an investigation of measures of modern and old-fashioned sexism

12
MELANIE A. MORRISON, TODD G. MORRISON, GREGORY A. POPE and BRUNO D. ZUMBO AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM ? (Accepted 8 January, 1998) ABSTRACT. Working from the Principle of Beneficence we make the case that there is a need for adequate measures of sexism in men and women. Swim, Aikin, Hall and Hunter (1995) suggest that modern sexism represents a constellation of beliefs in which individuals: (a) disavow women’s present-day discrimination; (b) reject women’s demands for political and economic power; and (c) disapprove of policies designed to promote gender equality. To assess this construct, Swim et al. (1995) developed the Modern Sexism Scale (MSS), and to date, few studies have assessed its reliability and factor structure. Using data from 187 Canadian participants, confirmatory factor analysis was used to assess the unidimension- ality of the MSS and Swim and associates’ (1995) Old-Fashioned Sexism Scale (OFSS). Results do not support Swim et al.’s conceptualization of modern sexism as a unidimensional construct; however, the OFSS fit a unidimensional model. The implications of these findings are discussed. In presenting his feminist framework for social indicators research, Michalos (1992) states that when considering a moral basis for quality-of-life research we operate from a Principle of Beneficence: One ought to try to act so that one’s actions tend to impartially improve the quality of life. This impartiality implies, in good part, that we strive for a society where sexism is nonexistent. Men and women should have equal opportunities for achieve- ment, receive equal pay for equal work, should be equally free to express their beliefs and opinions, and equally encouraged to work toward personal fulfillment. Over and above the measures of objec- tive circumstances (income, employment, equitable distribution of wealth, and elimination of poverty) there is a need for measures of subjective circumstances. In particular, there is a need for adequate ? Preparation of this article was supported, in part, by a doctoral fellowship to the second author from the Social Sciences and Humanities Research Council of Canada. Social Indicators Research 48: 39–50, 1999. © 1999 Kluwer Academic Publishers. Printed in the Netherlands.

Upload: melanie-a-morrison

Post on 29-Jul-2016

214 views

Category:

Documents


2 download

TRANSCRIPT

MELANIE A. MORRISON, TODD G. MORRISON, GREGORY A. POPE andBRUNO D. ZUMBO

AN INVESTIGATION OF MEASURES OF MODERN ANDOLD-FASHIONED SEXISM?

(Accepted 8 January, 1998)

ABSTRACT. Working from the Principle of Beneficence we make the case thatthere is a need for adequate measures of sexism in men and women. Swim, Aikin,Hall and Hunter (1995) suggest that modern sexism represents a constellation ofbeliefs in which individuals: (a) disavow women’s present-day discrimination; (b)reject women’s demands for political and economic power; and (c) disapprove ofpolicies designed to promote gender equality. To assess this construct, Swim etal. (1995) developed the Modern Sexism Scale (MSS), and to date, few studieshave assessed its reliability and factor structure. Using data from 187 Canadianparticipants, confirmatory factor analysis was used to assess the unidimension-ality of the MSS and Swim and associates’ (1995) Old-Fashioned Sexism Scale(OFSS). Results do not support Swim et al.’s conceptualization of modern sexismas a unidimensional construct; however, the OFSS fit a unidimensional model.The implications of these findings are discussed.

In presenting his feminist framework for social indicators research,Michalos (1992) states that when considering a moral basis forquality-of-life research we operate from a Principle of Beneficence:One ought to try to act so that one’s actions tend toimpartiallyimprove the quality of life. This impartiality implies, in goodpart, that we strive for a society where sexism is nonexistent.Men and women should have equal opportunities for achieve-ment, receive equal pay for equal work, should beequally freetoexpress their beliefs and opinions, and equally encouraged to worktoward personal fulfillment. Over and above the measures of objec-tive circumstances (income, employment, equitable distribution ofwealth, and elimination of poverty) there is a need for measures ofsubjective circumstances. In particular, there is a need for adequate

? Preparation of this article was supported, in part, by a doctoral fellowship tothe second author from the Social Sciences and Humanities Research Council ofCanada.

Social Indicators Research48: 39–50, 1999.© 1999Kluwer Academic Publishers. Printed in the Netherlands.

40 MELANIE A. MORRISON ET AL.

measures of the level of sexist attitudes (i.e., sexism) in men andwomen.

The purposes of this paper are to (a) highlight and build upon thelink between racism, a topic that has been extensively studied, andsexism, and (b) discuss two measures of sexism. After reviewingsome of the literature on the measurement and conceptualization ofsexism, we will investigate the reliability and factor structure of theModern Sexism Scale (MSS; Swim, Aikin, Hall and Hunter, 1995)and the Old-Fashioned Sexism Scale (OFSS; Swim et al., 1995)using a heterogeneous sample of Canadian participants. To ourknowledge, previous research on measures of sexism has focusedalmost exclusively on university/college students.

Old-Fashioned racism is expressed in sentiments such as whitesupremacy, black inferiority, and racial segregation. According toMcConahay, Hardee and Batts (1981), such blatant expressions ofracial hatred have become unacceptable in mainstream society. Asa result, old-fashioned racist beliefs have been replaced by modernracism, in which racist sentiments are expressed in a more subtleand indirect form. Duckitt (1992) states that modern racism is rootedin the assumptions that African Americans violate cherished valuessuch as self-reliance, discipline, and obedience, and are makingillegitimate demands for greater economic and political power.

Due to the shared minority status of women and African Ameri-cans, racism and sexism are often represented as parallel phenomena(e.g., Reid, 1988; Smith and Stewart, 1983). For example, Smith andStewart (1983) assert that the content of racist and sexist stereotypesshare many similarities (i.e., women and African Americans areboth portrayed as “intellectually inferior, lacking in ambition anddrive, emotional, dependent, and childlike” [p. 20]). Reid (1988)also underscores the similarities between racism and sexism. Specif-ically, the author reports that “both groups have had to deal withthe expectation and assumption that white men are better suitedto . . . supervisory and managerial positions” (p. 204). Also, womenand African Americans have been excluded from prestigious com-munity organizations and restrictions have been placed on theiradmittance to post-secondary educational institutions (Reid, 1988).Finally, Brewer (1988) emphasizes that racism and sexism mayresult, in part, from similar cognitive processes. The author contends

AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM 41

that race and sex are readily discernible physical stimuli and, thus,initiate automatic categorization processes along ingroup/outgroupdimensions.

In accordance with earlier theorizing on modern racism, Swimet al. (1995) suggest that modern sexists disavow women’s present-day discrimination, reject their demands for political and economicparity, and disapprove of policies designed to promote gender equal-ity. The authors also identify other ways in which modern sexismmay be viewed as an analogue to modern racism. Specifically, tradi-tional beliefs about characterological differences between men andwomen have been supplanted by more subtle manifestations of prej-udice. Furthermore, although national opinion polls suggest thatmany individuals endorse parity between the sexes, concomitantstructural and behavioral changes indicative of gender equality havenot occurred. For example, research indicates that: (a) women work-ing full-time earn approximately 40% less in salary per annum thanmen (Wilson, 1991); (b) women remain disproportionately repre-sented in low-paying occupations which provide minimal opportu-nity for career advancement (Armstrong and Armstrong, 1994); and(c) women, irrespective of their employment status, spend more timethan men performing child care and domestic activities (Biernat andWortman, 1991; Douhitt, 1989).

Swim et al. (1995) developed the Modern Sexism Scale (MSS)to measure covert prejudice against women and the Old-FashionedSexism Scale (OFSS) to measure traditional sexist beliefs aboutwomen. The authors reported that the MSS is reliable (i.e., alphacoefficients for the scale ranged from 0.75 to 0.84). In addition,the MSS appears to possess construct validity. For example, aspredicted, participants high in modern sexism were more likelythan those low in modern sexism to engage in gender-biased votingbehavior. Neither participants’ gender nor their scores on the OFSSpredicted voting preference. Finally, the authors conducted a prin-cipal components factor analysis to investigate whether modern andold-fashioned sexism are conceptually distinct. A two-factor solu-tion was produced, with items from the MSS loading on one factor,and items from the OFSS loading on the other.

These results are encouraging. However, further investigationof the psychometric properties of the MSS and OFSS is impera-

42 MELANIE A. MORRISON ET AL.

tive if researchers are to gauge accurately the usefulness of thesemeasures. To date, there are only two published accounts of thepsychometric properties of the MSS (Campbell, Schellenberg andSenn, 1997; Swim et al., 1995). However, the usefulness of thesestudies is limited by their reliance on university students. Again,the purpose of the present study is to investigate the reliability andfactor structure of the MSS and OFSS using a heterogeneous sampleof Canadian participants.

METHOD

Participants

A sample of 187 Canadian participants (88 males; 99 females)ranging in age from 16 to 58 years (M = 26.4, SD = 10.1) wasused. Chain referral sampling was used to target members of thegeneral population (i.e., acquaintances of the first two authors wereinstructed to provide copies of the questionnaire to their friends andrelatives). Ethics regulations were satisfied by use of a cover sheetwhich clearly indicated that participation was voluntary and that allresponses would be anonymous and confidential.

Instruments

Modern Sexism Scale (MSS). The MSS measures subtle manifesta-tions of sexist attitudes toward women. It consists of eight items anduses a five-point Likert-type scale (1 = strongly agree; 5 = stronglydisagree). Scale scores can range from 8 to 40, with higher scoresdenoting greater levels of modern sexism.

Old-Fashioned Sexism Scale (OFSS). The OFSS measures overtsexist attitudes toward women. It contains five items and uses a five-point Likert-type scale (1 = strongly agree; 5 = strongly disagree).Scale scores can range from 5 to 25, with higher scores representinggreater levels of old-fashioned sexism. Items on the MSS and OFSSare listed in the Appendix.

AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM 43

Factor Models

Single factor models were used to separately test the MSS andOFSS. Specifically, working from the covariance matrices, maxi-mum likelihood estimation via LISREL 8.14 (Joreskog and Sorbom,1995) was used. Maximum likelihood methods are sensitive tomultivariate normality; therefore, the data were screened for multi-variate outliers by examining Mahalanobis’ distance. Eight outlyingcases were identified, and subsequently deleted from the analyses.

Before one fits a two-factor model, it is necessary to confirm thateach separate measurement model adequately fits (Bollen, 1989;Joreskog and Sorbom, 1989). If one does not initially test for thefit of one-factor models, a two-factor model that provides poor fitwould be difficult to interpret. Thus, in the present study, the extentto which the MSS and OFSS separately fit a one-factor model wasexamined.

Goodness of fit. In accordance with Marsh, Balla, and McDonald’s(1988) recommendations, multiple indices were used to assess thefit of the model to the data. The following indices were examined:(a) the goodness-of-fit index (GFI; Joreskog and Sorbom, 1989);(b) the goodness of fit which adjusts for the number of degrees offreedom (AGFI; Joreskog and Sorbom, 1989); (c) the Tucker-Lewisindex (TLI; Tucker and Lewis, 1973) which is also referred to asthe Non-Normed Fit Index (NNFI; Bentler and Bonett, 1980); (d)the Normed Fit Index (NFI; Bentler and Bonett; 1980); (e) the rootmean square residual (RMSR); (f) the root mean square of approx-imation (RMSEA; Steiger and Lind; 1980); and (g) the chi-squaretest of fit between the predicted and obtained covariance matrices. Injudging the adequacy of the fit, the larger the GFI and AGFI values(i.e., values above 0.90), the better the fit (Bollen, 1989). A non-significant chi-square value is an additional indication of model fit,as well as RMSR and RMSEA values of less than or equal to 0.1 foracceptable fit, and 0.05 or lower for good fit (Thompson, Coovert,Richards, Johnson and Cattarin, 1995). Finally, the TLI (NNFI) andNFI should be above 0.90 (i.e., lower values suggest the model canbe substantially improved).

It should be noted that for confirmatory factor analysis therequired sample size is determined, in part, by the number of para-

44 MELANIE A. MORRISON ET AL.

TABLE I

Means, standard deviations, and reliabilities for the ModernSexism and Old-Fashioned Sexism Scales

Modern Sexism M SD Item

Scale items reliability

#1 2.50 1.11 0.34

#2 2.03 0.91 0.47

#3 2.64 1.12 0.25

#4 2.54 1.06 0.36

#5 2.60 1.19 0.40

#6 2.13 1.20 0.12

#7 2.10 0.92 0.55

#8 2.64 1.00 0.27

Scale 19.18 5.46

α = 0.79

Old-Fashioned Sexism M SD Item

Scale items reliability

#1 1.41 0.71 0.56

#2 1.98 1.10 0.12

#3 2.28 1.10 0.23

#4 1.66 0.95 0.53

#5 1.47 0.76 0.39

Scale 8.80 3.10

α = 0.68

Note. Item reliabilities, defined as the squared multiple corre-lations, were obtained from the LISREL output (Dillon andGoldstein, 1984).

meters being estimated in the model. In the present study, thelargest number of parameters being estimated is 16 (i.e., in fittinga one-factor model for the MSS, there are 8 parameters in both theLAMBDA-X and an additional 8 in the THETA-DELTA matrices).Therefore, the estimated parameters to sample size ratio exceeds11:1, which is adequate for our analyses (Dillon and Goldstein,1984).

AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM 45

TABLE II

Correlation matrix of the Modern Sexism Scale

Item 1 Item 2 Item 3 Item 4 Item 5 Item 6 Item 7 Item 8

Item 1 1.00

Item 2 0.31 1.00

Item 3 0.10 0.45 1.00

Item 4 0.52 0.34 0.23 1.00

Item 5 0.26 0.49 0.43 0.41 1.00

Item 6 –0.002 0.40 0.41 0.004 0.36 1.00

Item 7 0.63 0.50 0.30 0.48 0.37 0.37 1.00

Item 8 0.26 0.24 0.24 0.32 0.38 0.38 0.22 1.00

RESULTS

Item and scale means, standard deviations, and reliability estimatesfor the MSS and OFSS are listed in Table 1. Tables 2 and 3 listthe correlation matrices for the modern and old-fashioned sexismscales, respectively. The levels of internal consistency obtained inthe present study (for MSS = 0.79; for OFSS = 0.68) are comparableto those reported by Campbell et al. (1997) and Swim et al. (1995).It should be noted, however, that item #6 of the MSS and item #2 ofthe OFSS had particularly poor reliabilities (see Table 1). The scalemeans clearly indicate that, as a whole, the sample was below thescale mid-point on both modern sexism (95% CI = 18.4, 20.0; scalemid-point = 24) and old-fashioned sexism (95% CI = 8.4, 9.3; scalemid-point = 15).

MSS. The obtained fit statistics are: GFI = 0.83; AGFI = 0.69;TLI (NNFI) = 0.67; NFI = 0.74; RMSR = 0.12; and RMSEA =0.16. Furthermore, the chi-square test was statistically significant,χ2 (20) = 120.2,p < 0.01. These indicators clearly demonstratethat a unidimensional model does not provide an adequate fit for thedata.

Because of the lack of unidimensional fit, we conducted anexploratory factor analysis (maximum likelihood estimation withdirect oblimin rotation) with a two factor solution. This model fit the

46 MELANIE A. MORRISON ET AL.

TABLE III

Correlation matrix of the Old-Fashioned Sexism Scale

Item 1 Item 2 Item 3 Item 4 Item 5

Item 1 1.00

Item 2 0.32 1.00

Item 3 0.34 0.07 1.00

Item 4 0.50 0.25 0.48 1.00

Item 5 0.52 0.20 0.18 0.45 1.00

data,χ2 (13) = 13.8,p > 0.39, with nearly all of the standardizedresiduals less than 0.05 in magnitude. Factor one was comprised ofitems #1, #4, and #7 and appears to represent modern sexist attitudestoward present day discrimination. The second factor was comprisedof items #2, #3, #5, and #6 and appears to represent negative atti-tudes toward policies designed to promote gender equality. The twofactors were correlated at 0.36. Item #8 was dropped from the scaledue to the fact that it did not load on either factor.

OFSS. The obtained fit statistics are: GFI = 0.95; AGFI = 0.86;TLI (NNFI) = 0.81; NFI = 0.89, RMSR = 0.056; and RMSEA =0.14. The chi-square test was statistically significant,χ2 = (5) =22.7,p < 0.01. In this case, the fit indices are less consistent thanin the MSS with all indicators except the RMSEA and chi-squaretest suggesting that a unidimensional model fits. Therefore, whilenoting the two discrepant indices, it may be concluded that the OFSSreflects a unidimensional construct and the item responses may belegitimately summed for a composite score – large scores representold-fashioned sexist attitudes toward women.

DISCUSSION

The analysis does not support Swim et al.’s (1995) conceptualiza-tion of the MSS as a unidimensional measure of modern sexism.However, the analysis provides evidence that the OFSS is a unidi-mensional measure of old-fashioned sexism. Guilford (1954, ascited in Briggs and Cheek, 1986) emphasizes the importance of a

AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM 47

unidimensional scale and states that “any test that measures morethan one common factor to a substantial degree . . . yields scores thatare psychologically ambiguous and difficult to interpret” (p. 356).The apparent multidimensionality of the MSS suggests that oneor more of the scale’s items may delineate aspects of sexism thatdiffer qualitatively from one another. Our revised MSS is comprisedof two correlated sub-scales reflecting: (a) modern sexist atti-tudes toward present day discrimination, and (b) negative attitudestoward policies designed to promote gender equality. Our resultssuggest that composite scores on these two sub-scales are moreclearly interpreted than Swim et al.’s 8-item composite of modernsexism.

Many of the studies exploring sexism, in both its old-fashionedand modern forms, use university students as subjects. Relianceon this sample in previous assessments of modern sexism mayhave contributed to its unidimensional factor structure. Accordingto Rubin and Peplau (1975), the liberal milieu characteristic of mostuniversities may enhance students’ awareness of injustice, therebyincreasing their sensitivity to issues such as racism and sexism.University students’ heightened sensitivity to racism and sexismmay result in them adopting a more inclusive interpretation of theseforms of prejudice. In particular, university samples may interpretall items on the MSS as falling under the rubric of sexism, whereas,non-student samples may endorse a more restrictive view (i.e., theymay not believe that all items on the MSS constitute examples ofsexism).

In conclusion, it is recommended that the psychometric proper-ties of the MSS continue to be examined with non-student popu-lations. Specifically, the exact nature of its dimensionality and itsconceptual distinction from old-fashioned sexism must be deter-mined. This type of research will enable social scientists to gaugethe usefulness of this measure and, subsequently, to assess thepervasiveness of modern sexism.

48 MELANIE A. MORRISON ET AL.

APPENDIX

The Modern Sexism Scale (MSS)

1. Discrimination against women is no longer a problem inCanada.∗

2. Women often miss out on good jobs due to sexual discrimina-tion.

3. It is rare to see women treated in a sexist manner on television.∗4. On average, people in our society treat husbands and wives

equally.∗5. Society has reached the point where women and men have equal

opportunities for achievement.∗6. It is easy to understand the anger of women’s groups in Canada.7. It is easy to understand why women’s groups are still concerned

about societal limitations of women’s opportunities.8. Over the past few years, the government and news media have

been showing more concern about the treatment of women thanis warranted by women’s actual experiences.∗

Old-Fashioned Sexism Scale (OFSS)

1. Women are generally not as smart as men.∗2. I would be equally comfortable having a woman as a boss as a

man.3. It is more important to encourage boys than to encourage girls

to participate in athletics.∗4. Women are just as capable of thinking logically as men.5. When both parents are employed and their child gets sick at

school, the school should call the mother rather than the father.∗

Note. Items with an asterisk require reverse scoring. Questions orig-inally pertaining to the United States were modified to pertain toCanada.

REFERENCES

Armstrong, P. and H. Armstrong: 1994, The Double Ghetto: Canadian Womenand Their Segregated Work (McClelland and Stewart, Toronto, ON).

AN INVESTIGATION OF MEASURES OF MODERN AND OLD-FASHIONED SEXISM 49

Bentler, P. M. and D. G. Bonett: 1980, ‘Significance tests and goodness of fit inthe analysis of covariance structures’, Psychological Bulletin 88, pp. 588–606.

Biernat, M. and C. B. Wortman: 1991, ‘Sharing of home responsibilities betweenprofessionally employed women and their husbands’, Journal of Personality andSocial Psychology 60, pp. 844–860.

Bollen, K. A.: 1989, Structural Equations with Latent Variables (John Wiley, NewYork, NY).

Brewer, M.: 1988, ‘A dual process model of impression formation’, in T. K. Srulland R. S. Wyer, Jr. (eds.), Advances in Social Cognition (Erlbaum, Hillsdale,NJ), pp. 1–36.

Briggs, S. R. and J. M. Cheek: 1986, ‘The role of factor analysis in thedevelopment and evaluation of personality scales’, Journal of Personality 54,pp. 106–148.

Campbell, B. E., G. Schellenberg and C. Y. Senn: 1997, ‘Evaluating measures ofcontemporary sexism’, Psychology of Women Quarterly 21, pp. 89–102.

Dillon, W. R. and M. Goldstein: 1984, Multivariate Analysis: Methods andApplications (John Wiley, New York, NY).

Douhitt, R. A.: 1989, ‘The division of labour within the home: Have gender roleschanged?’ Sex Roles 20, pp. 693–704.

Duckitt, J.: 1992, The Social Psychology of Prejudice (Praeger Publishers, NewYork, NY).

Joreskog, K. G. and D. Sorbom: 1989, LISREL 7 User’s Reference Manual(Scientific Software, Mooresville, IN).

Joreskog, K. G. and D. Sorbom: 1995, LISREL 8.14 (Scientific Software,Chicago, IL).

Marsh, H. W., J. R. Balla and R. P. McDonald: 1988, ‘Goodness-of-fit indexes inconfirmatory factor analysis: The effect of sample size’, Psychological Bulletin103, pp. 391–410.

McConahay, J. B.: 1986, ‘Modern racism, ambivalence, and the modern racismscale’, in J. F. Dovidio and S. L. Gaertner (eds.), Prejudice, Discrimination, andRacism (Academic Press, Orlando, FL), pp. 91–125.

McConahay, J. B., B. B. Hardee and V. Batts: 1981, ‘Has racism declined inAmerica? It depends on who is asking and what is asked’, Journal of ConflictResolution 25, pp. 563–579.

Michalos, A. C.: 1992, Social Indicators Research From A Feminist Perspec-tive. Seminar on Social Indicators, Instituto de la Mujer (Women’s Institute),Ministerio de Asuntos Sociales, Madrid, Spain.

Reid, P. T.: 1988, ‘Racism and sexism: Comparisons and conflicts’, in P. A. Katzand D. A. Taylor (eds.), Eliminating Racism: Profiles in Controversy (PlenumPress, New York, NY), pp. 203–221.

Rubin, Z. and L. A. Peplau: 1975, ‘Who believes in a just world?’ Journal ofSocial Issues 31, pp. 65–89.

Smith, A. and A. J. Stewart: 1983, ‘Approaches to studying racism and sexism inblack women’s lives’, Journal of Social Issues 39, pp. 1–15.

50 MELANIE A. MORRISON ET AL.

Steiger, J. H. and J. C. Lind: 1980, May, ‘Statistically-based tests for the numberof common factors’, Paper presented at the Annual Meeting of the PsychometricSociety, Iowa City, IA.

Swim, J. K., K. J. Aikin, W. S. Hall and B. A. Hunter: 1995, ‘Sexism andracism: Old-fashioned and modern prejudices’, Journal of Personality andSocial Psychology 68, pp. 199–214.

Thompson, J. K., M. D. Coovert, K. J. Richards, S. Johnson and J. Cattarin:1995, ‘Development of body image, eating disturbance, and general psycho-logical functioning in female adolescents: Covariance structural modelingand longitudinal investigations’, International Journal of Eating Disorders 18,pp. 221–236.

Tucker, L. R. and C. Lewis: 1973, ‘A reliability coefficient for maximumlikelihood factor analysis’, Psychometrika 38, pp. 1–10.

Wilson, S. J.: 1991, Women, Families, and Work (McGraw-Hill Ryerson, Whitby,ON).

Melanie A. MorrisonDepartment of PsychologyUniversity of Northern British ColumbiaPrince George, BCCanadaV2N 4M9Phone: (250) 960-5710E-mail: [email protected]