University of Pennsylvania
We developed a questionnaire that measured policy attitudes toward gender equality and separation of males and females, beliefs about differences in the suitability of males and females for various roles, and general moral principles concerning gender equality and inequality. We gave the questionnaire to 109 U.S.\ students and 89 foreign students (some from immigrant families), mostly from India, but some from the Middle East. The questionnaire was internally consistent: policy beliefs were predictable from factual beliefs and general moral principles. An interaction predicted from the syllogistic structure of the questionnaire was found. It also distinguished males and females, foreign and domestic subjects, more acculturated and less acculturated foreign subjects, and Islamic and Hindu foreign subjects. These results indicate that the measures are valid. Factual beliefs were correlated with moral principles, perhaps because both lead to the same policy recommendations and people want to be consistent.
We report here the development of a questionnaire that takes a new approach to the measurement of gender-role attitudes. It is based on the idea that attitudes, in principle, involve specific prescriptive beliefs about policies, for example, beliefs about whether men and women should share housework equally, or whether boys should be encouraged in mathematics more than girls. Each of these policy attitudes can be derived logically from factual beliefs and general prescriptive, moral, principles (Hare, 1952). For example, the specific policy attitude that boys should be encouraged more in mathematics might derive from a factual belief that boys are better in mathematics and from a general moral principle that people should be encouraged to do what they are likely to be good at. The questionnaire allows us to examine the internal consistency of policy attitudes, moral values and beliefs.
The questionnaire is also designed to be useful in a wide variety of cultures, and we provide validation data based on student subjects from several cultural backgrounds. Gender-role attitudes are important around the world. For example, when women receive more education and more power within the household, their desired fertility decreases, and this helps to reduce the birth rate in many countries that suffer from excessive population growth (Bongaarts, 1994; Paine et al., 1992). Abuse of women and girls is also a world wide problem that may be related to the kinds of gender-role attitudes we measure.
The kinds of factual beliefs we assess are those discussed in the literature on gender differences (e.g., Maccoby & Jacklin, 1974; Kimball, 1989). We do not assume that answers to these items are correct or incorrect.
Our questionnaire measures policy attitudes rather than gender-typed personality traits. Thus, it is not in the tradition of the Terman-Miles M-F Test (1936), the M-F Scale of the Strong Vocational Interest Blank (1943), the Mf scale of the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway and McKinley, 1943), and the Femininity Scale of Gough's California Psychological Inventory (CPI, 1964). Nor does it follow more recent tests that have dropped the assumption that masculinity and femininity were opposite ends of a continuum (Bem, 1974, 1975, 1981; Spence, Helmreich & Stapp, 1974). The beliefs section of our questionnaire is closer to tests of stereotypes of women (Belk & Snell, 1986), although we are concerned specifically with abilities and dispositions that have implications for the specific policies we examine. Policy attitudes, however, must be measured in their own right (as done by Hartman & Hartman, 1983, and Spence & Helmreich, 1972a). They need be inferred from beliefs, personality traits, or stereotypes. (Of course, personality traits and attitudes are sometimes found to be related, as found by Spence, 1993, and to a lesser degree by Archer, 1989). We know of no previous studies that measured gender-role beliefs and policy attitudes separately.
Many of our items may be arranged in the form of syllogisms, such as:
As the examples illustrate, the moral principles are, in a sense, general versions of the policy items. We thus measure general moral prescriptions, specific prescriptions (policies), and factual beliefs required for the policies. We expect that subjects will hold reasonably consistent attitudes for each type of item, so that the items of each of the three types can be used as a scale. Our interest is thus not so much in examining within-syllogism consistency (as done by McGuire, 1960) but rather in the use of the syllogism form as a template for generating items. Because both premises must be accepted in order for the conclusion to be expected, we hypothesize a statistical interaction in predicting policy attitudes from moral principles and factual beliefs: the contribution of each predictor should be greater when the other term is high.
In addition to examining the internal consistency of our questionnaire, we validate it in two ways. First, we use it to predict gender differences, on the ground that women ought to have more liberal gender-role attitudes than men (i.e., attitudes that are more pro-women). Second, we use it to predict differences among cultures thought to have different gender-role attitudes: Indian Hindu, Islamic, and U.S. Judeo-Christian. The Indian and Islamic samples were tested in the U.S., so we also look at acculturation to U.S. culture as a predictor. These comparisons of genders and cultures also allow us to ask the substantive question of whether group differences in policy attitudes are mediated by differences in moral values, beliefs, or both. Effects of sex and culture on gender-role attitudes have been found by Lottes and Kuriloff (1992), Fine-Davis (1989), Belk et al. (1989), and Miller (1984).
The questionnaire was developed in two steps. In the first step, a first draft was administered to 59 subjects. The final draft was a revision made on the basis of analysis of the first draft. All analyses of the first draft alone agree with identical analyses of the final draft, so we report only the data from that draft. (Most of the first-draft subjects were from the U.S., so the effects of cultural differences were not tested.)
Subjects were 109 American students, solicited by advertising and paid $6/hour for completing this questionnaire and others, and 89 ``foreign'' subjects (including children of immigrant families), also mostly students. There were 96 males and 102 females overall, with both sexes roughly equally represented in both samples (U.S. and foreign). The median age was 20 (range, 16 to 46). Among the foreign students, 59 were Hindu, 23 Moslem (Islamic), and 7 some other religion. Of the Hindus, 30 were from India, 26 born in the U.S. Of the Moslems, 10 were from the Middle East, 8 from Pakistan, and 4 from the U.S.
The final questionnaire, contained in the Appendix, contains items divided into three sections - policy items, belief items, and moral items. Subjects rated the extent to which they agreed with each of the items on a scale of 1 to 5, where 1 was described as ``strongly agree,'' 2 was ``moderately agree,'' 3 was ``neutral,'' 4 was ``moderately disagree,'' and 5 was ``strongly disagree.''
Most policy items were of three types, all concerned with distribution of goods or burdens between males and females: equality items held that males and females should be treated the same or should do the same things; status-quo items supported current inequality; reversal items supported actively reversing present inequality. Some policy items did not concern distribution but, rather, relations between males and females or rules of etiquette.
There were, in all, 75 policy items, 22 belief items and 9 moral items, making a total of 106 items. In addition, all subjects were asked a series of demographic questions: age, gender, level of education, profession, religion, strength of religious beliefs, country of birth and ethnic/racial background.
To measure acculturation, foreign subjects were also asked about the number of friends they had from their own country vs. the U.S., their fluency in English, the number of publications they read relating to their country of origin vs. the U.S., and their membership in organizations relating to their home country vs.\ other themes. We defined acculturation as an equally weighted composite of the z-scores of the proportion of friends that were American, the proportion of publications read that were American, the difference between number of home-country and American organizations, and fluency in the native language. (The difference was used for organizations because many subjects belonged to none.)
Factor analyses supported the formation of three subscales of the policy items: Equality consisting of equality and status-quo items (the latter with negative loadings); Reversal consisted of the reversal items; and Separate consisted of the items concerned with separation of the sexes (some reverse scored). Most of the belief items were used to form a single Belief scale, and likewise for the Moral items. The items used (and the direction of their scoring, when different within a scale) are shown in the Appendix with letters indicating the scales in question. Reliabilities (alpha) of the scales were, respectively: Equal, .964; Reversal, .774; Separate, .685; Belief, .907; Moral, .734. Naturally, the scales with more items were more reliable; Separate used only 6 items, and Moral only 8.
All of the scales except for Reversal correlated strongly with each other, suggesting a single general source of individual differences in gender-role attitudes. Subjects who favor equal treatment are also those who oppose separation and who believe that the sexes are equally qualified. Table 1 shows these correlations. All the correlations among the scales in the table are significant at , but it is apparent that those involving the Reversal scale are much lower than the others. (The correlation between Reversal and Moral is the highest of these because the Moral scale contained some items that favored reversing the usual gender roles.)
Correlations among scales, and of the scales with Gender (male=1, female=0), Sample (foreign vs. U.S.), Acculturation (within the foreign sample), and Islam (vs. Hindu, within the foreign sample). All variables are defined here so that lower numbers represent more liberal beliefs or ends of dimensions that are expected to correlate with such beliefs (female, U.S., more acculturated, Hindu).
The only subjects who strongly favored unequal treatment (in the policy questions) were those who believed both that males and females differed (in the belief items) and that such differences justified unequal treatment (in the moral items). A regression of Equal on Belief, Moral, and the interaction of Belief and Moral yielded a strong interaction term (overall , for the interaction term). Moral was also a significant predictor (), but Belief was not significant unless the interaction term was dropped. (Separate analyses indicated that the predictive power of the Belief and Moral scales was not peculiar to the item. In particular, we compared individual syllogisms to control syllogisms consisting of items from the same scales but with the conclusion terms, the policies, mismatched with the premises. The real syllogisms did not differ from the controls in the extent to which the premises predicted the conclusions.)
The results were similar for Separate. The only subjects who favored separation of the sexes were those who believed both that males and females differed and that unequal treatment was justified. Again, in the regression of Separate on Belief, Moral, and their interaction yielded a strong interaction (, for the interaction). Again, Moral was also a significant predictor (), but Belief was not.
Females and males differed on all scales as shown by the correlations with gender in Table 1 ( for all). Females favored equality, favored efforts to reverse roles, opposed separation, and believed that males and females were more equal in what they could do. The policy difference in Equality and Separate between males and females is apparently mediated by the differences in Moral and Belief: when gender was added to the regression of Equality on Moral, Belief, and their interaction, the Gender effect was not significant, and the same result obtained for prediction of Separate.
The American and foreign samples also differed, with Americans favoring equality, opposing separation, and believing that the sexes are more equal, as shown in Table 1 (Sample) ( for these correlations). But the two groups did not differ in attitude toward Reversals. The group differences in Equality and Separation remained significant after including Moral, Belief, and their interaction in a multiple regression for each (although the coefficients decreased). These policy differences are therefore not, apparently, fully mediated by differences in Moral and Belief.
Subjects more acculturated to the U.S. (according to the Acculturation variable) were more in favor or equality, against separation, and inclined to believe that the sexes were equal ( for these correlations) but did not differ in their attitude toward reversals. The correlations of Acculturation with Equality and Separate were, however, apparently mediated by effects of Acculturation on Belief and Moral, since the Acculturation coefficient became nonsignificant when Belief, Moral, and their interaction were included in a regression. (Although women were more acculturated than men [, ], the correlation between Acculturation and the policy measures remained significant when gender was included, as did the effect of gender, and Gender and Acculturation did not interact.)
For the foreign sample, we defined a variable Islam, which was 1 for Islamic subjects, 0 for Hindu subjects, and missing otherwise. Islam was uncorrelated with Acculturation or Gender. As shown in Table 1, Islamic subjects were opposed to equality, in favor of separation, and inclined to believe that the sexes were unequal. Islam remained strongly related to Equality and Separation even when Belief, Moral, and their interaction were included in regressions, so the effect of Islam on policy attitudes is not mediated by its relation to Moral and Belief. Within the Moslem group, all these effect were correlated strongly with strength of religious belief, although this did not hold for the Hindu group.
Although the effects of Acculturation on Equality and Separation was little reduced when Islamic subjects were excluded (and were still significant at one tailed, despite a small sample), the effect of Sample (foreign vs. U.S.) clearly and completely disappeared. It is apparent, then, that the effects of Sample, and the fact that these effect are not mediated through Moral and Belief, are completely due to the inclusion of the Islamic subjects in the foreign sample.
A few items have been left out of all analyses because they don't fit neatly into scales. However, we include them in the Appendix because they may be useful for other purposes.
The questionnaire seems reliable and valid. It would make sense to use parts of it in subsequent studies. We grouped the items together so that it could easily be used in this way. For example, the Equality scale could be used on its own, or the Belief scale. Each is sufficiently reliable.
The strong interaction between Belief and Moral in predicting the two policy scales, Equality and Separate, supports the logical structure of the questionnaire. Syllogisms ought to involve interactions, since both premises must be true for the conclusion to follow.
Further validation of the questionnaire comes from the correlations of its scales with each other, with gender, with acculturation, and with religion (Moslem vs. Hindu). Although the scales also distinguish U.S. vs. foreign subjects in general, most of this difference is because of the presence of the Moslem subjects in the foreign sample. However, we should note that most of the foreign sample were fairly well acculturated to the U.S., so it seems likely that South Asians tested in their home countries would be more opposed to equality, etc.
One finding of interest is the correlation between Belief and Moral (.575), which remained strong and significant when other variables were included in the regression in all possible combinations (Gender, Acculturation, Islam, Sample). The moral statements were not written to refer to any specific facts; they were contingent on factual assumptions. Although the attitudes measured by these two scales can be used to justify the same policies, there is no logical connection between them. The same kind of result was found by Ellsworth and Ross (1983): beliefs in the morality of the death penalty correlated with factual beliefs about its deterrent value. People seem to adjust their factual and moral beliefs so that they point to the same conclusions, even though they need not.
Archer, J. (1989). The relationship between gender-role measures: A review. British Journal of Social Psychology 28, 173-184.
Belk, S. S., & Snell, W. E. (1986). Beliefs about women: Components and correlates. Personality and Social Psychology Bulletin, Vol. 12(4), 403-413.
Belk, S. S., Snell, W. E., Holtzman, W. H., & Hernandez-Sanchez, J. E. (1989). The impact of ethnicity, nationality, counseling orientation, and mental health standards on stereotypic beliefs about women: A pilot study. Sex Roles, 21(9-10), 671-695.
Bem, S. L. (1974). The measurement of psychological androgyny. Journal of Consulting and Clinical Psychology, 42, 155-162.
Bem, S. L. (1975). Sex role adaptability: One consequence of psychological androgyny. Journal of Personality and Social Psychology, 31, 634-643.
Bem, S. L. (1981). Gender schema theory: A cognitive account of sex typing. Psychological Review, 88, 354-364.
Bongaarts, J. (1994). Population policy options in the developing world. Science, 263, 771-776.
Ellsworth, P. C., & Ross, L. (1983). Public opinion and capital punishment: A close examination of the views of abolitionists and retentionists. Crime and Delinquency, 29, 116-169.
Fine-Davis, M. (1989). Attitudes toward the role of women as part of a larger belief system. Political Psychology, 10, 287-308.
Hare, R. M. (1952). The language of morals. Oxford: Oxford University Press (Clarendon Press).
Hartman, M. & Hartman, H. (1983) Sex-role attitudes of Mormons vs. non-Mormons in Utah. Journal of Marriage and the Family, 45, 897-902.
Kimball, M. M. (1989). A new perspective on women's math achievement. Psychological Bulletin, 105, 198-214.
Lottes, I. L., & Kuriloff, P. J. (1992). The effects of gender, race, religion, and political orientation on the sex role attitudes of college freshmen. Adolescence, 27, 675-688.
Maccoby, E. E., & Jacklin, C. N. (1974). The psychology of sex differences. Stanford, CA: Stanford University Press.
McGuire, W. J. (1960). A syllogistic analysis of cognitive relationships. In M. J. Rosenberg, C. I. Hovland, W. J. McGuire, R. P. Abelson, & J. W. Brehm (Eds.), Attitude organization and change (pp. 65-111). New Haven: Yale University Press.
Miller, K. A. (1984). The effects of industrialization towards men's attitudes toward the extended family and women's rights: A cross cultural study. Journal of Marriage and the Family, 46, 153-160.
Paine, P., Pasquali, L., Wright, M. D., & Rosa, S. C. (1992). Explaining fertility in urban-dwelling Brazilian women: Importance of traditional sex-role attitudes and religious orthodoxy. Psychological Reports, 71, 559-567.
Spence, J. T. (1993). Gender-related traits and gender ideology: Evidence for a multifactorial theory. Journal of Personality and Social Psychology 64, 624-635.
Spence, J. T., & Helmreich, R. (1972). The Attitudes toward Women Scale: An objective instrument to measure attitudes toward the rights and roles of women in contemporary society. Catalog of Selected Documents in Psychology, 2, 66-67.
Spence, J. T., Helmreich, R., & Stapp, J. (1974). The Personal Attributes Questionnaire: A measure of sex role stereotypes and masculinity-femininity. Catalog of Selected Documents in Psychology, 4, 43-44.
These are the items used in the final questionnaire. Scales (not included in the questionnaire) are indicated at the beginning of each item: E for Equality, R for Reversal, S for Separate, B for Belief, and M for Moral.
(E+): The responsibility of taking care of infants should be equally divided between parents, irrespective of their gender.
(E-): The wife should have primary responsibility for child care.
(E-): Boys should be encouraged to do things that boys usually do and girls should be encouraged to do things that girls usually do.
(E+): Boys and girls should be encouraged to do the same things.
(R): Boys should be encouraged to do things that girls usually do and girls should be encouraged to do things that boys usually do.
(R): There should be special quotas for women in government jobs.
(R): Schoolteachers should devote more effort to encouraging girls in science and math than they devote to boys.
(E+): Schoolteachers should encourage girls and boys equally in science and math.
(E-): Schoolteachers should devote more effort to encouraging boys in science and math than they devote to girls.
(E-): Schoolteachers should devote more effort to encouraging girls in English, social studies, and languages than they devote to boys.
(E+): Schoolteachers should treat encourage girls and boys equally in English, social studies, and languages.
(R): Schoolteachers should devote more effort to encouraging boys in English, social studies, and languages than they devote to girls.
Laws regarding sexual harassment and rape should be made much more favorable to women.
Because most pornography is particularly demeaning to women, films and magazines depicting women as sex objects should be regulated by law.
(E+): There should be no differential dress codes for men and women at workplaces, for example, codes saying that women must wear skirts.
(R): Girls should be given priority in using gyms, and other sports facilities in schools.
(E+): Girls and boys should be given equal access to gyms and other sports facilities in schools.
(E-): Boys should be given priority in using gyms, and other sports facilities in schools.
(R): Girls should be given priority in using computers in schools.
(E+): Husbands and wives should share equally in housework such as cooking, washing dishes, and housecleaning.
(E+): The husband and wife should have equal responsibility to contribute to the family income by working.
(E-): The husband should have primary responsibility for contributing to the family income by working.
(e-): Women in the military should be kept out of certain combat roles.
The Catholic Church should allow women to be priests.
The Muslim religion should allow women to go to work without their faces covered.
(E+): Families should spend just as much money on the education of daughters as on the education of sons.
(R): Families should spend more money on the education of daughters as on the education of sons.
(E-): Families should spend less money on the education of daughters as on the education of sons.
(E+): Families should provide equal medical care to daughters and sons.
(R): Families should provide more medical care to daughters than to sons.
(E-): Families should provide less medical care to daughters than to sons.
(E-): Governments and hospitals should provide more health care facilities for men than for women.
(E+): Governments and hospitals should provide equal health care facilities for men and women.
(R): Governments and hospitals should provide more health care facilities for women than for men.
(E+): Families should provide daughters with as much inheritance as sons, and as much authority over the use of inherited funds.
Women should be allowed to have maternity leave without fear of losing their job while they are away.
Men should be allowed to have paternity leave without fear of losing their job while they are away.
(e+): High schools should spend as much money on girls sports as on boys sports.
(E+): Women should have equal access to health clubs and recreational facilities in the workplace.
(E+): Husbands and wives should have equal roles in decisions about investments.
(E+): Husbands and wives should have equal roles in decisions about spending money.
(E+): Husbands and wives should have equal roles in decisions about where to live.
(E+): Husbands and wives should have equal roles in decisions about the education and care of their children.
Women who stay home and take care of their children should be allowed to declare themselves as employed for tax purposes.
(R): Special fellowships should be available to encourage women to go to graduate school in engineering, mathematics, economics, and other fields where women are underrepresented.
(E-): Women's sports at colleges should receive less funding than men's sports.
(E+): Fathers and mothers should be treated equally by the law in child custody cases. It should be the parent's circumstances that matter, not the parent's gender.
People in jobs traditionally done by women should be paid more, compared to people in jobs traditionally done by men.
Men should give up their seat to women on the train or bus.
(E+): Women and men should not be treated any differently on buses and trains in terms of seating.
(e-): Parents should spend more effort teaching girls to take care of their appearance than they spend teaching boys about this.
(S+): There are occasions on which women and men should eat separately.
(S+): Trains should have separate cars for women.
(S-): At a social dinner, men and women should stay together rather than separate.
(S-): In religious services, men and women should be together.
(S+): In college student residences, men and women should not live together on the same floor.
(R): More men should go into nursing and elementary-school teaching.
Medicine is a more appropriate career for a woman than engineering.
(r): More men should be foreign-language teachers.
(E+): In college classes, professors should call on men and women equally when the students raise their hand.
(r): In college classes, professors should call on women more than men when students raise their hands.
Men should not think badly of women who ask them for a date.
(e-): Virginity is more desirable in a woman than in a man.
(E+): In relationships between men and women, disloyalty should be equally censured for women and men.
(E-): In relationships between men and women, disloyalty should be censured more for women than for men.
The most important criterion in the choice of marriage partners should be the need for love and companionship.
Individuals should choose their own marriage partners rather than letting their parents decide.
(e+): Marriage partners should be about equally well educated.
(e-): Husbands should be better educated than their wives.
(S): Marriage partners should socialize with other people as a couple.
(E+): The husband and wife should have equal responsibility for household work other than child care.
(E-): The wife should have primary responsibility for house work other than child care.
Both parents should have equal say in the decision to have a child.
Men should have more say in the decision about whether to have a child.
Women should have more say in the decision about whether to have a child.
(B): If one parent is to care for a child, the child develops better if it is the mother rather than the father.
(B): Boys are naturally better at math and science than girls.
(B): Girls are naturally better at English, social studies, and languages than boys.
(B): Boys are naturally better at most sports.
(B): Boys need sports activities for their psychological development more than girls do.
(B): Boys are naturally better at learning to use computers.
(B): Human beings evolved for women to do the work at home and men to do the work outside of the home.
(B): Men are more capable than women of killing the enemy in war.
(B): Women have more medical problems than men.
(B): Men are better at making decisions about money.
(B): Women are better at making decisions about child care.
(B): Human beings evolved so that men have authority in the family.
(B): It is more difficult for a woman than for a man to stand up on a train or bus.
(B): Men and women are naturally interested in different topics for conversation.
(B): Men are naturally more concerned than women with matters of the spirit.
(B): Men are naturally better religious leaders than women.
(B): Men are more rational than women.
(B): Human beings evolved so that the man pursues the woman in courtship, not the other way around.
(B): Human beings evolved so that men need more sex partners than women.
(B): Women are better suited than men to child care.
(B): Men are better suited than women to work outside of the house.
(B): Men are better suited for higher education than women.
(M-): If men are naturally better at a particular task than women, we should leave things as they are if more men than women wind up doing that task.
(M-): If women are naturally better at a particular task than men, we should leave things as they are if more women than men wind up doing that task.
(M+): Even if men are naturally better than women at something, we should try to make sure that equal numbers of men and women do that thing.
(M+): Boys and girls should have equal opportunity and should not be treated differently.
(M-): Boys should be encouraged to do what boys are naturally good at, and girls should be encouraged to do what they are naturally good at.
(M+): There should be concerted efforts to raise children with a non-sexist orientation, at home and in schools.
(M+): Child rearing should try to compensate for natural differences between the sexes, even if that requires encouraging boys to do "girlish" things and encouraging girls to do "boyish" things.
It is not the business of government to worry about the different roles played by men and women.
(M-): Women and men should be respected in different ways.
Measurement of gender-role attitudes, beliefs, and principles
This document was generated using the LaTeX2HTML translator Version 96.1-c (Feb 29, 1996) Copyright © 1993, 1994, 1995, 1996, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html -html_version 3.1 -split 0 -no_subdir -ascii_mode pp.tex.
The translation was initiated by Jonathan M Baron on Sat Mar 30 16:32:52 EST 1996