sex and sexism - a comparison of male and female sex role attitudes

Upload: black-blog-attack

Post on 07-Apr-2018

223 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    1/16

    Sex and Sexism: A Comparison of Male and Female Sex-Role Attitudes

    Author(s): Marie Withers Osmond and Patricia Yancey MartinSource: Journal of Marriage and Family, Vol. 37, No. 4, Special Section: Macrosociology of theFamily (Nov., 1975), pp. 744-758Published by: National Council on Family RelationsStable URL: http://www.jstor.org/stable/350825 .

    Accessed: 13/09/2011 06:49

    Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

    JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

    content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

    National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access to

    Journal of Marriage and Family.

    http://www.jstor.org

    http://www.jstor.org/action/showPublisher?publisherCode=ncfrhttp://www.jstor.org/stable/350825?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/350825?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=ncfr
  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    2/16

    Sex and Sexism: A Comparison of Male andFemale Sex-Role Attitudes*MARIE WITHERS OSMOND**

    ANDPATRICIA YANCEY MARTIN***The Florida State UniversityAn analysis of sex-role attitudes by sex of respondent highlights the complexityoqfthe sex-roles concept. The examination qf the roles of both sexes in terms of.fourtheoretical components revealed the fbllowing results. Males and females tend toshow the least divergence over (1) macrolevel social change issues and (2) the familialroles of both sexes. They differ most over (3) extrafamilial roles of.females and (4)the stereotypes of both sexes. Both men and women tend to be most traditional inthe area of familial roles and most modern in relation to macrosocial change issues.A conceptualization of a general sex-role attitude continuum is suggested which em-phasizes the degree of sex-differentiation and sex-typing rather than the fiequentlyused equality-inequality framework. Results of the data analysis (N = 480 college-level juniors and seniors) are interpreted in line with the differentiation/typingconceptualization. Suggestions for the future focus on the need for refinement inconceptualization and operationalization/measurement of the sex roles and sex-roleattitudes of both sexes.

    Recent studies of sex-role attitudes can bedivided into two broad categories. First arethose which have focused on the form andsubstance of the attitudes themselves and/oron the correlation of various social-psycho-logical traits with these attitudes (cfq.McKeeand Sherriffs, 1959; Steinmann and Fox,1970; Ferber and Huber, 1975; Mischel,1974; and Hesselbart, 1974 and 1975).Second, a group of studies has concentratedon the association of sex-role attitudes withvarious characteristics of subjects other thansocial-psychological ones; for example,demographic variables, socioeconomic sta-tus, family situation or background, and soforth (cf' Papanek, 1969; Sterrett andBollman, 1970; Meier, 1972). Evidence fromboth types of studies, however, has consis-tently underscored one empirical result: thatmales and females differ in their sex-role

    attitudes. The existence of differences by sexin the amount of "sexism" expressed is afinding which, though often reported, isseldom analyzed. The present researchrepresents a beginning effort to fill this gap.Three empirical questions provided theimpetus for this study: (1) Are theredifferences between males and females intheir attitudes regarding both male andfemale roles? (If so, what distinguishes malefrom female attitudes?) (2) How, if at all, aremales who hold traditional sex-role attitudesdifferent from females with traditionalattitudes? And (3) how, if at all, are maleswith modern sex-role attitudes distinguishedfrom females with modern attitudes? Inshort, our questions involve an exploration ofvariations in sex-role attitudes by sex ofrespondent, both for our total sample and forthe "traditional" and "modern" subsamplesat each end of the sex-role attitudecontinuum.

    BACKGROUNDThree features of the present studydistinguish it from much of the research onsex roles which has been published previous-ly. First, our measure of sex-role attitudes is a32-item Likert scale, which is intended to tap

    *This paper was presented in the session on SexRoles at the annual meeting of the American SociologicalAssociation, San Francisco, California, August, 1975.** Department of Sociology, Florida State University,Tallahassee, Florida 32306.***School of Social Work and Institute for SocialResearch, Florida State University, Tallahassee, Florida32306.

    744 JOURNAL OF MARRIAGE AND THE FAMILY November 1975

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    3/16

    several components of sex roles and ofsex-role related assumptions and behaviors.Second, we analyze attitudes at each end ofthe sex-role continuum; that is, at both the"modern" and "traditional" ends. Third,and perhaps of greatest significance, weattempt to analyze attitudes towards the rolesof both sexes. Among the sociologists whohave recently urged the latter are Komarov-sky (1973), Keller (1974), and Stein (1974).In line with our third goal, we attempt toclarify both previous and current conceptu-alizations of the sex-role attitude continuum.Although one conceptualization has beenpredominant in the literature, at least twothemes appear to have emerged. The first ofthese, and the one which has received themost attention, is the equality versusinequality continuum (for example, seeMeier, 1972; Lipman-Blumen, 1972; andBayer, 1975). Second, and the conceptional-ization which we believe to be the moreheuristic, is a continuum with dichotomousdifferentiation and typing of social roles onthe basis of sex at one end and nondifferen-tiation and/or nontyping of roles on the basisof sex at the other extreme (see Hefner et al.,1974; Brabant and Garbin, 1974; Kutner andBrogan, 1974; Lipman-Blumen, 1973; andChafe, 1972).In much of the literature, the equality-inequality continuum tends to focus atten-tion, either explicitly or implicitly, on femaleroles. The notion of "sex-equality" or"egalitarianism" appears to referprimarilytothe acceptance of females in the occupationalsphere on an equal par with males. There isseldom any suggestion of how male roles willbe (or should be) changed thereby. For thosewho have utilized this "equality" continuum,the attitude that "women belong in thehome" is generally labeled as the traditionalor "sexist" end; and the attitude that womenshould have "equality" with males in thework sphere is labeled variously as the"liberal," "modern," or "contemporary"end.Closer to our current thinking is theconceptualization of a sex-role continuum asone which involves lesser or greater degrees ofsex differentiation and typing. Since weattempt explicitly to deal with the social roles.of both sexes, not just of females, we findHefner et al.,'s (1974) delineations of the twoends of this continuum most useful. They

    define "traditional" sex roles as those whichare based on polar, dichotomous conceptionsof the nature and roles of men versus women."Modern" roles, on the other hand, arecharacterized by flexible and dynamictranscendence of sex-role constraints; that is,"modern" definitions of social roles are notspecified by "sex."' For a person whoseattitudes fall at the modern end of thecontinuum, sex would simply no longer be arequisite characteristic for the occupancy ofsocial positions or for the performance ofsocial roles.While additional work is required todevelop and refine this continuum further webelieve it holds sufficient promise to justifythe effort.2

    METHODOLOGYData Collection

    The data for this study were collected byquestionnaire from undergraduate studentsselected via a stratified random sample ofjunior- and senior-level classes at a large,state-supported university in Florida. Allsections of junior- and senior-level classes(excluding laboratory or recitation sessions;field, intern or practicum experience; DIS's;etc.) were stratified according to one of theeight colleges or schools of the university.3Then, within each of these divisions, a fiveper cent systematic random sample of classeswas drawn, and permission of the instructorto administer the questionnaires during aregularly scheduled class period was sought.Thirty classes were selected via the samplingprocedure, and 28 participated. All data werecollected during a two-week period inmid-July (1972), and 480 completed ques-tionnaire schedules were obtained. Of these,225 were from males and 255 from females.The questionnaire schedule consisted of 81

    'Of course, as Kemper (1974) notes, the biologicalroles of impregnation and childbearing cannot be alteredand will remain sex-specific.2Toview sex roles from this perspective may enhanceour sensitivityto the many varieties of "division of laborby sex" which currently exist (across families as well asacross cultures). At a minimum, we believe the latterconceptualization, in comparison to the equality-ine-qualityone, places the scholarlyexamination of sex roles(and sex-role attitudes) in less of an evaluative light.3These included Arts and Sciences, Nursing, HomeEconomics, Social Welfare (and Criminology), Business,Education, Library Science, and Music.

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 745

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    4/16

    questions on a variety of topics, including:female and male sex roles; sex-role change; avariety of values and attitudes; current socialissues; and a number of family, school, anddemographic variables.4 Most of the ques-tions were highly structured although some(such as those dealing with abortion, legaliza-tion of drugs, etc.) were open-ended.Measurement of Sex-Role Attitudes

    Included in the questionnaire was theOsmond-Martin 32-item Sex-Role Attitude(SRA) scale, a Likert-type scale with fiveresponse categories (from strongly agree tostronglydisagree). Table 1 presents the itemsin the scale in terms of four generalcomponents: (1) familial roles of females andmales; (2) extrafamilial roles of each sex; (3)stereotypes of male/female characteristicsand behaviors; and (4) social change asrelated to sex roles. Although we conceived ofthe scale as tapping a single theoreticaldimension (modern to traditional sex roles, asdefined above), we thought it inadvisable tolimit the contexts of such a dimension.Our a priori grouping of the items wasbased on two general considerations. First,since sex-role definitions and behaviors are,emotionally, highly-charged issues, it is onlyrealistic to expect that the attitudes (modernor traditional) evoked may be dispersed withregard to the types of roles associated withthem. Forexample, it might be unwise to taponly family roles since attitudes in that realmcould be changing more (or less) rapidly thanthose in a work context. Second, ourmethodology required prior theorizing. In-creasingly, researchers are being warned thatmeasurement must rest on a theoretical base(cf Smith, 1974; Osmond and Martin, 1974).We therefore attempted to make our a priori

    conceptualizations as explicit and specific aspossible.",Reliability of the SRA scale was examinedthrough a Likert-scaling item analysis.Resulting t-tests showed that 31 of the 32items discriminated between the top andbottom quartiles of the sample withassociated probabilities of .001 or less. Theitem (see Table 1, item 25, Battle of Sexes)which did not meet this criterion (.01 < p

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    5/16

    TABLE 1. SEX-ROLE ATTITUDE (SRA) SCALE ITEMS GROUPED ACCORDING TO A PRIORI CON-CEPTUALIZATION*I. Familial Roles**(1) Women with children in grammar school should, if at all possible, stay at home rather than work (Gram-mar School Children).(2) Women with preschool children should not work-if at all possible (Preschool Children).(3) Whoever is the better wage-earner, wife or husband, should be the breadwinner (Wage-Earner).

    (4) It is possible for women to satisfy their needs for achievement through their husbands (Achieve ThroughHusbands).(5) Men should have more freedom to do such things as cook and care for children, if they so desire (Men Freeto Cook).(6) A man's self-esteem is severely injured if his wife makes more money than he does (Man's Self-EsteemHurt).(7) Men should take the same amount of responsibilityas women in caring for home and children (Men Re.Home/Children).(8) A husband who is the breadwinner in the family should make all the important decisions (Husband MakeDecisions).II. Extrafamilial Roles(9) I would feel uncomfortable if my immediate supervisor at work was a woman (Woman Supervisor).(10) To a great extent, women are less able to make a career commitment than men are (Female-Male Com-mitment).(11) Females should be encouraged to plan for a career, not just a job (Females Plan Career).(12) I would vote for a woman for President of the United States (Woman President).(13) Women are less capable of making important decisions than men are (Female-Male: Important Deci-sions).(14) Men are more capable of assuming leadership than women are (Male-Female: Leadership).III. Stereotypes of Male/Female Nature and Behaviors(15) Women generally prefer light conversations over rational discussions (Light Conversation).(16) There is considerable evidence that men, in general, are a "superiorspecies" to women (Male Super-Spe-cies).(17) Women really like being dependent on men (Females Like Dependence).(18) Career women generally are neurotic (Career Women Neurotic).(19) Females should go ahead and pamper males-"Tell him how great he is"-because that's a useful way toget what they want (Pamper Men).(20) Men should stop appraising women solelyon the basis of appearance and sex appeal (Female Sex Appeal).(21) Either consciously or unconsciously, most women would like to be men (Females Like to be Male).(22) The way men and women behave is more a result of their genetic make-up than of the way they werebrought up (Genetic Make-up).(23) Women are as capable as men of enjoying a full sex life (F-M Enjoy Sex).(24) Since men have a natural urge to dominate and lead, women who challenge this actually threaten the wel-fare of society (Male Dominance Natural).IV. Social Change as Related to Sex-Roles(25) Unlike the race riots, the "battle between the sexes" will never involve violence on any large scale (BattleSexes).(26) There should be low-cost, high-quality child-care centers for working women (Child Care Centers).(27) Men need liberation equally as much as women do (Male Liberation).(28) Men's clubs and lodges should be required to admit women (Clubs/Lodges).(29) Women should get equal pay with men for doing the same jobs (Equal Pay).(30) Women should have equal job opportunities with men (Equal Job Opportunities).(31) Women can attain true equality in this country only through a really drastic change in the social structure(Drastic Change).(32) The Equal Rights Amendment related to sex should be ratified as soon as possible (ERA).

    *In parentheses, after each item, is a code name for reference to the item later in the text.**It should be emphasized that the items were presented to subjects for their responses in a quite different orderthan shown in Table 1. Persons desiring a copy of the scale as it was administered may obtain one by writing to theauthors.

    tion of percentage differences between malesand females in their responses to the 32 scaleitems; (2) t-tests for the significance ofdifferences in male-female scores on eachitem; and (3) utilization of the AutomaticInteraction Detector (AID) program: (a) to

    examine whether the scale items compliedwith an additive or multiplicative model; and(b) to analyze the configurational patterningof item responses by sex of respondent.Because of the relative dearth of AIDapplications in the sociological literature, it is

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 747

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    6/16

    perhaps wise to review the basics of themethod and to justify its use in the presentcontext.AID involves a repeated one-way analysisof variance process which follows the basicprinciple of least squares. The focus is onpower in reducing error in a criterionvariable: that is, on the relative importanceof various predictors rather than on theirstatistical "significance." The test of impor-tance, reduction in the error sum of squares,is expressed at each step of the analysis foreach predictor in terms of the proportion ofvariation in the (dependent) group that wouldbe explained if a specific predictor were usedto split the group into two new groups. Thecriterion (or dependent) variable may beeither binary or measured at the intervallevel, while the predictor (or independent)variables must be in the form of classifica-tions.AID has been used most often when theresearch purpose was specification of ex-planatory models. According to Sonquist(1970:74), however, a basic aspect of theoriginal rationale for developing AID was theexpressed concern of socio- and psychometri-cians (cf. Stouffer et al., 1952; Meehl, 1950)with obtaining adequate representation of anunderlying dimension by a set of interactingindicators. They also recognized the need fora searching technique that would permitexamination of the possibility that the shapeof the effects of one indicator on anotherdiffered in various parts of the sample. Insum, as Sonquist implies, these early basesfor AID's development can continue to serveas guidelines for its present use.

    RESULTSItem Responses by SexTo address our first research question,concerned with sex differences on sex-roleattitude items, Table 2 was constructed. Theintent of Table 2 is to present the percentagedifferences between females and males intheir responses to each of the 32 scale itemsand, in particular, to do so at each "end"(modern and traditional) of the attitudinalcontinuum. The first column represents themore modern end of the response continuum(reflected scores of either one or two on afive-point scale). Since we expected, fromprevious research, that the items would elicita more modern response from females, the

    figure in column one is the percentage offemales who answered one or two on the itemminus the percentage of males who did so. Aplus sign (+) before the difference indicatesthat more females than males did respond (asexpected) in a modern fashion; a minus sign(-) indicates that fewer females than malesgavea modern response. Similarly, the secondcolumn represents the more traditional end ofthe continuum (reflected scores of four or fiveon a five-point scale). Here we reverse thesubtraction so that the figure in column twoindicates the percentage of males whoanswered four or five on the item minus thepercentage of females who did so. A plus sign(+) indicates more males than femalesresponded traditionally, while a minus sign(-) means fewer males responded tradi-tionally. One example should make theprocess clear. For item nine, Woman Super-visor, 16 per cent more females than maleschecked the modern response categories (88.6per cent of the females checked categoriesone or two minus 72.4 per cent of the maleswho likewise checked categories one or two);17 per cent more males than females checkedthe traditional response categories (20.0 percent of the males checked categories four orfive minus 3.1 per cent of the females check-ing these categories).In general the results were as expected.More males than females gave a traditionalresponse (to 25 of the 32 items), while morefemales than males gave a modern response(to 29 of the 32 items). Thus the sign of thepercentage difference in both columns(modern and traditional) of Table 2 is mostoften positive (+). T-tests of differencesbetween male and female mean scores oneach item showed associated probabilities ofless than .05 on 23 items. In other words,males and females were differentiated on 72per cent of the items.With regard to reversals in expectedpatterns, only for item 6, Man's Self-EsteemHurt, were males considerably less likely thanfemales to give a traditional response (that is,agree with the item) while females were lesslikely to give a modern one. In short, on thisitem, men were significantly more modernthan women were (21 per cent fewer malesresponded traditionally). Three other itemsshowed more modest reversals in expectedresponse patterns: item 3, Wage-Earner (twoper cent fewer males responded traditional-

    748 JOURNAL OF MARRIAGE AND THE FAMILY November 1975

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    7/16

    TABLE 2. SEX-ROLE ATTITUDE (SRA) SCALE ITEMS BY PERCENTAGE DIFFERENCES IN FEMALE(N= 255) AND MALE (N=225) RESPONSES AT EACH END OF THE ATTITUDINAL CONTINUUM% Males Minus %% Females Minus % Females GivingMales Giving Modern Traditional

    Items Response to item' Response to itemI. Familial Roles

    (1) Grammar School Children***' + 15% +11%(2) Preschool Children + 3 + 1(3) Wage earner + 9 - 2(4) Achieve Through Husband*** +19 + 7(5) Men Freeto Cook*** +13 + 6(6) Man's Self-Esteem Hurt*** -24 -21(7) Men Re. Home/Children + 2 + 1(8) Husband Makes Decisions** + 7 + 3II. Extrafamilial Roles

    (9) Woman Supervisor*** +16 +17(10) F-M CareerCommitment*** +19 +11(11) Females Plan Career*** +13 0(12) Woman President + 4 + 5(13) F-M: Important Decisions*** + 18 + 10(14) M-F: Leadership*** +28 +21III. Stereotypesof Male/Female Nature and Behavior(15) Light Conversation*** +22 + 9(16) Male Super-Species*** +30 + 3(17) Females Like Dependence +14 - 6(18) CareerWomen Neurotic*** +26 + 7(19) Pamper Men** +12 + 6(20) Female Sex Appeal*** +12 + 7(21) Females Like to be Male*** + 19 + 1(22) Genetic Make-up*** + 9 + 2(23) Females Enjoy Sex* + 6 + 1(24) Male Dominance Natural* +10 0IV. Social Change as Related to Sex Roles(25) Battle of Sexes** + 4 +10(26) Child-CareCenter** + 6 + 5(27) Male Liberation + 8 0(28) Clubs/Lodges - 2 3(29) Equal Pay*** + 6 + 3(30) Equal JobOpportunities*** + 9 9(31) Drastic Change + 6 +8(32) ERA 0 + 3

    'Onceeach of the items was properlyreflected (so that its positive or negative direction was taken into account), ascore of 1 or 2 always indicated a modern response and a score of 4 or 5, a traditional one.2t-testfor significance of differences between the sexes significant at: p < .001 = ***p

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    8/16

    Natural; and item 27, Male Liberation), anequal proportion of females and males gave atraditional response; and on one item (item32, ERA) an equal proportion gave a modernresponse. We point this out primarily tocaution future researchersagainst generaliza-tions made on the basis of only one end of anattitudinal continuum. That is, simplybecause no sex differences appear at one endof the continuum, one cannot conclude (as iswitnessed by the above examples) that thereare no differences at the opposite pole.If we review the results in Table 2 by our apriori conceptualization of the four groupingsof items presented in Table 1, we are able tonote which areas provoke the greatestdisagreement between females and males;which areas evoke the least amount ofdisagreement between the sexes; and,conjointly with each of these, which items inwhich areas evidence more complex responsepatterns.Men and women in our sample tend todisagree most consistently in the areas of"extrafamilial roles" (Component II, Table1) and of "stereotypes" (Component III,Table 1). However, there is an importantdistinction between responses in these twoareas. It is in the context of extrafamilialroles that females and males differ signif-icantly in their responses at each end of thecontinuum.sThe stereotype items, on theother hand, elicit considerably more modernresponses for the women, but the maleresponses do not "balance" the femaleones here in the sense of being stronglytraditional. 9These findings suggest two con-clusions. First, the basic issue (or conflict)with regard to sex-role behavior is in therealm of female-male participation in theeconomy and the polity (extrafamilial roles).

    Second, both sexes in our college sampleappear to have had some "consciousnessraising" with regard to "sexist" stereotypes.That is, although women are rejectingstereotypical statements to a greater extentthan are men, still men are not currentlyendorsing such stereotypes to any extent. Wereturn to these findings in our subsequentanalysis.Turning to the areas where women andmen in the sample appear to disagree least,the contexts are "familial roles" (ComponentI, Table 1) and "social change as related tosex roles" (Component IV, Table 1). Of theeight items concerned with female-malefamilial roles, five items exhibit significantdifferences in response by sex. Further, oneof these is the Man's Self-Esteem Hurt item(number 6) where responses were opposite tothe expected direction. Even more striking, ofthe eight items dealing with macrosocialchange of sex roles only four significantlydifferentiated men and women; and one ofthese four is the Battle of Sexes item (number25) which was the single item revealed by theitem analysis as unable to differentiate theupper and lowerquartiles of ourtotal sample.We speculate on the meaning of responsepatterns to these two rather intriguing items,Self-Esteem and Battle, in later sections ofour paper. The point to be underscored hereis that we find evidence of the greatestconsensus between the sexes in those areaswhere prior research has led us to expect themost controversy; that is, over the familialroles of both sexes and over macro, socialchange issues. The following multivariateanalysis provides an additional perspective onthese results.AID Analysis by Sex

    Thus far, we have reported on a bivariateanalysis of the relationship between asubject's sex and his/her response to each ofthe 32 items in our SRA scale. In this sectionwe utilize the AID computer program todifferentiate between the sexes in a multi-variate analysis of the same scale items. Thatis, with sex of subject as the criterion variable(male = 1; female- 0), AID is used toassist us in determining the relative impor-tance of the various scale items fordifferentiating between the sexes.10We em-

    8On four of the six items in this group, males andfemales differ by ten per cent or more at each end of thecontinuum. Further, t-tests show associated probabilitiesof .05 or less for five of the six items.

    90Onight of the ten stereotype items, at least ten percent more women than men gave a modern response.Men, however, on the traditional end of the continuumwere about equally as likely as women to respondtraditionallyon five of the items (#'s 16, and 21 through24), less likely to do so on one of them (#17), and onlymoderately more likely on the other four items(percentages varying from six to nine per cent on items15 and 18 through 20). It should be noted, however, thatmale-female responses to nine of the ten stereotype itemsdo differ significantly (p

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    9/16

    ploy AID to analyze both the total sample(N = 480) and two subsamples of subjects,one at each end of the SRA continuum.Analyses of the subsamples allow us toaddress our second and third researchquestions regarding the differences between(1) traditional males versus females and (2)modern males versus females. The subsam-ples were established on the basis of thesubjects' summated, Likert SRA scale scores.Subjects were divided at the median (of 76)on the scale; 244 fell at or above the medianand became our "traditional" group, and 236fell below the median, becoming our"modern" group.Let us preface these results with a briefoverview of the application of AID in thepresent context. First, the criterion variablein the AID analyses is, in fact, a binary or"dummy" variable of sex of subject and, asconstructed here, tells us the proportion ofmales in any given group (or subgroup) of thesample. The effector ("predictor") variablesin the analyses are the 32 SRA items. TheAID algorithm actually proceeds by compar-ing all of the scale items on the basis of theirrespective abilities to "divide the sample" soas to maximize the difference betweenresultant subgroups in terms of their sexcomposition (i.e., the proportion of malesubjects in these groups)."I

    Second, the measure used to rank order thescale items (on the basis of their abilities to"explain proportion of total variance") is theAID partial beta coefficient (f32) of theindividual items. Here our interest is notdirected toward explaining variance in thecriterion of sex of subject. We are asking theprogramto detect the comparative abilities ofour SRA items in distinguishing (ordifferentiating) men from women in amultivariate model. Thus, we interpret theitems' "variance values," or AID Beta-squared coefficients, primarily as roughindicators of their relative importance fordelineating the sexes. 12With the above explanations in mind, wenow consider the AID analysis results aspresented in Table 3 both for the total sampleand for the "modern" and the "traditional"subsamples. The various scale items whichentered one or more of the three analyses areshown in Table 3 along with each item's Beta-squared coefficient and its rank order ofimportance (as based on Beta-squared). Thestarred (*) items are those which resulted indivisions of the sample that were judged asreliable/stable on the basis of the threecriteria of split reducibility, split eligibility,and subgroup size (see Footnote 11). Since itis quite likely that with a larger sample size agreater number of items would haveproduced such reliable divisions of the

    (ranking) of scale items when a criterion is available (see,for example, Wang and Stanley, 1970). For our purposesthis method has two major disadvantages. First, eachindicator has its effect measured over the entire sample(while AID measures the effect of each on each subgroupof the sample). Second, interactioneffects either have tobe assumed away or "built in" at the beginning of theanalysis when the indicators are defined. Further, otherrestrictive assumptions involved in regression (e.g.,interval level of measurement, linearity, etc.) are notrequired by AID."This is a sequential procedure in which the program

    examines the full data using each predictor, and witheach searches for the best single division according tothat predictor. "Best" (Sonquist, Baker, and Morgan,1971:9) means the largest reduction in predictive errorfrom knowing to which of two subgroups on thatpredictor each case belongs (and, for our problem, theproportion by sex of those subgroups). This is a repeatedstepwise operation in which, at each step, the programselects for examination and possible further splitting thegroup with the largest remaining unexplained variance.The process stops when one or more of several presetcriteria are not met. The criteria we utilized were (1) the"split reducibilitycriterion," i.e., additional reduction inproportion of error variance must be at least .015; (2) the"size of subgroup criterion," i.e., the split must result in

    both subgroupshaving at least 25 cases each; and (3) the"split eligibility criterion," i.e., the group must containat least .010 of the total sum of squares to become acandidate for splitting (Sonquist and Morgan, 1964:22-23).12Theformula for (32X s as follows:

    032X= ~TSSiX--ITSSjXi jTSST

    where i is over all parent groups split by predictor X, andj is over all new groups formed by splitting a parentgroupon predictor X. The AID Beta-squared coefficientindicates the proportion of the total variation actuallyexplained by each of the variables in the analysis. TheseAID coefficients sum to R-squared. Beta-squared in AIDmay be interpreted as an approximate counterpart to"multiple partials" (or squared multiple partialcorrelationcoefficients) in least-squares regression analy-sis. R-squared has exactly the same interpretation forboth methods. There is, however, a major distinctionbetween the two. Multiple partials indicate the amountof unique variance which a variable is able to explain in

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 751

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    10/16

    TABLE 3. SCALE ITEMS WHICH ENTERED THE AID ANALYSES, WITH SEX OF SUBJECT (MALE 1;FEMALE = 0) AS THE CRITERION (FOR THE TOTAL SAMPLE AND TWO SUBSAMPLES)TraditionalTotal Sample Modern Subsample Subsample

    (N= 480) (N= 236) (N= 244)Partial Partial PartialBeta Rank Beta Rank Beta RankItem Numbers and Names Coefficient Order Coefficient Order Coefficient Order(13) F-M Important Decisions .107* 1 -- .021 8(6) Man's Self-Esteem Hurt .068* 2 .087* 2 .119* 1(9) Woman Supervisor .064* 3 .017* 8 .091* 2(25) Battle of Sexes .046* 4 .052* 4 .015 10(18) CareerWomen Neurotic .045* 5 .081* 3(17) Females Like Dependence .035* 6 -- .076* 3(8) Husband Makes Decisions .028 7 -- .015+ 9(4) Achieve Through Husband .021 8 --(21) Females Like to be Male .018 9 --(3) Wage Earner .016 10 .041 5 --(14) M-F Leadership .015* 11 .126* 1 .039 4(30) Equal Job Opportunities .014+ 12 .016 9 --(15) Light Conversation .014 13 --(31) Drastic Change .011 14 --(10) F-M Career Commitment .008 15 -- .027 6(20) Female Sex Appeal -- .034 6 .015* 11(27) Male Liberation -- .030 7 .013 12(32) ERA -- .012+ 10 --(24) Male Dominance Natural -- .012 11 .034 5(23) Females Enjoy Sex -- .007 12 --(29) Equal Pay - -- .026 7(2) Preschool Children -- - .013 13(28) Clubs/Lodges .- -- 006 14

    R2 = .505 R2 = .513 R2 = .506*Asterisks identify those items which resulted in highly stable/reliable divisions of the sample. (See Footnote 11,for greater explanation.)+Apparent ties in the f2's were not ties before rounding to three decimal places.

    sample, results in Table 3 should be viewedmore as suggestive than as definitive.'3 Consider first the results from the totalsample analysis. Here, 15 of the 32 scaleitems entered the AID configuration modeland 14 of these explained at least one per centof the variance in sex of subject (thecriterion). Seven items (those starred in Table3 under the total sample column) produceddivisions of the sample which can beconsidered highly stable/reliable. The R-squared coefficient shows that approximately50 per cent of the variation in our criterion isexplained by those 15 items, i.e., there is asubstantial association between sex of subjectand sex-role attitudes.

    some dependent variable, "above and beyond all othervariables" and/or with "all other variables controlled."With AID, no variables are "controlled." That is, theAID analysis progresses step-by-step according to whichpredictor variable can divide a group (or subgroup) ofthe sample on the basis of the criterion (or dependent)variable. Therefore, the AID partial beta coefficient mayindicate for a given item its ability to separate the twosexes in some subgroup of the sample but not necessarilyin the "original group" (or the total sample), and notnecessarily in comparable parts of the sample at thesame time. In reference to both of these latter points, theinterested reader should consult Sonquist's (1970)explanation of evidence for additivity versus interactionas related to the symmetry (or asymmetry) of the "tree"which results from the divisions of the sample by the AIDalgorithm.

    "What is meant here is that we should concludeneither that only those items are important whichproduced reliable divisions of the sample nor that thoseitems are unimportant which failed to do so. Additionalgrounds for caution in interpreting Table 3 center on theproblem of multicollinearity. Because of the relatively

    high intercorrelation between many of the scale items,some items may have emerged late in the analysisprimarilyas a result of their independence of items whichentered earlier and not as a result of their substantivesignificance. Likewise, items which are not shown inTable 3 may be missing for the opposite reason; that is,they were highly associated with items which entered theanalysis early.752 JOURNAL OF MARRIAGE AND THE FAMILY November 1975

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    11/16

    Relating the total sample results in Table 3to our four a priori groupings of items (Table1), the importance of female extrafamilialroles for differentiating between the sexes isyet again emphasized. Note that among theitems ranked from one to six, the first and thethird concern women as decision-makers andas supervisors, respectively. Further, we findat least one item among the first six fromeach of the other three components. Man'sSelf-Esteem, which was considered to be afamilial-role item, ranks second; Battle ofSexes, a social change item, fourth; and twostereotype items, Neurotic and Dependence,rank fifth and sixth, respectively. In short,the fact that items are represented from eachof our a priori groupings provides evidencethat : (1) the concept of sex-role attitudeshas a number of fairly distinct components;and (2) when attempting to understand therelationship between sex roles and sex ofsubject, items which reflect each of thesecomponents should be included.One other feature of the total sampleanalysis was provocative to us: the evidenceof non-additivity (or interaction) of the items.AID is unique, relative to many analysistechniques, in that it does not force aparticular form on the data, i.e., neitheradditive nor cumulative (Sonquist, 1970:106-111). Rather, the investigator is allowed to seethe form which the data actually take for theparticular sample. In our case, the resultspoint to a predicted, albeit complex,bifurcation of the sample into "modern" and"traditional." The portion of the sample(N = 158) which split toward the modernend of the continuum contained significantlyfewer males (25 per cent) than did the group(N = 322) which split toward the traditionalend (57 per cent). Also, the first division ofthe sample resulted in a separation ofsubjects who "very strongly disagreed" that"Women are less capable of makingimportant decisions than men are" from allsubjects who gave other responses. Subse-quently, this very modern group split againand again in a cumulative fashion (i.e., onlyon the modern end of the responsecontinuum) over four additional items. Onthe traditional end, the 322 subjects werefurther divided in a more symmetrical pattern(i.e., splitting on both modern and tradi-tional response categories) and over a largernumber of items. We do not go into the

    details of these splits because only furtherresearch with different samples can showthem generalizable. The results, however, doprovide impetus for further research. In ourongoing work, we are (1) testing to see if acumulative, Guttman-type, scale is moreappropriate (than the Likert-type) for"modern" subjects (subjects who have beenidentified, by other criteria, as displayinglesser degrees of sex-role typing or ascrip-tion); (2) examining the group of "deviant"males at the very modern end of thecontinuum, and a comparable group of"deviant" females at the very traditional endof the continuum, to learn if and how theirbackgrounds are distinguished from those oftheir more typical same-sex counterparts.The subsample analyses extend ourunderstanding of males and females at eachend of the sex-role continuum. It is especiallyrevealing to contrast the more egalitarianmales with their female counterparts (refer tothe third and fourth columns of Table 3). Inaddition to the Self-Esteem and Battle items,the modern males tend to distinguishthemselves from the modern females on threeapparently interrelated items: Women asLeaders (item 14), Career Women andNeurosis (item 18), and Women as Supervi-sors (item 9). All three of these items mightbe construed as indicative of the inability ofthese "otherwise modern" males to fullyaccept women in a professional work role.That is, while rejecting most of the femalestereotypical items and responding in amodern fashion to a majority of thefemale/male familial roles and sex-rolechange items, they yet agreed: that careerwomen are neurotic; that a female supervisorwould make them uncomfortable; and thatmen aremore capable of assuming leadershipthan women are.The other two items which differentiate thesexes in the modern subsample are #6, Man'sSelf-Esteem Hurt, and #25, Battle of Sexes.On the Self-Esteem item, the "otherwisemodern" females responded more tradition-ally than their male counterparts. On theBattle item, the "otherwise modern" malesresponded more traditionally than thefemales. It should be recalled (from theprevious section on item analysis) thatthese are the very two items whichsignificantly differentiated females and malesin the two areas where the sexes appeared to

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 753

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    12/16

    disagree least (familial roles and macrosocialchange). Here, the AID analysis has extendedanother clue with regard to interpretation.The Self-esteem item proved importantthroughout the AID analyses (ranking eitherfirst or second in importance for the totalsample and for both subsamples). Thus wecomment on it more fully in the discussion atthe end of our paper. The Battle item,however, was rejected in the Likert analysisand ranks as tenth in importance in the AIDanalysis of the traditional subsample.14 Themajor difference between the sexes overBattle emerges in the AID analyses asbetween males and females at the modernend of the sex-role continuum. It is ourhypothesis (though we have no means oftesting it) that this item separates some of theself-styled "radical" women in our samplefrom their "liberal, but not radical" malecounterparts. At any rate, the power of theBattle item in the AID analyses cautionsresearchers against premature rejection of anitem on the basis of only one analyticmethod. 15Whereas the differences between modernmales and females tended to focus on femaleprofessional roles, differences between thesexes in the traditional subsample appear toinclude a wider variety of realms. Traditionalmen differ from traditional women byexpressing dislike for a woman supervisor(item 9); by agreeing that men are morecapable of assuming leadership than women(item 14); and by believing that women areless able than men to make a commitment toa career (item 10). All three of these itemsrefer to extrafamilial roles of females, andappear to parallel responses of the males inthe modern subsample. However, traditionalmen also agree that men have a natural urgeto dominate (item 24), and disagree that menshould stop appraising women on the basis ofsex appeal (item 20). 16

    The traditional females, however, areperhaps the greatest surprise of our study.They, like the modern females, agree that amale's self-esteem is hurt if his wife makesmore money. However, they go on to take astance which is considerably more traditionalthan the men's on several other items. Thesewomen are characterized by: (1) agreementthat women like being dependent on men(item 17); (2) agreement that women withpreschool children should not work (item 2);and (3) disagreement that men needliberation (item 27). While the divisions ofthe traditional subsample based on the lattertwo items (Preschool and Male Liberation)failed to produce especially reliable splits,still they seem noteworthy as an instance ofwomen responding in the opposite directionfrom what one might expect. In sum, womenand men who are traditional in their sex-roleattitudes not only show a more complex arrayof attitudes but also display greater sexsegregation in such attitudes than do theirmodern counterparts.DISCUSSIONThe intention of this research has been to

    highlight the differences in male and femalesex-role attitudes at both ends of the sex-rolecontinuum (modern and traditional). Wehave evidence not only that such differencesare qualitative but that they also differ interms of the intensity with which they areendorsed.17 Since we argue that the sex-rolecontinuum is more usefully defined in terms oflesser or greater degrees of sex differentiation

    '4We should note that 75 per cent of the men and 65per cent of the women in our sample gave a traditionalresponse to this item.' For elaboration on this point and a comparison ofscaling techniques based on these same SRA items andsample, see Martin and Osmond, forthcoming.'"While item 20 "explains" less variance in thetraditional than in the modern group, it neverthelessproduced a more reliable division in the traditionalsubsample.

    "Here we wish to affirm the potential utility of theAID algorithm for sociological research. It is not onlyhelpful (when employed with its companion program,Multiple Classification Analysis, MCA) in exploring fornonlinear and nonadditive relations among predictorvariables, but, further, we believe it has as yetunexplored potential as a data reduction or scalingtechnique. The authors are currently completing a paperwhich compares AID as a scaling technique withtraditional Likert item analysis and with principal factoranalysis (see Martin and Osmond, 1975). Nonetheless,we should perhaps sound a cautionary note to potentialusers of the AID computer program. Like other methodsof data analysis, AID is not able to compensate for suchproblems as multicollinearity or measurement error.Further, since AID places so few restrictions on one'sdata, the potential complexity of splitting patterns whichcan emerge from its use (especially, say, if the researcherhas utilized variables on more than one level ofabstraction) can be quite misleading. We wish,therefore, to endorse strongly Sonquist's (1970) repeatedadmonition to use AID only after a great deal oftheorizing about one's topic has been done.

    754 JOURNAL OF MARRIAGE AND THE FAMILY November 1975

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    13/16

    and sex typing of roles (rather than in termsof equality-inequality), it seems appropriateto ask how our data analysis relates to thisconceptualization.One way to address this is to evaluate thefour SRA components (Table 1) in terms ofthe degree to which our subjects showed atendency to differentiate by sex and to sextype the various characteristics and rolesdelineated under each. First, we discuss whatcan be considered as the additive componentson this continuum, the areas in which: thegreatest amount of sex typing occurs for bothmales and females (the familial component);and where the least amount occurs for bothsexes (the social change component). Second,we examine the even more intriguing, albeitcomplex, non-additive components, the areasin which sex-role attitude is specified by sexof subject (the extrafamilial and stereotypecomponents).The fact that the subjects in our sampleresponded to the familial role component(Component I, Table 1) with the greatestamount of sex differentiation and sex typingis, on reflection, perhaps predictable. It is notdifficult to conceive of the family as the mostconservative of our social institutions. Ingeneral, we found acceptance by members ofboth sexes of a fairly sharp sexual division oflabor within the family, with the femalecontinuing to have primary responsibility forhome and child care. In our opinion, thissuggests that little genuine sex-role changehas occurred in the family realm. Since ourresults are from a highly educated sample ofyoung adults, we might anticipate (althoughit must be tested) that a lower-middle-classsample might respond in the traditionaldirection to an even greater extent. Further,if the attitudes of upper-levelcollegeyouth canbe taken as harbingers of social change, thesefindings suggest that relativelylittle change inthe sex distribution of familial roles is to beexpected in the near future. At any rate, inregards to the study of sex-role attitudes, ourresults caution against the use of only familialrole items.There is one family role item, Man'sSelf-Esteem Hurt (#6 in Table 2), that elicitedunanticipated responses. As mentionedearlier, 56 per cent of the males disagreedthat a man's self-esteem would be hurt if hiswife made more money than he (the"modern" response), while 56 per cent of the

    females agreed that it would be hurt (the"traditional" response).18 Moreover, thisitem proved to be important throughout theAID analyses.At least two alternative explanations of the"meaning" of the Self-Esteem item seemsplausible. The first rests on the assumptionthat the males' responses are accurate orvalid reflections of reality. The females'responses, therefore, must be presumed to beerroneous; that is, for some reason, womenhave misperceived that a wife's higherearnings would "bother" her husband'sself-esteem. There may even be some basis forassuming that the responses of our malesubjects to the Self-Esteem item reflectactual role change for males. McKee andSherriffs conclude that during a time ofsex-role change, " . . . the sex with higherstatus in the society will be able to expressovert change sooner than the sex with lesssecurity" (1959:362). Thus, it could be thecase that item 6 has tapped a significantaspect of recent sex-role change for males,one which females have so far failed toperceive. Why women have thus far notrecognized this "change" would be aninteresting question for future research.In contrast to the first, a second possibleexplanation rests on the assumption that thefemales' perceptions arethe valid ones. Here,it is assumed that the responses of our malesubjects are erroneous and that they reflecteither a conscious or an unconscious "socialdesirability" bias. That is, there may exist inour society an implicit normative proscriptionagainst a husband's "being bothered" if hiswife makes more money. Since there aredefinite norms that a man must work andearn money, while the norms regardingfemale employment and earnings are muchless distinct, perhaps males consider thisissue either to be unreal or else unlikely toarise. In this event, they may have respondedto the Self-Esteem item in terms of how theyfelt they should respond or in terms of howthey perceived other men like themselves

    t'Item 6 is one of only three of our 32 scale-items forwhich the 'modal' responses of the two sexes were atopposite ends of the continuum (the other two being item1: Grammar School Children and item 14: M-FLeadership). Further, Self-Esteem is the only item ofwhich it is true that the modal response of each sex wasat the opposite end of the SRA continuum from thatwhich was anticipated.

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 755

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    14/16

    would respond. Along this same line, it is alsopossible that our male subjects may haveresponded in the manner they perceivewomen would want them to respond. Again,McKee and Sherriffs observed that "men subjects are more perceptive of whatwomen desire in them as attributes than arewomen subjects insightful about the currentdesires of men for characteristics in them"(1959:363). Of course, the McKee andSherriffs study is over 15 years, old and theconclusions which they reached may nolonger hold.While both of the foregoing explanationsare no doubt inadequate, still it is likely thatthe true explanation falls somewhere betweenthe two positions outlined above. Our datacertainly lead us to conclude that males'self-esteem, as related to husband-wiferelations, is a topic which requires additionalinvestigation. This seems particularly so inlight of the central position which "husband'sself-esteem" has been given in much of thefamily literature on marital satisfaction andmarital intactness (for example, see Scan-zoni, 1966 and 1968; also, Osmond andMartin, 1974).The group of items on which our subjectsshowed the least amount of sex differentia-tion and sex typing was the social changecomponent (#IV, see Table 1). It can berecalled that all eight items in this componentdeal essentially with macrolevel social issues;thus, they may have represented, on apersonal level, the least "threatening" itemsin our scale. That women would respond in amodern fashion to such items as Equal Pay,Equal Jobs, Child Care, and the ERA seemsreadily understandable. Reasons for themales' modern responses, however, are lessapparent. Perhaps males perceived that ifsuch macrolevel changes did occur, theywould have few direct effects on their lives. Asfor potential indirect effects (for example,better jobs or higher pay for their wives,mothers, and daughters), our male subjectsmay have considered the possibilities andevaluated them positively. Public opinionpolls on the ERA, for example, have shownrepeatedly that equally as many men aswomen (and sometimes more) are in favor ofthe ratification of this amendment by thestates. This finding is replicated in thepresent sample where 55 per cent of bothsexes are in favor of the ERA.

    Turning now to the components whichproved non-additive in terms of ourconceptualized sex-role attitude continuum,the task of interpretation becomes moredifficult. That is, the quite distinctiveresponse patterns of the sexes to theextrafamilial roles and stereotype compo-nents (Components II and III, Table 1) makethe above type summarization (over bothsexes) inappropriate.In our opinion, one of the most significantfindings revealed by our data analysis is theidentification of the "nerve center" ofextrafamilial sex roles as the issue over whichmen and women disagree the most. Our dataappear to indicate that females may wellencounter strong resistance to sex-rolechanges which involve the assumption ofsupervisory, decision-making, and leadershiproles outside of the family. The males in oursample seem least willing to accept women inthese roles, and the females appear equally asinsistent not only that women should occupysuch positions but that they can perform inthem equally as well as men can. It isinteresting to consider these results in light ofthe goal of current Affirmative Actionprograms to advance, over the next fewyears, sizeable numbers of females to justsuch leadership positions. Judging from ourresults, once women attain these positions,what kinds of experiences might we expectthem to have? The frequently documenteddiscrepancy between "attitudes and behav-iors" might suggest that while males insubordinate positions (with female super-ordinates) might not particularly like theirstatus, they would nevertheless go along withit.On the other hand, the above discussionmay be quite premature. Perhaps the issue isnot what women would experience once inleadership positions, but whether or not maleattitudes such as those described here willprevent women from ever attaining suchpositions in the first place. This point seemscritical in light of current feminist chargesthat it is males who have, in the past,systematically excluded women from top-level, decision-making positions in all walksof life (in industry, business, politics,religion, and so forth).19 Issues such as the

    '9We are reminded of Kemper's recent (1974)discussion of how ascribed statuses tend to becomeassociated with differential amounts of social power.756 JOURNAL OF MARRIAGE AND THE FAMILY November 1975

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    15/16

    above certainly call for additional research.In a highly complex, industrialized societysuch as ours, most of our extrafamilialcontacts and interactions take place in formaland/or complex organizations. Hopefullyfuture research on the attitudes andbehaviors of men and women as they relate toroles outside the family will follow the recentlead of Miller, Labovitz, and Fry (1974) andexplore the organizational correlates of sexbias which tend to result in the exclusion offemales from positions of influence andleadership.The fact that we reserved stereotypes as thefinal component for discussion has beendeliberate. This is an area where the itemresponses yield the least interpretable resultsof our study.Attitudinal studies of minority groups arefrequently studies of stereotypes only. Oftenthese studies are based on the assumptionthat the degree to which subjects endorsecertain stereotypes is indicative of the degreeto which they are anti- (or pro-) change inminority group status. While we agree thatsuch an assumption may at times bewarranted, it seems increasingly apparentthat reality is more complex than theassumption implies. We are reminded that aperson may endorse, at once, two stereotypeswhich contradict each other; and that suchattitudes are often poor predictors of actualbehavior. In brief, while we found definitemale-female differences in the tendency toendorse stereotypes, we are somewhathesitant to interpret their meaning or toemphasize their significance. For example, asrevealed by our subsample analyses, tradi-tional females are more likely than tradi-tional males to endorse some of the stereo-types of women. Further, while we foundmore women than men rejecting the stereo-types of both sexes, we also found menonly moderately more likely than women toendorse them. More particularly, our re-luctance is based on what seem to beinconsistencies in the present findinas: thesame males who moderately endorsed most ofthe stereotypes were, at the same time,supportive of social changes which wouldbenefit women and unsupportive of women'soccupancy of supervisory and leadershippositions outside the family. We are able toconclude, then, only that additional inquiry iscalled for regarding the nature of the

    relationship between sex-role stereotypes andother aspects of both sex-role attitudes andbehavior.

    REFERENCESAndrews. Frank M., James M. Morgan, and John A.Sonquist1967 Multiple Classification Analysis:A Report on aComputer Program for Multiple RegressionUsing Categorical Predictors. Ann Arbor:Sur-vey Research Center, The University of Michi-gan.Bayer, Alan1975 "Sexist students in American colleges: a de-scriptive note." Journal of Marriage and theFamily 37:391-397.Brabant, Sarah and A. P. Garbin1974 "Sex of student as defined sociologically andhigher educational persistence." Paper pre-sented at the annual meeting of the SouthernSociological Society, Atlanta, Georgia.Chafe, William H.1972 The American Woman: Her Changing Social,Economic, and Political Role, 1920-1970. NewYork:Oxford University Press.Ferber, Marianne Abeles and Joan Huber1975 "Sex of student and instructor: a study of stu-dent bias." American Journal of Sociology 80:949-963.Haavio-Mannila, Elina1967 "Sex differentiation in role expectations andperformance." Journal of Marriage and theFamily 29:568-578.Hefner, Robert, Virginia D. Nordin, Meda Rebecca andBarbara Oleshansky1974 "A model of sex-role transcendence: role pol-arity and sex discrimination in education."National Institute of Education Contract

    (#NIE-C-74-0144). Ann Arbor:Department ofPsychology, University of Michigan.Hesselbart, Susan1975 "Education, ethnic stereotypes, and questionformat." Sociology and Social Research 59:266-273.Keller, Suzanne1974 "Components of male and female roles: sometheoretical comments." Paper presented at theannual meeting of the American SociologicalAssociation, Montreal, Canada.Kemper, Theodore D.1974 "On the nature and purpose of ascription."American Sociological Review 39(December):844-853.Komarovsky, Mirra1973 "Presidential address: Some problems in roleanalysis." American Sociological Review 38(December):649-662.Kutner, Nancy G. and Donna Brogan1974 "An investigation of sex-related slang vocabu-lary and sex-role orientation among male andfemale university students." Journal of Mar-riage and the Family 36(August):474-484.

    November 1975 JOURNAL OF MARRIAGE AND THE FAMILY 757

  • 8/4/2019 Sex and Sexism - A Comparison of Male and Female Sex Role Attitudes

    16/16

    Lipman-Blumen, Jean1972 "How ideology shapes women's lives." Scien-tific American 226(January):33-42.1973 "Role de-differentiation as a system response tocrisis." Sociological Inquiry43(2):105-129.Martin, Patricia Y. and Marie W. Osmond1975 "The measurement of sex-role attitudes: acomparison of three scaling techniques."Forthcoming.McKee, J. P. and A. C. Sherriffs1959 "Men's and women's beliefs, ideals, and self-concepts." American Journal of Sociology 64:356-363.

    Meehl, P. E.1950 "Configurational scoring." Journal of Consult-ing Psychology 14(2):165-171.Meier, Harold C.1972 "Mother-centeredness and college youth'sattitudes toward social equality for women."Journal of Marriage and the Family 34(Febru-ary):115-121.Mischel, Harriet N.1974 "Sex bias in the evaluation of professionalachievements." Journal of Educational Psy-chology 66:157-166.Miller, Jon, Sanford Labovitz, and Lincoln Fry1974 "Differences in the organizational experiencesof women and men: resources, vested interests,and discrimination." Paper presented at theannual Meeting of the American SociologicalAssociation, Montreal, Canada.Millman, Marcia1971 "Observations on sex role research." Journalof Marriage and the Family 33(November):772-775.

    Morgan, J. N. and J. A. Sonquist1963 "Problems in the analysis of survey data: anda proposal." Journal of the American Statisti-cal Association 58:415-434.

    Nassziger, C. and K. Nassziger1974 "Development of sex role stereotypes." TheFamily Coordinator 23:251-258.Osmond, Marie W.1973 "Family stability and economic dependency."Unpublished Ph. D. dissertation, Florida StateUniversity.Osmond, Marie W. and Patricia Y. Martin1974 "Multivariate model building: an inductive

    approach to a contingency theory of maritalviability." Paper presented in Theory Con-struction Workshop, National Council onFamily Relations, St. Louis, Missouri.Scanzoni, John H.1966 "Family organization and the probability ofdisorganization." Journal of Marriage and theFamily 28(November:407-411.1968 "A social system analysis of dissolved andexisting marriages." Journal of Marriage andthe Family 30(August):452-461.Smith, Kent W.1974 "Forming composite scales and estimatingtheir validity through factor analysis." SocialForces 53(December): 168-180.

    Sonquist, John A.1970 "Multivariate model building: the validationof a search strategy." Ann Arbor:Survey Re-search Center, The University of Michigan.Sonquist, John A., E. L. Baker, and James N. Morgan1971 Searching for Structure (Alias, AID III). AnnArbor:Survey Research Center, The Universityof Michigan.Sonquist, John A. and J. N. Morgan1964 The Detection of Interaction Effects. AnnArbor:SurveyResearch Center, The Universityof Michigan.Stein, Robert1974 "In search of a masculine identity: theoreticalformulations and duplications for women-menrelationships." Paper presented at the annualmeeting of the National Council on FamilyRelations, St. Louis, Missouri.Steinmann, Anne and D. J. Fox1970 "Attitudes toward women's family role amongblack and white undergraduates." The FamilyCoordinator 19:363-368.Sterrett, J. E. and S. R. Bollman1970 "Factors related to adolescent's expectations ofmarital roles." The Family Coordinator 19:353-356.

    Stouffer, S. A.1952 "Atechnique for improving cumulative scales."Public Opinion Quarterly 16(2):273-291.Wang, Marilyn W. and Julian C. Stanley1970 "Differential weighting: a review of methodsand empirical studies." Review of EducationalResearch 40(December):663-703.

    758 JOURNAL OF MARRIAGE AND THE FAMILY November 1975