Over the past decade, interest in the causes, consequences, and correlates of sexism has increased manifold. Where racial prejudice and discrimination were topics of concern to sociologists of religion during the 1960s and 1970s, gender has similarly emerged in the 1980s as worthy of our attention. Wilson (1978) claims that “religion is probably the single most important shaper of sex roles.” Nevertheless, empirical documentation linking the two variables is scarce. Although comprehensive in scope, much of what has been written remains descriptive, somewhat polemical, and impressionistic (see, for example, Lampe, 1981; Ruether, 1975, 1981; Roberts, 1984). Much of the empirical work that does exist merely touches on the interrelationships between religion and gender by conceptualizing religion minimally as either a simple correlate or as one of many variables in a multivariate model (see, for example, Mason and Bumpass, 1975; Lipman-Blumen, 1972; Dempewolff, 1974; and Henley and Pincus, 1978). Other investigations have given the religious variable short shrift by measuring it with a single indicator such as attendance or affiliation (see, for example, the study by Brinkerhoff and MacKie, 1984, which was constrained by secondary data analysis). In short, the current state of empirical research on the relationship between religion and gender is quite unsophisticated. The purpose of this paper is to investigate this matter somewhat more thoroughly.

But why posit a relationship between religion and gender? A traditional perspective on gender relations and the status of women appears to be deeply rooted in Western religious thought. Although our religious heritage does contain material supportive of female-male equality, “there can be little doubt that the parables, stories, teachings, and gospels of the Judeo-Christian tradition that our culture has chosen to emphasize are those that perpetuate gender-stereotyping” (Richardson, 1981: 100, emphasis in original deleted). Among the religious sources of gender traditionalism are the following:

1. Although English translations of scripture are sometimes more sexist than the original Hebrew, Greek, and Aramaic versions, elsewhere it is obvious that the scripture was written exclusively for men (Roberts, 1984: 350). Examples are the Ten Commandments (“Thou shalt not covet thy neighbor’s wife . . .”) and the covenant made by the Israelites with God during the Exodus, which made circumcision a requirement (Walum, 1977: 130).

2. Relatedly, although Old Testament attitudes towards women are inconsistent (the book of Proverbs depicts women as sources of great wisdom), the “overall effect is clearly that women have a subordinate position” (Roberts, 1984: 351).

3. The New Testament conveys ambivalent messages concerning gender. The Gospels depict an attitude toward women that was unusually egalitarian for the time and culture. Jesus “violated Judaic law by touching menstruating women, by speaking alone with a woman not his wife, by allowing women to witness and testify to his resurrection” (Walum, 1977: 129). However, the Epistles foster the doctrine of female inferiority. “Wives, submit yourselves unto your own husbands, as unto the Lord. For the husband is head of the wife, even as Christ is the head of the church.” (Ephesians 5:22–23.)

4. Although the Protestant reformers abolished the requirement of clergy celibacy, traditional gender imagery occurs in the writings of Luther, Calvin, and Knox (Roberts, 1984: 355). For example, ambivalence is seen in the teaching of Martin Luther in that men and women are equally called to life in Christ but that “the ‘natural’ order of the universe decrees women’s lesser status and obedience to her husband” (Langdon, 1980: 38).

5. Religious imagery appears to buttress male supremacy. God is father, judge, shepherd, king. Jesus and the twelve apostles are all male. The Christian tradition has two divergent views of the nature of womanhood: Eve, the first “sinner” responsible for humankind’s expulsion from paradise; and Mary, immaculately conceived virgin and mother of God. However, no matter how revered the Virgin Mary is, she remains human and, as such, subordinate to the masculine Trinity. Theologian Mary Daly (1975: 156) concludes: “If God in ‘his’ heaven is a father ruling ‘his’ people, then it is in the ‘nature’ of things and according to divine plan and the order of the universe that society be male-dominated.”

6. Finally, church organization has traditionally assigned different roles to women and to men. The hierarchy of church structure is consistent with both religious teaching and the fact that most of the major prophets and leaders of Western religions have been men. With few exceptions men are the authority figures—deacons, priests, clergymen, bishops, cardinals, popes. While many of the large Protestant denominations have agreed recently, after much deliberation, to ordain women, so far the numbers involved are small and many women clergy have been diverted into marginal roles (Roberts, 1984: 362). According to Bullough (1973, quoted in Roberts, 1984: 360), “regardless of what a religion teaches about the status of women, or what its attitudes toward sex might be, if women are excluded from the institutions and positions which influence society, a general misogynism seems to result.”

How is contemporary religion related to gender attitudes? Religion is broadly defined as a combination of beliefs, values, and behaviors which provides an overall worldview (see McGuire, 1981). It is well documented that in preindus-trial societies, religion permeated all aspects of social life. With increasing industrialization and secularization, religion became more privatized. It has, for the most part, withdrawn from the public sphere, becoming closely linked with the family. There is an intrinsic relationship between religion and family values. The family, the primary agent of gender socialization, derives many of its ideas about gender from religion. Ruether (1974: 9) claims (perhaps extravagantly) tnat religion is “undoubtedly the single most important shaper and enforcer of the image and role of women.” Moreover, the family determines the child’s exposure to subsequent socialization agents, including the church, and interprets the meaning of these secondary influences (see D’Antonio, 1980; Lampe, 1981). McMurry (1978: 83), who regards religion and family as reactive institutions which exert conservative influences, concludes that “women who are exposed to more of this influence through greater involvement should be more traditional.” During the socialization process, certain attributes such as compassion, self-sacrifice, obedience, and humility may be internalized as both feminine and religiously appropriate (see Lampe, 1981: 29). Based on these observations, it is hypothesized that the more religiously involved will portray more traditional gender attitudes.

It may well be that formal involvement is not a necessary condition to produce traditional gender attitudes. That is, people may possess a worldview or belief system that is highly religious in orientation but be quite independent of any official religion or denomination. (Bibby, 1983; Geertz, 1966; McGuire, 1981: 75–104.) After all, religiosity may have multiple dimensions, with the ideological aspect being independent from the ritualistic one (Glock and Stark, 1965). For this reason one might hypothesize that the stronger the religious belief system, the more traditional the gender-role attitude independent of ritualistic involvement.

However, many people do belong to specific religious denominations which are characterized by coherent bodies of values, beliefs, and practices derived from prescribed doctrines and organizations. Such denominations can serve as socialization agents, teaching gender attitudes reflecting particularistic doctrine, beliefs, and rituals. For example, a denomination’s position on women in the clergy, or its opposition to the Equal Rights Amendment (Chalfant et al., 1981: 345; Johnson, 1983; Tedlin, 1978) may very well influence members’ gender attitudes. Moreover, denominations vary considerably in the extent and nature of women’s participation in their formal organizational structures (Driver, 1976). Such structures convey attitudes about gender—impressionable children see important church roles as a male prerogative and experience only males making ceremonial contact with the Deity. Based on this premise, it is hypothesized that gender attitudes are related to denominational affiliation. Some denominations are more conservative than others because they cling more tenaciously to the past as they adhere to a more literal interpretation of scripture. Therefore, one can posit that members from those denominations that are more fundamentalists will be more conservative on gender matters. (See Brinkerhoff and MacKie, 1984; Hesselbart, 1976; Thornton and Freedman, 1979; Thornton et al, 1983.) The term fundamentalist refers to such doctrinal beliefs as Bible literalness, divinity of Christ, salvation through Christ, and separation from the world (Ammerman, 1982; Hunter, 1981).

Since religious and gender socialization may both take place within the context of the official religion, one must distinguish between the affiliation one is born into and that with which one may later identify. These, of course, may differ. (Lazerwitz and Harrison, 1980.) Since a great deal of the socialization takes place when one is a child, one might expect the childhood denomination to have the major impact; however, Brinkerhoff and MacKie (1984) and Lipman-Blumen (1972) report one’s current religious identification to be more important. This may result from people associating and identifying with groups who have similar values and beliefs to theirs. If their ideas about gender and other important matters have changed over time, they may switch group identifications, including religions, to bring consistency. Hence, current identity may be more important than childhood affiliation. Also, for a specific denomination to have an impact upon gender attitudes, identification may be more important than actual belonging, which is at times only nominal at best. Finally, caution must be exercised in the discussion of denominational influence because greater variation exists within denominations than between them. This is especially true when denominations are categorized broadly into Catholics, Protestants, Jews and “Nones.” (See Bouma, 1973; Brinkerhoff, 1978; Means, 1966; Rhodes and Nam, 1970.) Extensive “within variation” is not unique to Protestant groups but also holds for Catholics (Greeley, 1964,1972). High “within groups variation,” therefore, may render denomination a rather weak correlate of attitudes and behavior. To illustrate, D’Antonio and Stack (1980: 406) report that upon investigating specific denominations, “the attitudes of Baptists more closely approach those of Catholics in all abortion situations than they do those of other Protestants.”

Recently, considerable attention has been given to the Moral Majority and its impact upon society. Although it purports to be “an organization established to restore moral values to society” (Roberts, 1984: 128), theologians do not concur that all the values and attitudes stressed by the Moral Majority are basic to Christianity, e.g., their support of war spending. “What we see in the Moral Majority is a powerful folk religion which has blended Christianity and Americanism” (Roberts, 1984: 129). Following this reasoning it is predicted that those people who adhere to the tenets of this unofficial, folk religion, the Moral Majority, will also tend to be traditional in terms of gender attitudes.

In summary, it is hypothesized that religious belief, religious behavior, denominational identification, and agreement with the principles of the Moral Majority all will be correlated with gender attitudes. However, the limited, existing empirical research reports contradictory findings regarding these relationships. On the one hand, Martin et al. (1980) found religious affiliation to be the best predictor of gender conservativism, and McMurry (1978) reported religious activity to be more highly correlated with it than are the standard demographic factors. On the other hand, Barrish and Welch (1980), on the basis of a multi-university sample, found support for none of the above relationships and concluded that background factors are more important. Why this disparity in findings? Barrish and Welch (1980: 72) suggest that McMurry’s data are old (a 1964 sample) and that social attitudes have changed and “derive more from current political concerns and imagery than from conventional religious belief systems.” Although employing different measures, our own recent work from a random, urban sample (Brinkerhoff and MacKie, 1984) supports the linkages between religion and gender attitudes.

Since the Brinkerhoff-MacKie (1984) study was based on secondary data, selected measures were deemed inadequate (for example, church attendance served as the single indicator for all of religiosity); consequently, more complete measures are required. Furthermore, some interesting findings invited further and more sophisticated investigation. The Mormons were by far the most traditional denomination on the gender attitude scale. They were even more conservative than Catholics with regard to the specific attitude toward abortion. However, the Mormon subsample was small. Overall, our Catholics were more liberal vis-a-vis gender than other studies have reported (Campbell, 1966; McMurry, 1978; Meir, 1972; Porter and Albert, 1977; and Seaman et al., 1971).

As far as the Mormons are concerned, the literature provides little guidance on their gender attitudes. Indeed, Thomas (1983: 280) recentlv stated that “research evidence is virtually nonexistent that compares the extent to which children reared in Mormon families either accept or reject traditional sex roles” (emphasis added). Nevertheless, Thomas concludes “that acceptance of traditional sex roles would be relatively high for active Mormon families.” Both the Mormon organizational structure and its ideological system would lead to such a conclusion. Women are not allowed to hold the Mormon priesthood. Where women may be leaders in some women’s organizations, the men by virtue of possessing the priesthood virtually control the exercise of power. During many socialization experiences in their youth organizations, boys and girls are segregated, presumably to be treated differently. These organizational characteristics may contribute to Mormons’ ideas about gender. With regard to beliefs, women are taught that their power is derived through their husbands within the organization. Mormon theology emphasizes women’s role in the home (Thomas, 1983) in this highly patriarchal system (O’Dea, 1957). This is not to imply that Mormon women are denigrated, for it is one of very few religions that posit a “Mother-in-Heaven” (Heeren et al., 1984; Roberts, 1984: 372; Thomas, 1983; Vernon, 1980: 279). According to Mormon beliefs, women, along with their husbands, can share in exaltation—they also can become Gods. With only this cursory excursion into the historical and modern Mormon social system, one might hypothesize that Mormons would tend to be traditional on matters of gender.

Finally, because of both the paucity and the conflicting findings in this area, the Barrish and Welch (1980: 72) admonition seems appropriate: “Until further studies on the religiosity-sexism linkage are conducted, it seems wise to view the relationship as highly tentative, and to follow Wuthnow’s (1973: 128) sage advice about maintaining a healthy skepticism toward simplistic depictions of religiosity as a force in the service of social conservativism.”

Methodology

The Sample. The data for this paper are derived from a 1983 questionnaire study of 938 students which was carried out at the University of Calgary (n = 355), an Alberta Bible college (n = 71), Brigham Young University (n = 236), and the University of Nebraska (n = 276). Comparisons between Canadian and American students and between state school and denominational school students should prove enlightening. The fact that these data were obtained from introductory social science students does, of course, restrict their generaliz-ability. On the one hand, youth’s views may inform us of the shape of the future. Wuthnow argues that “youth, being generally better educated and apparently more aware of current trends than many older people, are more likely to be exposed to such trends” and, further, that “young people may also be more receptive to cultural innovations than their elders since they are still in the process of forming basic values” (1976: 856). (See also Ryder, 1965: 848.) However, a number of studies show that gender attitudes vary with age, with younger respondents holding more egalitarian attitudes than older ones (Boyd, 1984; Thornton et al., 1983).

Measurement. The definition of religiosity, as well as its measurement, has been the focus of considerable attention. The number and composition of dimensions running through the concept have been explored. (For example, see Faulkner and Dejong, 1966; King and Hunt, 1975; Nudelman, 1971; Stark and Glock, 1968.) Although not without limitations, one can argue that measuring denominational identification is important because of differences in specific doctrines, values, beliefs, practices, and organizational structures which may influence gender attitudes. When focusing on denomination, two important distinctions must be made. First, the religious affiliations in which respondents were raised must be differentiated from those with which they currently identify. Furthermore, “identification” is more important than “belonging,” since nominal membership may have little or no actual impact. Second, denomination can be categorized broadly (Catholic, Protestant, Jew, and Nones) or specifically (such as Greek Catholic, Baptist). Brinkerhoff and MacKie (1984) reported “current identification,” specifically categorized, to be the strongest denominational predictor of gender attitudes. In the study reported here, identification with Current Denomination was tapped by asking “Do you presently feel a part of some religious group?” Responses then were classified somewhat broadly into “Nones,” “Catholics,” “Mainline Protestants,” “Conservative Christians” or “Fundamentalists,” and “Mormons.” (Mormons are maintained as analytically distinct because of their claim to be non-Protestant, the objective of building on a past study, and their “well-known” position vis-à-vis family/gender matters.)

Several measures in this study were created in a similar fashion and may be summarized briefly here. Batteries of items, or questions, to tap both behaviors and attitudes were answered on Likert-type formats. These items then were subjected to exploratory factor analysis, using principal factoring iterations (see Kim, 1975; Rummel, 1967; and Zeller and Carmines, 1980). Based on the magnitudes of the loadings and their rotational patterns, items that did not appear to contribute to the underlying variable in question were eliminated. Because the common variance explained was sufficiently high on first factors for all cases, and because the remaining factors had low eigenvalues and low loadings, only the first factors are employed here. Those items of a given battery meeting the criterion of a factor loading of .40 or greater (this may be seen as somewhat conservative, since Kerlinger [1979: 189] accepts the common criterion of .30) then were summed into the scales presented in Table 1. This approach to scale construction “has been a most popular and effective technique in social research” (emphasis in original, Lin, 1976: 192; see also Smith, 1974). Finally, Cronbach’s alpha (1967) is presented as a measure of reliability. The ranges of the summed scales, their means and standard deviations also are found in Table 1.

Table 1. Measuring Gender Attitudes and Religion: A Summary Table of the Parmeters in Scale Construction for Four University Samples

Factor Analysis:^a First Factors Summed Scale Values^b

^aUsing principle factoring with iterations (Kim, 1975)

^bThe higher the score, the higher the "machismo," the higher the religiosity scores

^cSample sizes vary due to nonresponses. In any scale where the respondent failed to answer 25 percent or more of the items, he/she was eliminated; for the occasional nonresponse, the mean of the items was employed for the missing response.

The Macho scale, which is the major dependent variable in this analysis, illustrates the operations summarized in Table 1. Villemez and Touhey (1977) built a scale of twenty-eight items which (with slight variations to make them applicable to both Canadian and American respondents) were administered using five-point, Likert-type response categories. As shown, eighteen of these items clustered together to account for 85 percent of the variance, had factor loadings greater than .40, and had a reliability coefficient of .89. When the eighteen items are summed into a scale, it has a range from 18 to 85, a mean of 52.1, and a standard deviation of 12.1. [1]

Two broad dimensions of religiosity commonly found in the literature include religious ideology (or belief) and ritualism (or behavior). Employing items derived, in part, and modified from many of these studies, five scales were developed which are used as variables to explore this study’s hypotheses. Religious Belief is derived from a set of twelve items having four response categories. Fundamentalism is a subset of four of these items and is based on Ammerman’s (1982) differentiation between fundamentalism and evangelicalism—the former, but not necessarily the latter, possessing the four qualities of biblical literalness, personal salvation through Jesus Christ, divinity of Jesus Christ, and separation from the world. Religious Behavior, or ritualism, includes twenty items of a personal nature (such as praying, reading the Bible) and of a congregational nature (such as attending meetings, contributing money to a church), presented on a five-point scale. Four of these items form a subset termed the Church Participation scale to illustrate formal, congregational involvement. Finally, although many argue the case for multidimensionality, the items in the Belief and Behavior scales are combined to form the Overall Religiosity scale. (In short, at least for these data, as reflected in Table 1, a common dimension seems to underlie these religiosity items.) A series of statements consistent with the views of the Moral Majority (for example, “prayer should be allowed in our schools”) were posed with a four-point response ranging from “strongly disagree” to “strongly agree.” Eleven of these clustered to form the Moral Majority scale.

Some (including Roof and Perkins, 1975) have suggested the importance of the variable “religious salience” in studies of religiosity. Respondents were presented with several aspects of the self and requested to rank the importance of each. Religious Salience was an evaluation of the importance of “myself as a religious person” on a four-point scale. Similarly, Gender Salience was measured by their ratings of “myself as a male or female person.”

It has been argued that religion breeds social conservativism and that the more “fundamental” denominations would be more conservative in their attitudes. Based on these assumptions, the denominations are grouped into five categories according to their mean scores on the four-item Fundamentalism scale. The Alliance, Baptist, Church of Christ, Evangelical, and Pentecostal denominations form the “fundamentalist” or Conservative Christian group, with means ranging from 14.75 to 15.5 (out of a possible score of 16.0); the other Protestants become the Mainline Protestants; while the Catholics (a mean of 11.8), Mormons (13.8), and Nones (8.9) were kept distinct. Both the seven non-Christians and the respondents from denominations with only one member were excluded since the latter may indicate individual rather than group fundamentalism.

Because attitudes and values are at least partially shaped during youth, selected measures of religious background may prove important to ideas about gender. For example, a question was posed to assess whether respondents still held to the beliefs taught in church while growing up (Persistent Believer). Finally, the parents’ religious affiliations also were tapped.

A series of demographic and background questions, measured by single indicators, were asked in order to serve as explanatory or control variables. These include age, sex, year in school, type of university, major, self-reported grades, marital status, community size, political orientation, ethnicity, socioeconomic status (based on father’s occupation using the Treiman [1977] international scale), and parents’ educational levels. Some of these demographic and self-evaluative measures are explored in the analysis for their impact on the relationships between religion and gender.

Data Analysis. Statistical analysis is largely descriptive and mainly exploratory. Initially, the hypotheses adumbrated earlier are cursorily examined using simple zero-order correlation coefficients. Where the independent variables are categorical in nature (for example, Denomination), dummy variable analysis is entertained. The commonly supported position of Labovitz (1970) is followed, treating ordinal data as if they were interval.

Finally, because no causal claims are made, four models are explored by nonhierarchical stepwise regression. This multivariate procedure allows several variables to enter into the predictive equation in the order of the magnitude of variance explained in the dependent variable, i.e., Macho in this case. By altering the predictor variables in various models, hints of the relative impact of different factors, interaction effects, evidence of spuriousness, and so on are suggested. This procedure should lead to the development of a more sophisticated model in this important field of endeavor.

Because the sample is composed of university students and must be considered nonrandom and nonrepresentative, tests of statistical significance are not presented. That is, inference is not the question in this exploratory analysis. With sample sizes of the magnitudes presented in this study, most correlation coefficients (R2s) are statistically significant.

Some Exploratory Findings

Simple Relationships. Initially, in order to explore the linkages between religion and gender, the various scales and dimensions of religiosity are correlated with three “indicators” of gender attitude conservativism: the dependent variable—the Macho scale—and two indicators dealing directly with attitudes about abortion. This approach both offers evidence of validity and fits the findings to a body of literature on abortion. Table 2 contains the correlations which suggest several tentative conclusions. First, it is rather obvious that the data support McMurry (1978) and Martin et al. (1980), rather than Barrish and Welch (1980). That is, the religious factors seem much more strongly related to all three gender attitude measures than do the demographic/background variables. (The measures are very similar to those used by Barrish and Welch; consequently, differences cannot be accounted for solely by different items.) Some might suggest that the relatively large number of Mormon students, who tend to be both religiously involved and traditional on gender attitudes, may have biased the results somewhat. The counter argument, of course, is that the large number of “Religious Nones” at the University of Calgary may bias the results in the opposite direction; that is, the two samples are “offsetting in their effects on religiosity measures. Comparative sociologists argue for the selection of research sites that allow for maximum variation in the independent variable(s); the strategy in the present study meets this requirement. We cannot focus on the Mormons as an important group without a similar sampling procedure. Finally, an earlier paper (Brinkerhoff and MacKie, 1984) using somewhat different measures on a random sample of an older, urban population corroborates the findings presented here.

To interpret the table further, note that the Overall Religiosity scale accounts for 22.3 percent of the variance in gender attitude as reflected by the Macho scale, while respondent’s sex accounts for only 7.8 percent of the variance. Sex is the best predictor of the Macho scale among the demographic variables, unless the religious affiliation of the parents is considered to be a background factor. (Sex is less powerful as a predictor of the abortion attitude, however.) Although at first glance these coefficients appear somewhat weak, in comparison with other studies, the Rs are quite respectable. Furthermore, when comparing the religious variables as predictors of gender attitudes with the demographic variables (for example, SES), one can only conclude that they are substantial.

Second, the pattern is consistent across the Macho scale and the two single indicators of gender. The higher the religiosity, the more conservative one is on matters of gender. Interestingly, and in agreement with Martin et al. (1980), one’s current religious identification appears to be the strongest religious predictor of Macho. In all three cases it is a stronger correlate than is childhood affiliation. Bear in mind that for this analysis Religious Affiliation and Identification have been classified into five groupings: Religious Nones, Catholics, Mainline Protestants, Conservative Christians (or Fundamentalists), and Mormons are treated as dummy variables. Based on previous research (Brinkerhoff and MacKie, 1984), even stronger relationships would be expected to result if more specific classifications were employed, such as Presbyterians, Pentecos-tals, Methodists, etc.

Third, selected variables must be considered correlates and not causal in nature. As a variable the Moral Majority scale seems to be strongly related to Macho and to both abortion items. Caution must be exercised in considering this finding in any causal sense; rather, it is probably another measure of conservativism which bears a distinctive religious connotation and, therefore, correlates highly with Macho. Indeed, this would lead to the prediction that Moral Majority also should correlate highly with current religious identification, just as Macho does (although the data are not presented here, the R2 is .426). Similarly, the self-report of political views, although not as highly related, also must be considered another general measure of conservativism.

Finally, the University affiliation is one “apparent” nonreligious variable that is strongly correlated with gender attitudes. However, is it really a nonreligious variable? Both the Bible college and BYU were selected for the study because they, in fact, are religious institutions. BYU is made up mainly of Mormon youth who are among those of this age group most strongly committed to the church, and undoubtedly the Bible college also is attended by strongly committed Conservative Christians; that is, the most committed youth may be in attendance at these schools rather than at alternative schools. The religious factors undoubtedly interact with the University variable to influence the relationship with gender attitudes. These interaction effects are examined later in the analysis to discover if it is the religious factor, or conversely the University factor, that correlates with conservative gender attitudes. It probably can be assumed that religion, in a time-ordering perspective, occurred prior to University attendance. In any event both religious identification and university are strongly related. (Treating the variables at the nominal level, the Lambda coefficient is .50, relatively high.) Time and space do not permit the discussion of other correlates in any depth, but it should be noted that self-perception as a religious person is much more strongly correlated with gender attitudes than is perception of self as gender, that is, as a male or female person.

Table 2. Religious and Demographic Factors' Impact on Gender Attitudes: Some Exploratory Zero-order Correlation

^aA positive sign means the greater the religiosity, the higher the Macho score (or more traditional in gender attitude).
^bCategorical variables requiring dummy variable analysis.

Categorical Predictors. The data in Table 2 provide some information about the direction of the relationships for selected variables; that is, overall, the greater the religiosity, the more traditional the gender attitudes. However, the categorical variables such as denomination, sex, marital status, and university do not disclose equivalent information regarding direction; that is, the coefficients do not indicate which classes are more conservative (for example, are males or females more conservative?). Therefore, the data in Table 3 go beyond the size of the coefficients for some of those nominal variables for which dummy variable analysis was employed earlier. Rather than present findings for all categorical factors in Table 2, the data pertinent to the dependent variables have been confined to the Macho scale and the item “Abortion Is Sin,” along with four of the independent variables.

Table 3. Current Religious Identification and Selected Demographic Variables by the Macho Scale and "Abortion is Sin" Item

^aThe constant is the predicted value for the omitted category, Mormon.
^bPredicted value for males.
^cPredicted value for married.
^dPredicted value for University of Nebraska

Table 3 illustrates that Current Religious Identification is strongly related to gender attitudes. In this table “Constant,” the reference category, refers to Mormons with a mean value of 60.512. These data indicate that the Mormons are only slightly more conservative on gender attitudes than are the Fundamentalists, followed by the Mainline Protestants and then the Catholics. The “Religious Nones” are the most egalitarian according to these data; of course, this finding is in agreement with others (see, for example, Dempewolff, 1974; Lipman-Blumen, 1972; McMurry, 1978), who all report the nonaffiliated to be less traditional. Catholics appear to be less traditional than either group of Protestants or than the Mormons; this result is contrary to those of Meir (1972), Campbell (1966), and McMurry (1978: 84) who reports “Catholics to be less favorable towards female equality.” On the other hand, it corroborates the findings of Brinkerhoff and MacKie (1984) when they used a different measure of gender egalitarianism on a random sample. Even on the “Abortion Is Sin” item on which Catholics have explicit doctrine, the Catholics are more liberal than the Mormons and Fundamentalists. Although the data are not reported here, other aspects of denomination (childhood affiliation, parents’ religious affiliation) follow the same basic pattern; that is, on the Macho scale, the least conservative are the Religious Nones, followed by the Catholics, with Mormons being the most traditional. On matters regarding abortion, the Mormons and Fundamentalists are always more conservative than the Catholics.

Interpreting the relationships between denomination and gender attitudes must be done with considerable caution due to the nature of the sample. Over 95 percent of the Mormons were sampled from BYU, while approximately 65 percent of the Conservative Christians came from the Bible college. Both groups may attract the most traditional or conservative of Mormon and fundamentalist youth. Cannon and Christensen’s (1978) research, for example, illustrates that Mormon youth at BYU are highly traditional and have become more so over time. Mainline Protestants and Catholics, if sampled from religiously sponsored institutions, may be relatively more conservative than those examined here from the Universities of Calgary and Nebraska. A “within groups” analysis (for example, comparing Conservatives at the Bible college with those from the public universities) might clarify this; however, there are too few non-BYU Mormons to entertain such analyses with these data. The earlier Brinkerhoff and MacKie (1984) study with a random sample of older adults corroborate the findings presented here. Mormons and Conservative Christians are more traditional on gender attitudes. The additional caveat that must be included involves the age of the respondents. The sample is highly educated, relatively young in age, and probably more “liberal” than the average adherent to their denominations.

With regard to other categorical variables, as expected females are more egalitarian than males. Respondents who are single, followed by those who are married, tend to be most conservative. Finally, on the University variable, those who attend the Bible college are slightly more conservative than BYU students (which might be expected, given the denominational characteristics of the schools), while those from the University of Calgary are the most egalitarian.

Some Exploratory Models. Up to this point, scales have been constructed, and selected simple bivariate and regression coefficients have been examined. These results suggest that the grouping of religious factors are more strongly associated with gender attitudes than are the background or demographic variables. Since these factors (such as denomination and sex) may interact to influence gender attitude (for example, Macho), multivariate techniques must be brought to bear on these relationships. This will enable an overall assessment of the impact of different variables on the gender attitude measures. Table 4 presents a summary of the proportion of variance in both Macho and the “Abortion Is Sin” indicator, as explained by four sets of independent variables considered jointly. This multiple regression analysis is divided into four models based both on technical concerns and on the theoretical and empirical import of the variables employed for prediction. Multiple regression requires the use of dummy variables for nominal level, or categorical, variables such as Denomination or University. Because more than two sets of dummy variables used simultaneously in an equation render interpretation somewhat difficult, this analysis is limited to a single dummy variable, Current Denomination Identification. (Sex and marital status may be considered noncategorical when conceptualized as the “degree of femaleness” and the “degree of ‘marriedness.’” In doing so, the usual caveat is required.) In a stepwise multiple regression approach, the option of deciding when a variable enters into the explanatory schema (equation) exists; however, unless the dummy variables are entered in at the beginning, the interpretation may become problematic.

Table 4. Proportion of Variance Explained for Selected Factors Affecting the Macho Scale and Abortion Attitude in Alternative Multiple Regression Models

^aAbbreviated factors are: Denomination = Current Denominational Identification; Moral = Moral Majority Scale; Sex = Sex;
Religiosity = Overall Religiosity, combining Belief and Behavior Scales; Self = Self as a Religious Person
^bDummy variable entered first in the equation.
^cCoefficients in paretheses refer to the amount of variance explained by that variable from the proportion of variance explained by all the variables entered into the equation to that point; they are R²s.

How are the variables to be placed in the predictive model chosen? Of course, the primary considerations are theoretical. Second, the findings in Tables 2 and 3 should provide direction. Therefore, based on theory, findings, and the above-mentioned technical restrictions, four models are presented in Table 4. In Model 1, Current Denomination was entered into the equation first, by design, where the Moral Majority scale, Sex, Marital Status, and the Overall Religiosity scale appear in the order of the magnitude of the remaining variance explained in the dependent (or predicted) variables, the Macho scale and Abortion Attitude. In Table 4, a summary table, Model 1 illustrates that the coefficient R2 is .263—or 26.3 percent of the variance in Macho is accounted for by Current Denomination. Of the remaining unexplained variance, Moral Majority enters first to explain an additional 17.1 percent, followed by Sex which accounts for 6.8 percent; Marital Status adds just .6 percent, followed by Overall Religiosity (.3 percent). The total variance explained in the Macho scale by these five variables is 51.1 percent; however, fully 50.1 percent was accounted for by only the first three variables. It is interesting to observe that when the Abortion Attitude is predicted, Current Denomination explains 43.9 percent of the variance, Moral Majority adds another 20 percent, and Overall Religiosity adds an additional .6 percent, while Marital Status and Sex contribute nothing. Note that for the other variables free to enter into the equation after Denomination, Moral Majority is the strongest predictor, with the remaining variables altering their order for the two dependent variables.

It may not be surprising, but the findings thus far would suggest that Denomination, Moral Majority, and Overall Religiosity may share some of the explanatory power. Removing Denomination from the equation and adding the “Self as Religious”—which is fairly strongly correlated with the gender items as reflected in Table 2—lends some confirmation to this notion, as shown in Model 2. When the “Self as Religious” variable is entered into Model 2, little is added to the prediction of the Abortion Attitude (.4 percent), and almost nothing to the explanation of Macho. It should be noted that the total variance explained is only slightly reduced for both gender attitude measures when Denomination is eliminated. As suspected, this would suggest that the religious variables (including assessing Self as Religious) appear to be interacting to influence gender.

Focusing on Model 3, with Moral Majority removed, the proportion of total variance explained decreases somewhat (about 5 percent for Abortion Attitude and 10 percent for Macho) from Model 2. This confirms that the independent variables, Denomination and Moral Majority, jointly influence gender attitudes, with Moral Majority being slightly stronger. When both are eliminated from the equation, as shown in Model 4, Religiosity emerges as the strongest predictor. In this four-variable model, Sex adds 9.2 percent in explaining Macho, but the other variables account for almost nothing. For the Abortion Attitude, Model 4 has only Religiosity making any real difference. It appears, then, that Religiosity is joining Denomination and Moral Majority in sharing much, but not all, of the variation explained. Model 1 remains the most powerful predictive model. Of course, other models have been examined for which the data cannot be presented here.

Some general findings might be usefully summarized. When Overall Religiosity is broken into its two components, the Belief and the Behavior scales, little difference in the amount of explained variance occurs; that is, they do no better separately than when combined, as presented in the four models. However, it is interesting that Belief always enters into the equation before Behavior when attempting to predict Abortion and does so about half the time with Macho. Recall that University is a strong predictor of gender attitudes at the bivariate level. Earlier it was suggested that this may be due, in part, to interaction with Denomination. If University is exchanged for Denomination as the dummy variable in Model 1, the findings are similar, with the amount of variance explained decreasing slightly (to 49.9 percent for Macho, and 61.6 percent for Abortion). Adding Self as Religious to the equations has only slight effect on the Abortion Attitude but almost no impact on the Macho scale. Probably much of the predictive power of the “Self as Religious” is shared by the other religious predictors. Among the demographic factors, Sex is of some modest influence on Macho but adds almost nothing to Abortion Attitude. Sex remains weak even when entered into the equation first, a model examined but not reported here. In fact, one can conclude that the religious factors have far greater influence in all multiple regression models when compared with any of the background factors. This finding concurs with McMurry (1978) and questions Barrish and Welch (1980).

Some Conclusions

This paper argues that religion has an impact on the traditionalism surrounding gender attitudes. Demographic variables have little import. The findings appear to be related to denominational identification, as well as independent belief systems and behaviors. The overall best predictor of gender attitudes is position on the Moral Majority scale; however, this should not be interpreted in a causal sense. Probably both degree of modernism on gender attitudes and position on the Moral Majority scale are linked to some other causal variable(s). Both are ideological systems which relate to another ideological system—one dealing with ideas about religion (for example, the nature of human beings, the ultimate meaning of life, etc.). Furthermore, it appears that gender attitudes are not strongly related—at least for this University sample—to the usual background factors, for example, sex, SES, etc. However, they do relate to denomination. This suggests that specific religious groups do have some degree of consistency or homogeneity in their beliefs and values about religion, the Moral Majority, and gender attitudes, providing a common worldview for their committed adherents.

As shown, the Mormons tend to be the most traditional denomination with regard to gender attitudes. Although data have not been thoroughly examined, one suspects that Mormons would score high on the Moral Majority scale, be highly committed on the Belief and Behavior scales, and so on. The Mormons’ position on women’s roles is entirely consistent with Brinkerhoff s (1978: 215) finding in a study of the impact of gender on high school youth’s educational and occupational orientations. He concluded: “Substantively, those Mormons who are highly involved or committed to the belief system and organization structure, aspire to goals consistent with the belief that woman’s major role is that of mother and homemaker.” In another study Brinkerhoff and MacKie (1984) reported that married Mormon adults are highly traditional on gender attitudes but are highly egalitarian with regard to reported behaviors—division of household labor and familial decision making. How might this discrepancy between attitudes and behavior be explained? Is it a function of the small sample, the measurement? Perhaps O’Dea’s (1957: 255) conclusion that Mormons have a conflict between “social idealism born of Mormon beliefs and political conservatism” may provide a hint. He notes (1957: 249) that early Mormons “came close to accepting the equality of women with men” while “accepting patriarchal ideals of family organization.” In short, man is the head of the familv, but woman is considered his equal in many ways; for example, both are eternal intelligences (O’Dea, 1957: 250). The Macho scale used in this study probably captures the Mormon patriarchal ideology, while selected behaviors illustrate the social idealism of egalitarianism.

In any event, for this study, it is fairly conclusive that Mormons possess a highly conservative gender ideology. One must remember that the findings involving denomination may be questioned because the respondents from BYU could be more committed and more traditional than average Mormon youth. On the other hand, since university students are usually somewhat more modern and closer to the forefront of societal change than are the rank and file, one might expect a random sample of highly religious Mormons to be even more traditional. Vernon (1980: 287) argues that women and men within the Mormons experience different churches, and he concludes that “the societal changes in the roles of women have already changed both churches, or the experience of both males and females within the church and promises to introduce even further changes.” With the high gender conservativism found among the Mormon students—the highly committed, intelligent, presumably future leaders—one can only speculate whether these changes will eventually alter the patriarchal belief system.

Bibliography

Ammerman, N. T. 1982. “Comment: Operationalizing Evangelicalism: An Amendment.” Sociological Analysis 43: 170–72.

Barrish, G., and M. R. Welch. 1980. “Student Religiosity and Discriminatory Attitudes Towards Women.” Sociological Analysis 41 (Spring): 66–73.

Bibby, R. W. 1983. “Searching for Invisible Thread: Meaning Systems in Contemporary Canada.” Journal for the Scientific Study of Religion 22(2): 101–19.

Bouma, G. D. 1973. “Beyond Lenski: A Critical Review of Recent ‘Protestant Ethic’ Research.” Journal for the Scientific Study of Religion 12: 141–55.

Boyd, M. 1984. Canadian Attitudes Toward Women: Thirty Years of Change. Ottawa: Women’s Bureau, Labour Canada.

Brinkerhoff, M. B. 1978. “Religion and Goal Orientations: Does Denomination Make a Difference?” Sociological Analysis 39 (3): 203–18.

Brinkerhoff, M. B., and M. M. MacKie. 1984. “Religious Denominations’ Impact upon Gender Attitudes: Some Methodological Implications.” Review of Religious Research 25(4): 365–78.

Bullough, V. L. 1973. The Subordinate Sex: A History of Attitudes Toward Women. Urbana, IL: University of Illnois Press.

Campbell, D. F. 1966. “Religion and Values Among Nova Scotian Students.” Sociological Analysis 27(Summer): 80–93.

Cannon, K. L., and H. T. Christensen. 1978. “The Fundamentalist Emphasis at Brigham Young University, 1935–1973.” Journal for the Scientific Study of Religion 17: 53–57.

Chalfant, H. P., R. E. Beckley, and C. E. Palmer. 1981. Religion in Contemporary Society. Sherman Oaks, CA: Alfred Publishing.

Cronbach, L. J. 1967. “Coefficient Alpha and the Internal Structure of Tests.” Pp. 132–67 in W. A. Mehrens and R. L. Ebel (eds.), Principles of Educational and Psychological Measurement. Chicago: Rand McNally & Company. (See also the 1951 original in Psychometrika 16: 297–334.)

Daly, M. 1975. “God Is a Verb.” Pp. 153–70 in U. West (ed.), Women in a Changing World. New York: McGraw-Hill.

D’Antonio, W. V. 1980. “The Family and Religion: Exploring a Changing Relationship.”Journal for the Scientific Study of Religion 19(2): 89–104.

D’Antonio W. F., and S. Stack. 1980. “Religion, Ideal Family Size, and Abortion: Extending Renzi’s Hypothesis.” Journal for the Scientific Study of Religion 19(4): 397–408.

Dejong, G. F.,J. E. Faulkner, andR. H. Warland. 1976. “Dimensions of Religiosity Reconsidered: Evidence from a Cross-Cultural Study.” Social Forces 54(June): 866–90.

Dempewolff,J. A. 1974. “Some Correlates of Feminism.” Psychological Reports 54 (April): 671–76.

Driver, A. B. 1976. “Review Essay: Religion.” Signs 2 (Winter):434–42.

Faulkner, J. E., and G. Dejong. 1966. “Religiosity in 5-D: An Empirical Analysis.” Social Forces 45: 246–54.

Geertz, C. 1966. “Religion as a Cultural System.” Pp. 1–46 in M. Banton (ed.), Anthropological Approaches to the Study of Religion. London: Tavistock Publications.

Glock, C. Y., and R. Stark. 1965. Religion and Society in Tension. Chicago: Rand McNally.

Greeley, A. M. 1964. “The Protestant Ethic: A Time for a Moratorium.” Sociological Analysis 25(2): 20–30.

Greeley, A. M. 1972. The Denominational Society. Glenview, IL: Scott, Foresman.

Heeren, J., D. Lindsey, and M. Mason. 1984. “The Mormon Concept of Mother in Heaven: A Sociological Account of Its Origins and Development.” Journal for the Scientific Study of Religion23(4): 396–4ll.

Henley, N. M., and F. Pincus. 1978. “Interrelationship of Sexist, Racist and Homosexual Attitudes.” Psychological Reports 42 (February)83–90.

Hesselbart, S. 1976. “A Comparison of Attitudes Toward Women and Attitudes Toward Blacks in a Southern City.” Sociological Symposium 17 (Fall): 45–68.

Hunter, J. D. 1981. “Operationalizing Evangelicalism: A Review Critique and Proposal.” Sociological Analysis 42(4): 363–72.

Johnson, S. 1983. From Housewife to Heretic. New York: Anchor Press/Doubleday.

Kerlinger, F. 1979- Behavioral Research: A Conceptual Approach. New York: Holt, Rinehart and Winston.

Kim, J. 0. 1975. “Factor Analysis.” Pp. 468–514 in N. H. Nie, C. Hull, J. Jenkins, K. Steinbren-ner, and D. Bent (eds.), Statistical Package for the Social Sciences. New York: McGraw Hill.

King, M. B., and R. A. Hunt. 1975. “Measuring the Religious Variable: National Replication.” Journal for the Scientific Study of Religion 14: 13–22.

Labovitz, S. 1970. “The Assignment of Numbers to Rank Order Categories.” American Sociological Review 35: 515–25.

Lampe, P. E. 1981. “Androgyny and Religiosity.” International Journal of Women’s Studies 4(1): 27–34.

Langdon, M. E. 1980. “Images and Ideals of Victorian Women, 1820–1850.” Unpublished master’s thesis, Department of History, University of Calgary.

Lazerwitz, B., and M. Harrison. 1980. “A Comparison of Denominational Identification and Membership.” Journal for the Scientific Study of Religion 19(4): 36l-67.

Lin, N. 1976. Foundations of Social Research. New York: McGraw-Hill.

Lipman-Blumen, J. 1972. “How Ideology Shapes Women’s Lives.” Scientific American 266(Jan- uary)33–42.

Martin, P. Y., M. W. Osmond, S. Hesselbart and M. Wood. 1980. “The Significance of Gender as a Social and Demographic Correlate of Sex Role Attitudes.” Sociological Focus 13(October): 383–96.

Mason, K., and L. L. Bumpass. 1975. “U.S. Women’s Sex Role Ideology, 1970.” American Journal of Sociology 80 (March): 1212–19.

McGuire, M. B. 1981. Religion: The Social Context. Belmont, CA: Wadsworth.

McMurry, M. 1978. “Religion and Women’s Sex Role Traditionalism.” Sociological Focus 11 (2): 81–95.

Means, R. L. 1966. “Protestantism and American Sociology: Problems of Analysis.” Sociological Analysis 27: 128–37.

Meir, H. C. 1972. “Mother-centeredness and College Youths’ Attitudes Towards Social Equality for Women: Some Empirical Findings.” Journal of Marriage and the Family 34 (February): 115–21.

Nudelman, A. E. 1971. “Dimensions of Religiosity: A Factor-analytic View of Protestants, Catholics, and Christian Scientists.” Review of Religious Research 13: 42–56.

O’Dea, T. F. 1957. The Mormons. Chicago: University of Chicago Press.

Porter, J. R., and A. A. Albert. 1977. “Subculture or Assimilation? A Cross-cultural Analysis of Religion and Women’s Role.” Journal for the Scientific Study of Religion l6(4): 345–59.

Rhodes, A. L., and C. B. Nam. 1970. “The Religious Context of Educational Expectations.” American Sociological Review 35: 253–67.

Richardson, L. W. 1981. The Dynamics of Sex and Gender (2nd ed.). Boston: Houghton Mifflin.

Roberts, K. A. 1984. Religion in Sociological Perspective. Homewood, IL: Dorsey Press.

Roof, W. C., and R. B. Perkins. 1975. “On Conceptualizing Salience in Religious Commitment.” Journal for the Scientific Study of Religion 14: 111–28.

Ruether, R. R. 1974. Religion and Sexism. New York: Simon & Schuster.

Ruether, R. R. 1975. New Woman, New Earth. New York: Seabury Press.

Ruether, R. R. 1981. “The Feminist Critique in Religious Studies.” Pp. 52–66 in E. Langland and W. Gove (eds.), A Feminist Perspective in the Academy. Chicago: University of Chicago Press.

Rummel, R.J. 1967. “Understanding Factor Analysis.” Journal of Conflict Resolution 11(4): 440–80.

Ryder, N. B. 1965. The Cohort as Concept in the Study of Social Change.” American Sociological Review 30: 843–61.

Seaman, J. M.,J. B. Michel and R. C. Dillehay. 1971. “Membership in Orthodox Christian Groups, Adjustment and Dogmatism.” Sociological Quarterly 12 (Spring): 252–59.

Smith, K. W. 1974. “On Estimating the Reliability of Composite Indexes Through Factor Analysis.” Sociological Methods and Research 2(4): 484–51

Stark, R., andC. Y. Glock. 1968. American Piety: The Nature of Religious Commitment. Berkeley, CA: University of California Press.

Tedlin, K. 1978. “Religious Preference and Pro/Anti Activism on the Equal Rights Amendment Issue.” Pacific Sociological Review 21 (January): 55–66.

Thomas, D. L. 1983. “Family in the Mormon Experience.” Pp. 267–88 in W. V. D’Antonio and J. Aldous (eds.), Families and Religions: Conflict and Change in Modern Society. Beverly Hills, CA: Sage Publications.

Thornton, A., D. F. Alwin and D. Camburn. 1983. “Causes and Consequences of Sex-Role Attitudes and Attitude Change.” American Sociological Review 48: 211–27.

Thornton, A., and D. Freedman. 1979- “Changes in the Sex Role Attitudes of Women, 1962–1977: Evidence from a Panel Study.” American Sociological Review 44 (October): 831–42.

Treiman, D. 1977. Occupational Prestige in Comparative Perspective. New York: Academic Press.

Vernon, G. M. 1980. “Mormon Women.” Pp. 279–87 in G. M. Vernon, Mormonism: A Sociological Perspective. Salt Lake City: University of Utah Press.

Villemez, W. J., and J. C. Touhey. 1977. “A Measure of Individual Differences in Sex Stereotyping and Sex Discrimination: The ‘Macho’ Scale.” Psychological Reports 41: 411–15.

Walum, L. R. 1977. The Dynamics of Sex and Gender. Chicago: Rand McNally.

Wilson, J. 1978. Religion in American Society. Englewood Cliffs, NJ: Prentice Hall.

Wuthnow, R. 1973. “Religious Commitment and Conservatism: In Search of an Elusive Relationship.” In C. Glock (ed.), Religion in Sociological Perspective. Belmont, CA: Wadsworth.

Wuthnow, R. 1976. “Recent Pattern of Secularization: A Problem of Generations?” American Sociological Review 41 (October): 850–67.

Zeller, R., and E. G. Carmines. 1980. Measurement in the Social Sciences: The Link Between Theory and Data. Cambridge, England: Cambridge University Press.

Notes

[1] These means and standard deviations of the Macho scale are different from those of Villemez and Touhey (1977) and others using this scale (e.g., Banish and Welch, 1980) because the items were scored from 1 to 5 rather than from 0 to 4, and because only eighteen of the items met the rigorous criteria to be employed in the scale, rather than the full twenty-eight used by the others. All items for the various scales are available from the authors upon request.