women or men.

In view of this nonconfirmation of the dou-ble-burden explanation of women's perceivedagency and in view of the continued reliabilityof the finding itself, we considered yet anotherexplanation: Perhaps employed women areconsidered more agentic than their malecounterparts because they are more likely tohave chosen to be employed. Because agencyevidently typifies employees (perhaps becausemost jobs require and reward agentic behav-ior), observing that women often are employedby choice may have led perceivers to infer thatsuch women are agentic. This explanation ofthe effects of choice on stereotype formationfollows from correspondent inference theory(Jones & Davis, 1965), which suggests thatperceivers more often infer that an employeepossesses the personal qualities she or he man-ifests behaviorally if it appears that the em-ployee has freely chosen to work and has notbeen required to work by virtue of her or hislife situation or sex role.

In a test of this freedom-of-choice expla-

nation, subjects in a fifth experiment rated thepersonal attributes of a female or male em-ployee described as employed by choice oremployed out of necessity. Other subjects rateda female or male employee about whom nochoice information was provided.

Experiment 5Method

SubjectsA total of 110 females and 132 males participated. One

female and one male experimenter each randomly selectedhalf of the subjects by choosing persons seated in a coffeeshop or general library at Purdue University. Subjects'mean age was 20.59 years.

Procedure

Each subject read a brief description of a full-time em-ployee [e.g., "an average woman who is employed full-time and who is employed because she wants to work (andnot because she has to work)"] and rated this stimulusperson. The descriptions varied according to a 2 X 3 (femalevs. male X employed by choice vs. employed out of ne-cessity vs. no choice information) factorial design.


described as an average employed woman or an averageemployed man.

Choice of stimulus person to be employed. The stimuluspersons were described as employed because they want towork and not because they have to work or employedbecause they have to work and not because they want towork, or no information about employment choice wasprovided.

Measuring InstrumentsSubjects' beliefs about Stereotypic attributes and job

status were assessed with the measures described in Ex-

9 Consistent with a Sex of Stimulus Person X ParentalStatus of Stimulus Person interaction, F(\, 233) = 5.86,p < .05, male employees with children at home were per-ceived as considerably more communal than male em-ployees without children at home (Ms - 3.76 versus 2.97,respectively, p < .001), whereas female employees withchildren at home were not perceived as so greatly different(M = 3.59) from female employees without children athome (M - 3.25), although this difference remained sig-nificant for the females (p < .01).

10 For all five role descriptions, the female agency meanwas larger than the male mean. Yet this agency sex dif-ference became quite small for the stimulus persons forwhom marital and parental statuses were not described.However, in this female stimulus-person condition, therewere two unusually low agency scores.


Table 4Mean Ratings of Stereotypic Attributes of Female and Male Employees Who Varied in Marital andParental Statuses: Experiment 4

Stimulusperson


MaleCommunalAgentic

Married/children

3.773.70

3.863.59

Married/nochildren

3.383.83

3.103.54

Single/children

3.423.91

3.663.75

Single/nochildren

3.113.84

2.873.60

Nodescription

3.303.49

3.403.45

Note. Means are on a 5-point scale; larger numbers indicate greater communion or agency. Cell ns ranged from 22 to27. For communal, A/5e = 0.37; for agentic, MS, = 0.24.

periment 1. For the stimulus persons about whom no choiceinformation was provided, subjects used an 11-point scale,ranging from 0% to 100% chance, to rate the likelihoodthat the person worked "because she [he] has to and notbecause she [he] wants to."

Results

The principal data analyses were Sex ofStimulus Person X Choice of Stimulus Personto Be Employed ANOVAS.

Inferred Likelihood of Employment Out ofNecessity and Inferred Job Status

The average employed woman was judgedless likely to be employed out of necessity thanthe average employed man (Ms = 48.75% ver-sus 67.33%, respectively), F(l, 78) = 11.50,p< .01. Also, women were judged to havelower salaries than men (Ms = $16,330 versus$23,983, respectively), F(l, 235) = 55.81,;? <.001. The main effect of the stimulus person'schoice to be employed, F(2,235) = 17.78, p <.001, revealed that higher salaries were ascribedto employees working by choice (M =$23,265) or to employees without choice in-formation (M - $2-1,304) than to employeesworking out of necessity (M=$ 15,977,ps < .001).


Subjects' mean ratings of communion andagency appear in Table 5. On communion,only the main effect of choice proved signif-icant, F(2, 236) = 7.41, p < .001: Greatercommunion was ascribed to employees work-ing by choice (M = 3.43) or employees withoutchoice information (M = 3.35) than to em-

ployees working out of necessity (M = 3.10,ps < .01 or smaller).

On agency, the main effects of sex and choicewere both significant, as was the interactionbetween these variables. Women were per-ceived as more agentic than men (Ms = 3.54vs. 3.30, respectively), J^l, 236) = 12.34, p <.001. Consistent with the choice main effect,F(2t 236) = 89.54, p < .001, employees work-ing by choice were perceived as more agenticthan employees without choice information(Ms = 3.86 versus 3.59, p < .002, respectively),who in turn were more agentic than employeesworking out of necessity (M = 2.80, p < .001).Consistent with the Sex X Choice interaction,F(2, 236) = 3.46, p < .05, the tendency forwomen to be perceived as more agentic thanmen was significant for employees withoutchoice information (p < .001) and employeesworking out of necessity (p < .025) and non-significant for employees working by choice.Note also that for women, the agency of theemployee working by choice did not differfrom that of the employee without choice in-formation, and these two employees were moreagentic than the employee working out of ne-cessity (ps < .001). For men, the agency ofthe employee working by choice was greaterthan that of the employee with no choice in-formation (p < .001), and these two employeeswere more agentic than the employee workingout of necessity (ps < .001).

These ANOVA findings generally support thehypothesis that employed women are perceivedas more agentic than employed men, becausewomen have been observed to exercise greaterchoice about being employed. Thus it isworthwhile to examine, for employees about


Table 5Mean Ratings of Stereotypic Attributes of Femaleand Male Employees Who Varied in Choice toBe Employed: Experiment 5

Stimulusperson


MaleCommunalAgentic

Employedby choice

3.383.86

3.453.85

Employedout of

necessity

3.232.95

2.982.66

No choiceinformation

3.413.80

3.273.38

Note. Means are on a 5-point scale; larger numbers indicategreater communion or agency. Cell ns ranged from 40 to41. For communal, MS, = 0.31; for agentic, MS, = 0.27.

whom no choice information was provided,the correlation between freedom of choice andagency. Overall, the more likely an employeewas to be regarded as working out of necessityrather than choice, the lower was her or hisagency, r(78) = -.40, p < .001. However, thisrelation was considerably stronger for the fe-male employee, r(38) = -.55, p < .001, thanthe male employee, r(38) = -.09, ns.

Discussion

In Experiment 5, the average employedwoman was once again rated as more agenticthan the average employed man, and the find-ings of this experiment favored the hypothesisthat this sex difference stems from perceivers'observations that women are more likely thanmen to have chosen to be employed. One ofthe findings favoring this hypothesis is that theaverage employed woman was judged as morelikely than the average employed man to beworking by choice rather than necessity. Fur-thermore, on a correlational basis, the lesslikely subjects believed it was that the averageemployed woman works out of necessity, themore agency they ascribed to her.

Stronger evidence that sex differences inchoice to be employed underlie the greaterperceived instrumentality of female (vs. male)employees is provided by subjects' beliefsabout the agency of employees described aseither having or lacking freedom of choice.The choice information strongly affectedagency ratings in the expected direction: Re-

gardless of their sex, employees working bychoice were perceived as more agentic thanemployees working out of necessity. Further-more, perceived agency was similar for thewomen and men about whom choice infor-mation was provided. When employed bychoice, the woman and the man did not differin agency, although when employed out of ne-cessity, the woman was perceived as somewhatmore agentic than the man."

In addition, it should be noted that thecomparisons between stimulus persons of thesame sex are relatively consistent with ourfreedom-of-choice hypothesis. Because sub-jects believed that it was moderately likely thatthe average employed woman had freedom ofchoice, her agency should have been perceivedas similar to that of the woman employed bychoice. As expected, the agency of the averageemployed woman did not differ from that ofthe woman employed by choice, and it wasgreater than that of the woman employed outof necessity. Because subjects believed that itwas moderately unlikely that the average em-ployed man had freedom of choice, his agencyshould have been perceived as lower than thatof the man employed by choice. As expected,the agency of the average employed man waslower than that of the man employed by choicealthough higher than that of the man employedout of necessity.

Even though the communion findings arenot relevant to our hypothesis, it should benoted that lack of choice also lowered thestimulus persons' communal tendency. Thefact that the effect of choice on communionwas weaker than its effect on agency may ex-plain why women's greater freedom of choicedid not cause female employees to be perceivedas more communal than male employees, aswell as more agentic. Consistent with the weak

"The meaning of these comparisons is necessarilysomewhat ambiguous because the choice information mayhave had different implications for males than for females.For example, a man employed by choice may be perceivedas independently wealthy, whereas a woman employed bychoice may be perceived as married and merely in com-fortable financial circumstances. Similarly, a woman em-ployed out of necessity may be perceived as single or inespecially poor financial circumstances, whereas a manemployed out of necessity may be perceived as fairly typicalof all men (as suggested by the findings presented above).


effect of choice on communion, in this seriesof experiments, nonsignificant tendencies inthe female direction were found in the majorityof communion comparisons, and in Experi-ment 2, which had a large number of subjects,the comparison reached significance.

As we noted earlier, the relation betweenfreedom of choice and agency may be attri-butionally mediated. Because jobs are thoughtto require agentic behavior, observations thatmembers of a particular group are generallyemployed by choice may favor the correspon-dent inference that such people possess agenticpersonality attributes. It is also possible thatemployed women's greater freedom of choiceimplies that they are more qualified becauseof selection or self-selection in terms of agenticqualities. That is, mainly those women whoare agentic have chosen to worjt or have beenspecially selected to work (see also Footnote8). An additional explanation stems from thefinding that activities perceived to be voluntaryrather than required are associated with morepositive affect and higher involvement (Csik-szentmihalyi & Figurski, 1982). Therefore, thehigher ratings of the agency (and communion)of employees who freely chose to work mightbe a consequence of the positive affect thathas been observed to characterize this majorportion of such individuals' lives.

General Discussion

According to our framework, social struc-ture accounts for the content of stereotypes,or more exactly, the observed distribution ofgroups into various aspects of social structureunderlies stereotypes. Although such an ex-planation may account for stereotypes aboutother subgroups within societies (e.g., races),we have confined our investigation to genderstereotypes by hypothesizing that the observeddistribution of women and men into socialroles underlies these stereotypes. In particular,our experiments investigated whether per-ceivers' beliefs about women and men stemfrom their previous observations of womenand men in differing statuses within workhierarchies and in differing occupational roles.Because the content of gender stereotypesarises from perceivers' observations of people'sactivities and these activities are determinedprimarily by social roles, gender stereotypes

(operationally defined as beliefs that certainattributes differentiate women and men; seeAshmore & Del Boca, 1981) arise whenwomen and men are observed typically tocarry out different social roles.

Any theory of the content of gender stereo-types should account for the major perceivedgender differences documented in past re-search—namely, the agentic qualities ascribedto men and the communal qualities ascribedto women. These differences should be ac-counted for by one or both of the major dif-ferences in the way women and men are dis-tributed into social roles—namely, the con-centration of women in lower positions inhierarchies of status and authority, and in thehomemaker rather than the employee occu-pational role. In our research, only the home-maker-employee difference appeared to ac-count for the subjects' beliefs that women areespecially communal and men especiallyagentic. Although status differences did notaccount for these beliefs, it would be surprising,in view of the importance that sociologists haveaccorded to hierarchy as a fundamental aspectof social roles (e.g., Blau, 1964; Weber, 1947),if observed status differences did not underlieany beliefs about gender-differentiating traitsand behaviors. On the contrary, status differ-ences have been shown to account for the beliefthat women are more compliant than men(Eagly & Wood, 1982).12

In documenting that occupational roles un-derlie belief in female communal qualities andmale agentic qualities, our research has shownthat beliefs about women resemble those abouthomemakers and that beliefs about men re-semble those about employed persons. In ad-dition, our research has shown that beliefsabout what is typical of homemakers and em-ployees override beliefs about what is typicalof women and men, whereas the converse doesnot occur. Beliefs about what is typical of

12 Indeed, still other aspects of the distribution of menand women in society may affect beliefs about the sexes.For example, Kiesler (1975) suggested that because menoutnumber women as successful achievers in many oc-cupations, perceivers tend to evaluate an individual wom-an's achievements less favorably than an individual man's,even when their products are objectively equal. Perceivers'judgments of an individual's success may thus reflect theprobability of success for persons of the individual's gender.


women and men do not override beliefs aboutwhat is typical of homemakers and employ-ees.13 Instead, except for the greater agency offemale employees, women and men were per-ceived equivalently once their occupationalrole as homemaker or employee was specified.The underlying reason that occupational roleis more important than sex in determiningbeliefs about communal and agentic charac-teristics is that stereotypes concerning thesequalities have become associated with the sexesmainly because sex has been observed to co-vary with occupational role.

Although the finding that female employeesare perceived as more agentic than their malecounterparts was initially serendipitous, it iscompatible with our social-structural per-spective if it is interpreted in terms of perceivedchoice. The perception that employed womenare likely to be employed by choice arises notonly from perceivers' observations that women(and not men) are homemakers but also fromtheir belief that women's primary obligationis to the domestic role, a belief documentedby public opinion research (e.g., Mason, 1973;Mason, Czajka, & Arber, 1976). Because tra-ditionally the role of homemaker did not in-clude any obligation to seek employment out-side the home, employment tends to be re-garded as an optional or freely chosen aspectof married women's lives. Therefore, even theperception of employed women as more agen-tic than employed men reflects previous ob-servations of women's and men's different dis-tributions into (and obligations in relation to)domestic and paid employment roles.

Relation to Research on CognitiveBases of Stereotyping

Our research has implications for severalaspects of stereotyping that have been ex-amined recently in the research literature onsocial cognition. One highly relevant analysishas treated sex stereotypes as prior probabil-ities in a Bayesian model (Locksley, Borgida,Brekke, & Hepburn, 1980; Locksley, Hepburn,& Ortiz, 1982). Locksley and her colleaguesfound perceived sex differences in a target per-son's assertiveness only when subjects did nothave diagnostic, individuating informationabout the target's previous levels of assertive-

ness. This information caused the prior prob-abilities of women's and men's assertivenessto be revised. Similarly, in our research, evenextremely general information about a person'semployment status caused subjects to revisetheir estimates of women's and men's com-munal and agentic qualities.

Our findings, however, highlight the incom-pleteness of the Bayesian analysis, which doesnot explain why stereotypes have certain con-tent, even though it does model a process bywhich past behavior or other characteristicsthat are believed to predict future behavioroverride information about sex. To accountparsimoniously for the content of gender ste-reotypes, an investigator must find a behavioralor personal attribute that is diagnostic of theparticular set of attributes believed to char-acterize womeji and men—namely, the strongcommunal and weak agentic tendencies as-cribed to women and the weak communal andstrong agentic tendencies ascribed to men. Toexplain these beliefs, an attribute must (a) dif-ferentiate the sexes and (b) relate to perceivedagency and communion in opposite directions.As we have shown, these criteria are fulfilledby the occupational roles of homemaker andemployee. Yet we have shown that one's po-sition in hierarchies of status and authoritydoes not meet these requirements.

Although most social cognition research haslittle power to account for the content of ste-reotypes, research on "illusory correlation"(e.g., Hamilton, 1981; Hamilton & Gifford,1976) pertains to content. It has shown thatminority group members may be perceived tohave characteristics with low probability ofoccurrence merely because the group membersand such characteristics are both rare andtherefore distinctive. In our research womenconstitute minorities in some social roles (e.g.,as high-status employees), as do men in othersocial roles (e.g., as homemakers). Neverthe-

13 Deaux and Lewis (in press) have also reported thatrole information is more important than sex in determiningbeliefs about women and men, although their role de-scriptions included more than occupational cues (e.g., themasculine role was described as "head of the household,financial provider, a leader, and responsible for householdrepairs").


less, despite their rarity, such women and menappear to be assigned primarily the attributesthat correspond to the roles they occupy.

Our research also relates to the issue ofwhether there is a basic level of categorizationthat people commonly use in representingother people and ascribing attributes to them(Brewer, Dull, & Lai, 1981; Cantor & Mischel,1979; Taylor, 1981). In particular, categori-zations at this basic level are hypothesized tomaximize the richness, differentiation, andvividness of subjects' perceptions of people.Using such criteria, Cantor and Mischel (1979)concluded that categorization at the level ofsocial roles or "persona" is basic in personperception. It is notable that the research re-ported in the present article and our earlierresearch on beliefs about social influence(Eagly & Wood, J 982) point to the importanceof categorizations at the level of social roles(i.e., occupation and hierarchical status), whichare highly diagnostic of people's traits andabilities. One's sex can be considered less basicthan roles because it functions as a cue that,due to its previous association with social roles,provides indirect access to occupation and hi-erarchical status.

Finally, our framework is not inconsistentwith the idea popular in the social cognitionliterature that stereotypes function as proto-types (e.g., Cantor & Mischel, 1979) or sche-mata (e.g., Taylor & Crocker, 1981). Our claimthat observations of social roles underlie genderstereotypes is not meant to imply that thestimulus person's social role is ordinarily re-trieved for perceivers to infer her or his attri-butes. Instead, perceivers' prototype or schemaof a typical woman or man is retrieved, andjudgments are made in terms of the personalattributes already associated with it. Becausepeople's activities are determined primarily bytheir social roles, the prototypes or schematathat perceivers possess of woman and manconsist largely of attributes that covary withthe role assignments of women and men.Therefore, our analysis primarily elaboratesthe process by which stereotypes have acquiredparticular content rather than the processesthat occur when stereotypes are used to makejudgments. Yet our analysis includes the im-portant point that the use of gender stereotypesis under the control of cues that carry role

information because of the origin of these ste-reotypes in observed variation in role assign-ments. Gender stereotypes are not appliedwhen typical role arrangements are invalidatedby these cues.

Conclusion

To turn from issues of cognitive mediationtoward more general issues addressed in thestereotyping literature, we note that our anal-ysis bears on the classic "kernel-of-truth"question pertaining to the validity of stereo-types. In assuming that stereotypes validlyrepresent the social structure and division oflabor in a society, our approach is consistentwith that of other investigators who have im-plicated these factors as the kernel of truth ofethnic stereotypes (Brewer & Campbell, 1976)and gender stereotypes (Williams & Best,1982). As such, our approach deviates frompsychologists' more typical assumptions thatstereotypes stem from psychological factorssuch as perceivers' distortions (e.g., Katz &Braly, 1933; Lippmann, 1922) and biases in-herent in perceivers' cognitive processing (e.g.,Hamilton, 1979). However, the kernel-of-truthquestion is considerably broader than our so-cial structural hypotheses, and as Brigham(1971), Campbell (1967), and others havenoted, there is ample reason to believe thatstereotypes both represent and distort reality.Our research, by relating stereotypes to per-ceived distributions into social roles, providesa promising avenue for addressing the kernel-of-truth question.

This kernel-of-truth issue raises the questionof whether the various attributes associatedwith social roles reflect ingrained personalitytraits and .abilities characteristic of the typicaloccupants of the roles. In relation to home-maker and employee roles, for example, onemay wonder whether girls are socialized to ac-quire communal traits and boys to acquireagentic traits, with the result that women asa group are more suited to perceivers' conceptof the homemaker role and men as a groupto perceivers' concept of the employee role.Although we believe that socializing agentstend to prepare girls and boys for the socialroles that they believe these girls and boys willprobably occupy as adults (Eagly, 1983), our


theory of stereotypes does' not address theseissues of how people are prepared for rolesand recruited into them. According to ourframework, the proximal cause of gender ste-reotypes is the differing distributions of womenand men into social roles, whatever the cau-sation that lies behind these differing distri-butions.

We also note that our findings have someimplications for the issue of gender equality.The gender or "sex role" stereotypes that psy-chologists have made famous and enshrinedin their theories about gender and the methodsof gender research appear to be eradicated andeven partly reversed by information as generaland basic as the fact that the persons beingjudged are employed. A literal interpretationof our findings suggests that employed women,who constitute about half of American women(U.S. Department of Labor, 1980), are prob-ably not ordinarily perceived in terms of thefemale stereotype of low agency and highcommunion, because it is usually quite salientto an employed woman's friends, associates,and family members that she is employed. Yetit is no doubt incorrect to think that employedwomen escape the traditional female stereo-type altogether, because we do not fully un-derstand the conditions under which peopleapply occupational stereotypes in preferenceto gender stereotypes. It is very likely that thereare contexts in which an employed woman isperceived traditionally, even by people whoknow that she is employed.

It is also important to note that the stereo-types we obtained of employees do not portraywomen and men equally. Instead, these ste-reotypes favor the formerly excluded group:Employed women are perceived as more agen-tic than employed men (even though womenare believed to have lower wages). Yet it shouldbe kept in mind that the source of the extraagency appears to be the belief that womenhave freely chosen to be employed. Free choiceis, of course, absent for many women, a factacknowledged by our respondents in their es-timates of the likelihood that employed womenhave freedom of choice. Enhanced agencywould be ascribed mainly to relatively affluent,married women, because they could mostplausibly be regarded as having freely chosento enter the paid work force.

Another implication of our idea that socialstructure underlies beliefs about gender is thatchange in these beliefs probably must awaitsocial change. Our theory and findings suggestthat gender stereotypes—the beliefs thatwomen in general differ from men in general—will not disappear until people divide socialroles equally, that is, until child care andhousehold responsibilities are shared equallyby women and men and the responsibility tobe employed outside the home is borne equally.Interventions designed to affect ideas aboutgender through education and exposure to themedia (e.g., ensuring that textbooks have non-sexist portrayals of women and men) would,of course, have some impact in terms of ourtheory because beliefs about the distributionof people into social roles derive from indirectsources such as textbook portrayals as well asfrom direct experience. Yet daily life providesabundant direct experience with women andmen. Therefore, efforts to remove gender ste-reotypes educationally may have relatively littleimpact, compared with actual changes in thedistribution of the sexes into social roles.

References

Abramson, P. R., Goldberg, P. A., Greenberg, J. H., &Abramson, L. M. (1977). The talking platypus phe-nomenon: Competency ratings as a function of sex andprofessional status. Psychology of Women Quarterly, 2,114-124.

Ashmore, R. D., & Del Boca, F. K. (1981). Conceptualapproaches to stereotypes and stereotyping. In D. Ham-ilton (Ed.), Cognitive processes in stereotyping and in-tergroup behavior (pp. 1-35). Hillsdale, NJ: Erlbaum.

Bakan, D. (1966). The duality of human existence: Anessay on psychology and religion. Chicago: RandMcNally.

Bern, S. L. (1974). The measurement of psychologicalandrogyny. Journal of Consulting and Clinical Psy-chology, 42, 155-162.

Bishop, Y. M. M., Fienberg, S. E., & Holland, P. W. (1975).Discrete multivariate analysis: Theory and practice.Cambridge, MA: MIT Press.

Blau, P. (1964). Exchange and power in social life. NewYork: Wiley.

Block, J. H. (1973). Conceptions of sex roles: Some cross-cultural and longitudinal perspectives. American Psy-chologist, 28, 512-526.

Blood, R. O., Jr., & Wolfe, D. M. (1960). Husbands andwives: The dynamics of married living. Glencoe, IL:Free Press.

Brewer, M. B. (1979). In-group bias in the minimal in-tergroup situation: A cognitive-motivational analysis.Psychological Bulletin, 86, 307-324.


Brewer, M. B., & Campbell, D. T. (1976). Ethnocentrismand intergroup attitudes: East African evidence. NewYork: Sage.

Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions ofthe elderly: Stereotypes as prototypes. Journal of Per-sonality and Social Psychology, 41, 656-670.

Brigham, J. (1971). Ethnic stereotypes. Psychological Bul-letin, 76, 15-38.

Broverman, I. K., Vogel, S. R., Broverman, D. M., Clark-son, F. E., & Rosenkrantz, P. S. (1972). Sex-role ste-reotypes: A current appraisal. Journal of Social issues,28, 59-78.

Brown, L. K. (1979). Women and business management.Signs, 5, 266-288.

Campbell, D. T. (1967). Stereotypes and the perceptionof group differences. American Psychologist, 22, 817-829.

Cantor, K, & Mischel, W. (1979). Prototypes in personperception. Advances in Experimental Social Psychology,12, 3-52.

Clifton, A. K., McGrath, D., & Wick, B. (1976). Stereo-types of women: A single category? Sex Roles, 2, 135-148.

Cohen, J. (1977). Statistical power analysis for the be-havioral sciences (2nd ed.). New \brk: Academic Press.

Cronbach, L. J. (1951). Coefficient alpha and the internalstructure of tests. Psychometrika, 16, 297-334.

Csikszentmihalyi, M., & Figurski, T. J. (1982). Self-aware-ness and aversive experience in everyday life. Journalof Personality, 50, 15-28.

Davis, J. A. (1974). Hierarchical models for significancetests in multivariate contingency tables: An exegesis ofGoodman's recent papers. In H. L. Costner (Ed.), So-ciological methodology, 1973-74 (pp. 189-231). SanFrancisco: Jossey Bass.

Deaux, K., & Lewis, L. L. (in press). The structure ofgender stereotypes: Interrelationships among compo-nents and gender label. Journal of Personality and SocialPsychology.

Eagly, A. H. (1983). Gender and social influence: A socialpsychological analysis. American Psychologist, 38, 971-981.

Eagly, A. H., & Wood, W. (1982). Inferred sex differencesin status as a determinant of gender stereotypes aboutsocial influence. Journal of Personality and Social Psy-chology, 43, 915-928.

England, P. (1979). Women and occupational prestige: Acase of vacuous sex equality. Signs, 5, 252-265.

Feldman, J. M. (1972). Stimulus characteristics and subjectprejudice as determinants of stereotype attribution.Journal of Personality and Social Psychology, 21, 333-340.

Gillespie, D. L. (1971). Who has the power? The maritalstruggle. Journal of Marriage and the Family, 33, 445-458.

Hamilton, D. L. (1979). Acognitive-attributional analysisof stereotyping. Advances in Experimental Social Psy-chology, 12, 53-84.

Hamilton, D. L. (1981). Illusory correlation as a basis forstereotyping. In D. Hamilton (Ed.), Cognitive processesin stereotyping and intergroup behavior (pp. 115-144).Hillsdale, NJ: Erlbaum.

Hamilton, D. L., & Gifford, R. K. (1976). Illusory cor-relation in interpersonal perception: A cognitive basisof stereotypic judgments. Journal of Experimental SocialPsychology, 12, 392-407.

Hartmann, H. I. (1981). The family as the locus of gender,class, and political struggle: The example of housework.Signs, 6, 366-394.

Jones, E. E., & Davis, K. E. (1965). From acts to dispo-sitions: The attribution process in person perception.Advances in Experimental Social Psychology, 2, 219-266.

Kanter, R. M. (1977). Men and women of the corporation.New York: Basic Books.

Katz, D., & Braly, K. W. (1933). Racial stereotypes ofone hundred college students. Journal of Abnormal andSocial Psychology, 28, 280-290.

Kiesler, S. B. (1975). Actuarial prejudice toward womenand its implications. Journal of Applied Social Psy-chology, 5, 201-216.

Lippmann, W. (1922). Public opinion. New \ork: Harcourt,Brace.

Locksley, A., Borgida, E., Brekke, N., & Hepburn, C.(1980). Sex stereotypes and social judgment. Journalof Personality and Social Psychology, 39, 821-831.

Locksley, A., Hepburn, C., & Ortiz, V. (1982). Social ste-reotypes and judgments of individuals: An instance ofthe base-rate fallacy. Journal of Experimental SocialPsychology, 18, 23-42.

Manis, M. (1971). Context effects in communication: De-terminants of verbal output and referential decoding.In M. H. Appley (Ed.), Adaptation-level theory (pp. 237-255). New York: Academic Press.

Mason, K. Q (1973). Studying change in sex-role defi-nitions via attitude data. Proceedings of American Sta-tistical Association, Social Statistics Section, 138-141.

Mason, K. Q, Czajka, J. L., & Arber, S. (1976). Changein U.S. women's sex-role attitudes, 1964-1974. AmericanSociological Review, 41, 573-596.

Mennerick, L. A. (1975). Organizational structuring ofsex roles in a non-stereotyped industry. AdministrativeScience Quarterly, 20, 570-586.

Park, B., & Rothbart, M. (1982). Perception of out-grouphomogeneity and levels of social categorization: Memoryfor the subordinate attributes in in-group and out-groupmembers. Journal of Personality and Social Psychology,42, 1051-1068.

Parsons, T. (1955). Family structure and socialization ofthe child. In T. Parsons & R. F. Bales (Eds.), Family:Socialization and interaction process (pp. 35-131).Glencoe, IL: Free Press.

Scanzoni, J. (1982). Sexual bargaining: Power politics inthe American marriage (2nd ed.). Chicago: Universityof Chicago Press,

Smedley, J. W, & Bayton, J. A. (1978). Evaluative race-class stereotypes by race and perceived class of subjects.Journal of Personality and Social Psychology, 36, 530-535.

Spence, J. T., & Helmreich, R. L. (1978). Masculinity &femininity: Their psychological dimensions, correlates,& antecedents. Austin: University of Texas Press.

Tajfel, H. (1981). Human groups and social categories.New York: Cambridge University Press.


Taylor, S. E. (1981). A categorization approach to stereo-typing. In D. Hamilton (Ed.), Cognitive processes instereotyping and intergroup behavior (pp. 83-114).Hillsdale, NJ: Erlbaum.

Taylor, S. E., & Crocker, J. (1981). Schematic bases ofsocial information processing. In E. T. Higgins, C. P.Herman, & M P. Zanna (Eds.), Social cognition: TheOntario symposium (Vol. 1, pp. 89-134). Hillsdale, NJ:Erlbaum.

Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman,A. J. (1978). Categorical and contextual bases of personmemory and stereotyping. Journal of Personality andSocial Psychology, 36, 778-793.

Taynor, J., & Deaux, K.. (1973). When women are moredeserving than men: Equity, attribution, and perceivedsex differences. Journal of Personality and Social Psy-chology, 28, 360-367.

Triandis, H. C. (1977). Interpersonal behavior. Monterey,CA: Brooks/Cole.

Upshaw, H. S. (1969). The personal reference scale: Anapproach to social judgment. Advances in ExperimentalSocial Psychology, 4, 315-371.

U.S. Department of Labor, Bureau of Labor Statistics.(1980). Perspectives on working women: A databook(Bulletin 2080). Washington, DC: Government PrintingOffice.

Weber, M. (1947). The theory of social and economic or-ganization. Glencoe, IL: Free Press.

Williams, J. E., & Best, D. L. (1982). Measuring sex ste-reotypes: A thirty-nation study. Beverly Hills, CA: Sage.

Zuckerman, M. (1979). Attribution of success and failurerevisited, or: The motivational bias is alive and well inattribution theory. Journal of Personality, 47, 245-287.

Received March 28, 1983Revision received November 9, 1983

New Journal on Aging: Call for Nominations

By action of APA's Publications and Communications Board and Council of Rep-resentatives, the APA is publishing a quarterly journal called Psychology and Aging,the first issue of which will appear in 1985. Psychology and Aging will contain originalarticles on adult development and aging. The articles may be reports of research orapplications of research, and they may be biobehavioral, psychosocial, educational,methodological, clinical, applied, or experimental (laboratory, field, or naturalistic)studies. For more information about the new journal, see the November 1983 issueof the APA Monitor.

Nominations for the editor of Psychology and Aging are now open. Candidatesmust be members of APA and should be available to start receiving manuscripts inmid-1984 to prepare for issues published in 1985. To nominate candidates, preparea statement of one page or less in support of each candidate. Submit nominationsno later than April 2, 1984, to:

Martha A. StorandtDepartment of PsychologyBox 1125Washington UniversitySt. Louis, Missouri 63130(314) 889-6508

Documents

Gender Stereotypes Stem From the Distribution of …gribouts.free.fr/psycho/menace%20du%20st%e9r%e9o/... · to hold positions of lower status and authority, ... Experiments 1 and