11. The Determinants of Religious Behavior: A Theoretical Model and Empirical Test

By Marie Cornwall

Marie Cornwall, “The Determinants of Religious Behavior: A Theoretical Model and Empirical Test,” in Latter-day Saint Social Life: Social Research on the LDS Church and its Members (Provo, UT: Religious Studies Center, Brigham Young University, 1998), 345–372.

Chapter 11: The Determinants of Religious Behavior: A Theoretical Model and Empirical Test

Marie Cornwall


This paper examines five categories of factors which have been found to influence religious behavior (group involvement, belief-orthodoxy, religious commitment, religious socialization, and socio-demographic characteristics), suggests alternative ways of measuring the various factors, and presents a theoretical model. Rather than using standard measures of group involvement, measures of in-group, marginal, and out-group personal community relationships are used. The model is tested using data collected from Latter-day Saints living in the United States. Each of the various factors is found to influence religious behavior. Religious commitment has the strongest direct effect. Belief, personal community relationships, and religious socialization variables also influence behavior, but their influence is primarily indirect. Personal community relationships indirectly influence religious behavior by helping individuals maintain a religious world view and commitment to the norms and expectations of the religious group.

Durkheim’s study The Elementary Forms of Religious Life is not simply the sociological study of religion. He was primarily interested in demonstrating how societal processes motivate individual action. Durkheim concluded: “The only source of life at which we can morally reanimate ourselves is that formed by the society of our fellow beings; the only moral forces with which we can sustain and increase our own are those which we get from others” (1915, p. 473). The significance of moral communities as reinforcers of religious thought and practice remains a central issue within sociology as the debate about the extent of secularization in modern society continues to grow (Campbell 1971; Fenn 1978; Lyon 1985; Martin 1978; Shiner 1967; Stark and Bainbridge 1985; Wilson 1966). This paper examines the impact of personal communities on the development of a religious world view (Berger 1967; Cornwall 1987; Gaede 1976; Welch 1981) and adherence to behavioral expectations (Hougland and Wood 1980; White 1968). Previous theoretical and empirical work is reviewed, operational refinements using network methodology are made, and a theoretical model and empirical test are presented.

The research reported here is based on data collected from members of The Church of Jesus Christ of Latter-day Saints living in the United States. It is recognized that Mormons are a unique population. However, the disadvantage of that uniqueness is far outweighed by the opportunity to examine the social processes which influence religious belief and behavior within a specific religious group where particular attention can be given to the measurement of denomination-specific beliefs and behaviors. [1]


Research on the determinants of religious behavior has identified four categories of variables related to religiosity: group involvement (Gaede 1976; Hougland and Wood 1980; Lenski 1963; Roberts and Davidson 1984; Welch 1981; White 1968), belief-orthodoxy (Alston and Mcintosh 1979; Hoge and Carroll 1978; Hoge and Polk 1980; King and Hunt 1972; Lofland and Stark 1965; Stark and Glock 1968), religious socialization (Cornwall 1988; Greeley and Rossi 1966; Greeley, McGready, and McCourt 1976; Himmelfarb 1979), and sociodemographic variables (Alston and Mcintosh 1979; Gaede 1977; Hoge and Carroll 1978; Hoge and Polk 1980; Hougland and Wood 1980; Mueller and Johnson 1975; Roberts and Davidson 1984; Welch 1981). The empirical evidence suggests each set of factors has some degree of influence on religious behavior, although the amount of influence is greater for some than for others. At present there is no empirical research which examines all factors together, and little theory being developed which might guide such an effort.

Group Involvement

Lenski (1963) provided empirical evidence of the impact of the religious group more than two decades ago. Shortly thereafter, White (1968) proposed the interaction model of religious influence, and several studies (Finney 1978; Gaede 1976; Roberts and Davidson 1984; Welch 1981) have drawn upon his limited treatment of the subject. The most significant contributions of his perspective were the emphasis on religion as a group phenomenon and his insights regarding the effects of in-group and out-group interaction. White suggests, for example, that the person with strong interactive ties outside the religious group and weak interactive ties within the group is less likely to be influenced by the group and may eventually withdraw.

Taking a somewhat different approach, Berger (1967) noted the impact of group involvement on the development of a religious world view. Social reality is sustained by “conversations” with significant others, and plausibility structures (churches, families, and voluntary associations) provide a consistent world view for individuals living in a pluralistic world. Consistent with White, Berger recognized that the introduction of alternative world views can change one’s subjective reality. Effective socialization occurs to the extent that one’s subjective reality mirrors the objective reality of the group. It is therefore logical to assume that introduction to alternative realities via conversations with significant others whose subjective reality does not mirror the reality of the group will negatively influence the adoption of the group reality.

Both perspectives emphasize the impact of group involvement, but Berger is interested in the impact of group involvement on the acquisition and maintenance of a religious world view while White’s focus is on the impact of group involvement on religious behavior. Both perspectives suggest the impact of in-group and out-group interaction, but no empirical test has yet been undertaken, partly because good measures of in-group and out-group interactions have not been available.

In past research the measurement of group involvement has been limited to the use of two variables: frequency of church attendance, and the number of one’s five closest friends who belong to the same congregation or the same religious group (Finney 1978; Gaede 1976; Hougland and Wood 1980; Lenski 1963; Roberts and Davidson 1984; Welch 1981). The use of church attendance as a measure of group involvement is confusing because attendance is also a measure of the ritualistic dimension of religiosity. Using the number of close friends who are members of the same congregation or religious group is also inadequate. It is an imprecise measure of in-group interaction, it does not adequately measure the strength of these relationships, and it provides no information about the salience of group involvement for the primary relationships being measured. Such methodological problems have produced nonsignificant associations between group involvement and belief and behavior. For example, Lenski (1963) found a very low correlation between associational involvement (church attendance) and communal involvement (primary relations with those of the same socioreligious group). A much stronger correlation is likely if a distinction is made between relationship ties with people who are actively involved in the religious group and those who are only nominally involved, and especially if the relationship ties are within the same congregation or parish.

More precise measures of group involvement are now possible using methodology derived from network studies conducted by Fischer (Fischer 1982; Fischer et al. 1977). Network relationships are personal communities which consist of family, relatives, friends, and associates, or the set of people with whom the individual is directly involved. The focus of much of the social network research is to understand personal communities or social networks as the new form of Gemeinschaft in modern society. Networks are the personal connections by which society is structured and individuals are integrated into it.

These personal community relationships can be measured as (1) the number of in-group, marginal, or out-group ties, and (2) the strength of these ties. In-group ties are relationships with people who are active participants in the same religious group. Marginal ties are relationships with people who are members of the same religious group but who are not active participants, and out-group ties are relationships with people who do not belong to the same religious group.

The strength of a network tie is measured by the degree to which relations between individuals are multistranded. In the network literature, multistranded relations are assumed to be more enduring, supportive, and intimate (Boissevain 1974; Craven and Wellman 1973; Fischer et al. 1977). Fischer (1982) measured multistrandedness by counting the number of different types of interactions reported with a given individual: discussing work, helping around the house, discussing personal problems, advising on decisions, etc.

Using network methodology to operationalize group involvement, we can hypothesize that the number and strength of in-group ties are positively associated with religious belief and behavior, and the number and strength of out-group or marginal ties are negatively associated with religious belief and behavior.


Scholars have traditionally assumed that religious orthodoxy has a strong influence on behavior. Some suggest that belief is the major variable affecting religious involvement (Hoge and Carroll 1978; King and Hunt 1972; Lofland and Stark 1965; Stark and Glock 1968). A great deal of evangelical effort is based on the assumption that the greater one’s belief-orthodoxy, the more likely an individual is to participate (Roberts and Davidson 1984). But others are uncomfortable with such conclusions and question whether beliefs influence group involvement (as measured by attendance) or group involvement influences the formation of beliefs (Roberts and Davidson 1984; Welch 1981). Confusion has been partly alleviated by introducing new measures of group involvement and redefining the research question as an examination of the relative impact of belief-orthodoxy and group involvement on religious behavior and the direct and indirect effects of each. As suggested by Berger (1967), group involvement should strongly influence the development of belief-orthodoxy. Both Gaede (1976) and Welch (1981) found friendship ties within the local congregation were strongly associated with several measures of belief-orthodoxy. Using the same network methodology suggested above, Cornwall (1987) has also shown that both the number and strength of in-group ties are positively associated with belief and commitment and the number of out-group ties is negatively associated with belief and commitment.

If the influence of the group is primarily cognitive, as the sociology of knowledge perspective would suggest, then the primary function of the religious group is to help individuals maintain belief in and commitment to the norms of the group. On the other hand, the influence of the group may be primarily normative, influencing behavior through the operation of sanctions, as the interpersonal model of religious influence would suggest (White 1968).

Two studies have examined the influence of group involvement and belief-orthodoxy on religious involvement (Hougland and Wood 1980; Roberts and Davidson 1984). Findings from both studies suggest that group involvement has a greater influence on behavior than does belief-orthodoxy. However, both studies use fairly weak measures of religious belief. While measures of traditional Christian beliefs such as belief in God, Christ, and life after death should be included in any analysis, a full test of the impact of beliefs on behavior must include measures of the particularistic beliefs of the group.

On the basis of past research, we can hypothesize that both group involvement and belief orthodoxy have a direct effect on religious behavior. However, we would also expect that group involvement also has an indirect effect on behavior since group involvement directly influences belief.

Religious Commitment

This study also includes measures of religious commitment. While belief is a cognitive dimension of religiosity, commitment is an affective dimension (see Cornwall et al. 1986) and is a measure of the salience of religion in a person’s life. An empirical examination of the relationship between measures of belief, commitment, and behavior suggests commitment is an intervening variable between belief and behavior (Cornwall et al. 1986). The utility of commitment measures is that they provide an explanation for the low association between measures of religious belief and behavior found in some studies (Stark and Glock 1968). Thus one can believe, but congruent behaviors occur only when one is truly “committed” to the beliefs.

Along with belief and behavior, commitment is likely to be highly dependent on strong ties with other group members (Kanter 1972), and thus strongly influenced by the nature of one’s personal community relationships.

Religious Socialization

Research has focused on three agents of religious socialization: the family (Greeley and Rossi 1966; Greeley et al. 1976), the church, and peers (Cornwall 1988; Fee et al. 1981). Conclusions from such research are generally consistent with traditional conceptions of socialization: the family is the principal agent of religious socialization (Erikson 1950; Freud 1933), while peers and the religious institution are secondary agents.

In the broader sociological literature, Acock (1984) has suggested parent-child similarity may be as much a product of the intergenerational transfer of social status as the product of individual psychosocial influence. Glass, Bengtson, and Dunham (1986) have provided evidence to support the hypothesis that parental socialization also involves the “successful intergenerational transmission of class, race, religious affiliation, marital status, and other prominent social statuses that structure life experience and mold social attitudes.”

A similar perspective is offered by Himmelfarb’s research on religious socialization. He suggests “parents socialize their children by channeling them into other groups or experiences (such as schools and marriage) which will reinforce (have an additive influence on) what was learned at home and will channel them further into adult activities” (1979, p. 478). Research by Himmelfarb among Jews and by Cornwall (1988) among Latter-day Saints provides empirical support for the channeling effect of the family. Both studies suggest that socialization processes influence the development of a world view, but perhaps more importantly, also channel individuals into personal communities which in turn help maintain a person’s religious beliefs and commitment to religious norms.

Demographic Variables

A great deal of attention has been given to the influence of demographic characteristics on religious behavior (Alston and Mcintosh 1979; Davidson 1977; Hoge and Carroll 1978; Stark and Glock 1968; Welch 1981). Generally this has been the result of efforts to test deprivation theory, which argues that religion is a source of compensation for persons suffering from economic or social deprivation (Demerath 1965; Glock and Stark 1965; Glock, Ringer, and Babbie 1967; Niebuhr 1929; Pope 1942; Troeltsch 1931). The research suggests that the impact of social class, education, and occupation differs across religious groups (Mueller and Johnson 1975) and that the impact of these demographic characteristics may be changing over time (Nelson and Potvin 1980). However, research conducted within specific denominations (Gaede 1976; Hougland and Wood 1980; Roberts and Davidson 1984; Welch 1981) which examines the impact of demographic characteristics relative to other variables such as amount of in-group association and religious socialization demonstrates the need to include them. While demographic characteristics do not have a strong effect on religious behavior, these variables should not be dropped from causal models because they do explain variance (Hougland and Wood 1980; Welch 1981). Other studies (Gaede 1976; Roberts and Davidson 1984) have suggested that while demographic variables do not have a significant direct effect on religious involvement, they have an important indirect effect.

It generally has been assumed that demographic characteristics are indicators of one’s location in the social structure, and location in the social structure influences religiosity. For example, research has suggested that lower-class individuals are more privately religious while upper-class people are more publicly religious (Stark 1972; Stark and Bainbridge 1985). It could be that the influence of such variables as education, marital status, and region of the country is mostly a function of their influence on social relationships. Fischer (1982) has demonstrated that age, education, gender, and marital status influence the nature of one’s social relationships. And Cornwall (1985) has shown that education, gender, marital status, and region of the country influence the number of social ties within a religious group as well as the number of out-group ties. Much of the effect of demographic characteristics on religious behavior may be indirect through their impact on the nature of personal community relationships.

A Theoretical Model

Building upon the theoretical and empirical research reviewed above, a basic model of the predictors of religious behavior is presented in fig. 11.1. Religious socialization has a direct positive influence on personal community relationships. Personal community relationships influence religious belief and commitment, and both personal community relationships and religious belief and commitment influence conformity to the norms and expectations of the religious group. [2] Religious belief also influences commitment. Demographic characteristics influence the nature of one’s personal community relationships. No direct effect of religious socialization on religious belief and commitment is assumed. In addition, no direct effect of the demographic characteristics on religious behavior is assumed.



Respondents were randomly selected from complete membership lists from twenty-seven different Mormon wards (congregations) from all parts of the United States. These twenty-seven wards had previously been chosen from a larger sample of Mormon stakes (typically made up of from six to twelve wards) which had been selected randomly from the different administrative areas of the Church in the United States.

A membership roster was obtained from each of the ward units used in the sample, and a list of households was randomly selected from each roster. Using this procedure, a total of thirty-two participating and forty-eight nonparticipating families was obtained within each of the twenty-seven units. Level of religious participation was previously obtained from the local bishop. Nonparticipating households were oversampled because pretest data indicated they would be less likely to respond to the study than would participating families. One adult member in each household was designated for inclusion in the sample by a toss of a die. Adult children over the age of eighteen were included in the sampling universe. When only one adult lived in the household or when there was only one adult who was a member of the LDS church, that person was included in the sample. The final sample consisted of 1,874 members over the age of eighteen.

Figure 11.1: Theoretical Model

A thirty-two-page questionnaire with appropriate cover letters requesting participation in what was described as one of the most important studies of Mormon religiosity ever undertaken was mailed to all adults in the sample. Follow-up postcards and additional copies of the questionnaire were mailed to nonrespondents over the next eight-week period (Dillman 1978). Adjusting for the 576 “undeliverable” questionnaires, the response rate was 74 percent from active members and 48 percent from less active members. The overall response rate was 64 percent. Despite the efforts taken to get accurate membership rosters, church records for those who were not currently attending religious services were found to be seriously inaccurate. Most undeliverable questionnaires were returned from less active members who had moved and left no forwarding address. The sample is representative of Mormons who could be located using information available from current church records. Respondents, therefore, probably have more positive feelings towards the LDS Church and may be more religious than the general population of LDS Church members.

Measures of Religious Behavior

Three measures of religious behavior are used as dependent variables: Frequency of personal prayer and two measures of institutional modes of religiosity (frequency of church attendance and home religious observance). These three measures are used because they represent slightly different aspects of religious behavior. Frequency of personal prayer is a personal mode of religious behavior; church attendance is an institutional mode (for a discussion of personal and institutional modes of religiosity see Cornwall et al. 1986). Frequency of home religious observance is an institutional mode of religious behavior although it is a very private aspect of religious participation. Home religious observance is encouraged by Church leaders in the form of a weekly Family Home Evening. Occurring within the home, it is a private aspect of religious participation and is more difficult for outsiders to monitor but is expected of good Latter-day Saints.

Frequency of personal prayer is measured with one item: “Other than blessing the food, how often do you usually have personal prayer?” Response categories ranged from “never” to “daily” on a seven-point scale. Frequency of church attendance is measured with one five-point item ranging from “never” to “weekly.” Frequency of home religious observance is a four-item scale measuring frequency of family prayer, family religious discussion, family scripture reading, and family discussion about right and wrong.

Measures of Personal Community Relationships

Network data were collected using a modification of the techniques developed by Fischer and his associates (Fischer 1982; Jones and Fischer 1978; McCallister [Jones] and Fischer 1978). Respondents were asked to make a list of “all the adults who are important to you in your life.” They were to use only first name or initials and could list family members, friends, neighbors, co-workers, etc. Respondents were asked to think of

People you depend on for help with day to day problems. This includes taking care of your children, helping with work around the house, borrowing tools or equipment, etc.

People you see on a regular basis. You may have dinner with them, go to the movies together, or share hobbies or special interests.

People you talk with about personal worries or concerns, or whose advice you seek before making a decision.

Space was provided for respondents to list up to fifteen names. Respondents were also told that if they had more than fifteen people who were important to them, they should select the fifteen most important. Once the list was complete, respondents were asked to provide information about the nature of each relationship, about each person’s religious activity, and about the person’s membership in the LDS Church.

The mean number of names listed by respondents was 9.5; the median was 8.6. About 10 percent of the sample listed fewer than five people in their network, and 23 percent listed the full fifteen.

Six variables were included in the analysis as measures of group interaction. Three of the variables were measures of the number of in-group, marginal, and out-group ties. These variables were calculated by counting the number of names who were (1) participating members of the LDS Church (in-group), (2) members of the LDS Church but not currently participating (marginal), and (3) not members of the LDS Church (out-group). The other three variables were measures of the mean strength of in-group, marginal, and out-group ties. Strength was measured as the number of different types of interaction reported with a given individual as indicated above. The strength of a tie was calculated by scoring 1 for each type of interaction mentioned. The mean strength of ties was calculated by adding the strength scores of all persons listed as in-group, marginal, or out-group ties and then dividing by the number of in-group, marginal, or out-group ties.

Since none of the above measures specifically identified whether the respondent’s spouse was also LDS, another variable designating spouse membership was included (coded 1 if spouse was LDS and 0 if spouse was not LDS or if respondent was not currently married).

Measures of Religious Belief and Commitment

Four measures of religious belief and commitment are used in the analysis: two scales measuring the institutional model of religiosity and two scales measuring a more personal mode of religiosity (Cornwall et al. 1986).

1. Traditional orthodoxy is defined as belief in traditional Christian doctrines such as the existence of God, the divinity of Christ, life after death, Satan, and the Bible. These beliefs are not unique to Mormonism; acceptance or rejection of them is largely independent of affiliation with a particular religious group or institution.

2. Particularistic orthodoxy refers to the acceptance or rejection of beliefs peculiar to a particular religious organization. Particularistic orthodoxy includes acceptance of such doctrines as the prophetic calling of Joseph Smith and the current church president, the authenticity of the Book of Mormon, and the belief that The Church of Jesus Christ of Latter-day Saints is the only true church on earth. Four items were used to measure particularistic orthodoxy.

3. Spiritual commitment was measured using a five-item scale which focuses on degree of commitment to God. Items tap feelings such as loving God with all one’s heart, willingness to do whatever the Lord wants, and the importance of one’s relationship with God.

4. Church commitment is the affective orientation of the individual towards the religious organization or community. It measures the attachment, identification, and loyalty of the individual towards the church organization or religious community. Five items were used to measure church commitment.

Measures of Religious Socialization

Religious socialization was measured using seven variables. Three variables measured family influences. Respondents were asked to indicate the religious preference of their parents when the respondent was age twelve to eighteen. A dummy variable was created by coding one if both parents were present in the home and both were members of the LDS church and zero if one or both parents were not LDS or if one parent was not present in the home. [3] Respondents also reported the frequency of their parents’ church attendance. Parental church attendance is the average frequency of church attendance for both parents when the respondent was age twelve to eighteen. If only one parent was present in the home, the item is the frequency of that parent’s church attendance. Home religious observance experienced as a teenager was measured with a scale composed of four items measuring frequency of family prayer, family religious discussions, family Bible or scripture reading, and family discussions of right and wrong.

The amount of socialization received through church participation was measured by two variables: a measure of the frequency of attendance at religious services during the teenage years, and number of years of seminary. Seminary is a program of religious study offered to LDS students in the ninth through twelfth grades.

Respondents were asked to indicate how many of their friends were active members of the LDS church during their teenage and young adult years. Response categories included (1) none of them, (2) a few of them, (3) about half of them, (4) most of them, and (5) all of them. The friends variable is a measure of the proportion of one’s friends who were active LDS during the teenage years, and the decline variable is a measure of the amount of decline in proportion of active LDS friends between the teenage and young adult years. Decline was computed by subtracting the proportion of active friends who were LDS during the young adult years (nineteen to twenty-five) from the proportion of active friends who were LDS during the teenage years.

Demographic Measures

Five demographic and background characteristics were included in the analysis: (1) a dummy variable coded 1 if divorced, separated, or never married, (2) years of education completed, (3) gender (coded 1 if female and 0 if male), (4) age (included in the analysis because of its demonstrated effect on religious belief and commitment), and (5) region of the country—three dummy variables: West (California, Washington, Oregon, and Alaska), Rocky Mountain (Nevada, Idaho, Colorado, Arizona, Wyoming, and Montana), and East/South (all other states except Utah). These regional variables are included because Latter-day Saints are not evenly distributed throughout the United States. There are particularly large concentrations of Mormons in Utah, Idaho, and California.


To examine the relative and cumulative effects of the several variables in the analysis, all variables were regressed on three measures of religious behavior: personal prayer, frequency of church attendance, and frequency of home religious observance. In each analysis, age at baptism was entered first as a control variable. Latter-day Saints are baptized at the age of eight, but approximately one-third of the membership in the United States are converts baptized after the age of eighteen. Entering age at baptism as a control variable is one way to control for differences in socialization processes experienced by converts.

The effect of each variable can be assessed by first examining the zero-order correlation coefficients. However, since we are most interested in examining the effects of each category of variables, a stepwise regression analysis and an incremental F-test were also carried out. An examination of standardized beta coefficients for all analyses also suggested the need to examine the direct and indirect effects of all variables.

Table 11.1: Zero-order Correlations of Five Variable Sets with Religious Behavior


Personal Observance

Church Attendance

Home Religious Observance

Personal community relationship

Spouse member




# of in-group ties




# of marginal ties




# of out-group ties




Mean strength in-group




Mean strength marginal




Mean strength out-group























Religious socialization





Parental attendance




Home religious observance




Youth attendance




Seminary attendance




Friends active LDS




Decline in friends active LDS




Demographic characteristics





Gender (1= female)
















Rocky Mountain




East, South





An examination of the zero-order correlations found in table 11.1 indicates that the belief and commitment variables are most strongly associated with religious behavior, the social relationship variables have a somewhat weaker association, and the religious socialization items have the weakest association with religious behavior. Mixed results were found among the demographic variables. Age, gender, and education are significantly correlated with the dependent variables although the correlations are not large. There appears to be no effect of marital status and region.

The usefulness of examining the effect of in-group, out-group, and marginal ties is demonstrated by the correlation coefficients. Number and strength of in-group ties are positively associated with all three behavioral measures, while number and strength of marginal and out-group ties are negatively correlated with behavior items. However, an examination of the standardized regression coefficients suggests that the strength of ties does not add any additional predictive value to the model.

As can be seen in the first column of table 11.2, the commitment items explain the greatest amount of variance in behavior when entered first. Forty-two percent of the variance in personal prayer, 50 percent of the variance in church attendance, and 34 percent of the variance in home religious observance are accounted for by commitment. The high degree of association between commitment and behavior items may in part be an artifact of measurement. Factor analysis indicates that the items cluster into independent dimensions, although some item overlap does exist (Cornwall et al. 1986). The utility of including belief and commitment items is, however, suggested both theoretically and empirically. First, the belief items are consistently more highly correlated with commitment items than behavioral measures are, suggesting an important mediating effect. Second, the pattern of correlations for belief, commitment, and behavior scales with other variables in the model differ. For example, belief and commitment scales are consistently more highly correlated with personal community measures than with behavior measures (see Cornwall 1987). These findings suggest that the belief, commitment, and behavior items do tap different dimensions of religiosity.

Personal community relationships account for between 15 and 43 percent of the variance in religious behavior (depending upon the dependent variable under study) when entered first. The results of the stepwise analysis suggest (table 11.2, column 2) that once the personal community relationship and belief and commitment variables are entered in the equation, the other variables do not contribute much more to the explained variance. The additional variables explain at most five percent of the variance in religious behavior.

The most crucial test of the causal importance of a set of variables is the incremental F test. Each set of variables was dropped out of the multiple regression, one set at a time, with replacement for the other tests, R-squared figures were computed and compared to see if the increase in explained variance associated with each set was significant. The amount of unique variance explained by each set of variables is provided in the last column of table 11.2. Looking at the last column, commitment clearly makes the greatest unique contribution in the personal prayer model. However, in both the church attendance and home religious observance models, personal community relationships also make a substantial contribution. These findings suggest that institutional modes of religious behavior may be more susceptible to the influence of in-group, marginal, and out-group ties than are personal modes of religious behavior.

Path models allow the examination of both direct and indirect effects of the several variables in the model. Fig. 11.2 presents a path model for frequency of church attendance. [4] Church commitment has the strongest direct effect on church attendance, although particularistic beliefs also have a strong direct effect. In addition, the number of in-group and out-group ties, gender, and education have direct effects. Church commitment is most directly influenced by particularistic beliefs, although number of in-group ties and education also have an influence. The network variables have their greatest impact on particularistic beliefs and therefore a strong indirect effect on commitment. The channeling influence of two religious socialization variables is apparent. Complete LDS family and number of active LDS friends during the teenage years influence both in-group and out-group ties, which in turn have a significant impact on belief.


Several conclusions can be derived from this analysis. First, we have identified five factors and tested their influence on religious behavior. The results suggest that all five factors: personal community relationships, religious commitment, religious belief, religious socialization, and the demographic variables have a significant impact on religious behavior. The need to examine both direct and indirect effects is apparent, but each of these sets of factors should be considered in further research examining the determinants of religious behavior.

Figure 11.2: Path Model Demonstrating Direct and Indirect Effects of Selected Variables in Church Attendance Model

Second, the demonstrated impact of belief and commitment on behavior suggests a fruitful area for research. For example, belief in and commitment to the normative order of a religious group may be more important in predicting behavior than sanctions levied by the group for non-conformance. Interviews with non-participating Latter-day Saints give credibility to this interpretation. They are less likely to adhere to behavioral expectations primarily because they do not value them. It may be that there are actually two mechanisms of social control. The first and more powerful may be acceptance by the individual of a particular world view and commitment to the normative order. The second, of course, is the societal sanctions which punish or reinforce certain behaviors. Perhaps more attention should be given to the first.

Third, the significance of in-group, out-group, and marginal ties is also demonstrated. The zero-order correlation coefficients suggest an association between personal community relationships and religious behavior. Although much of this effect is moderated by the introduction of belief and commitment measures, a direct effect of number of in-group ties remains. While others (Hougland and Wood 1980; Roberts and Davidson 1984) concluded that relationship variables have the strongest influence on religious behavior, the inclusion of affective measures of religiosity suggests that commitment has the strongest influence. This does not mean, however, that personal community relationships have little impact on behavior. Rather, much of the effect of in-group and out-group ties on behavior is indirect through its direct effect on belief and commitment. The impact of social relationships on religious belief and commitment has been demonstrated elsewhere (Cornwall 1987). There is growing evidence (Gaede 1976; Welch 1981) that religious belief and commitment are highly dependent upon the extent to which an individual is integrated into a religious community. Interaction within or outside the group has its greatest impact on individual world views. And for the believer, the number of in-group ties is an added impetus for religious commitment. Durkheim stressed the importance of “collective sentiments and the collective ideas” which could not be achieved “except by the means of reunions, assemblies and meetings where the individuals being closely united to one another, reaffirm in common their common sentiments” (1915, p. 475). Perhaps in modern pluralistic society individuals carry with them personal communities that sustain and enhance these collective sentiments and ideas. These personal communities shape personal beliefs, level of commitment, and also have some effect on behavior. But while group sanctions influence behavioral conformity, the effective operation of sanctions may be dependent upon the extent to which the individual believes in and is committed to the group.

Fourth, findings from this research and Cornwall (1988) suggest the need to broaden our thinking about the socialization process. Socialization includes the transfer of attitudes and social statuses, as well as a process whereby individuals are channeled into friendships and experiences that maintain the beliefs and attitudes of the former generation. While the religious socialization variables had little direct impact on religious behavior, further analysis reported elsewhere (Cornwall 1987, 1988) suggests that parental attendance and home religious observance have a significant direct impact on adult belief and commitment. In addition, parental attendance and home religious observance also have a significant impact in that they channel individuals into friendship networks during the teen and young adult years which support and sustain the religious values taught in the home. Friendship patterns during the young adult years have a direct impact on the nature of personal community relationships as an adult.

Finally, gender and education influence religious behavior even after controlling for all other factors. The gender effect has been noted in several research efforts. (For a review see deVaus and McAllister 1987.) What is interesting is that gender remains important even when controlling for religious socialization, personal community relationships, and belief and commitment. It is important to note, however, that the gender effect is not large (Cornwall 1989). The positive effect of education has been demonstrated in prior research as well, particularly among Mormons (Albrecht and Heaton 1984).

The continuing viability of religion in modern pluralistic societies may be primarily a function of the survival of moral communities that reinforce religious belief, commitment, and behavior (Stark and Bainbridge 1985; White 1968). Lenski concluded almost three decades ago that “the subcommunity is an important instrument for extending the influence of religious groups in the life of the community” (1963, p. 327). More recently, Welch has concluded that “the survival of traditional commitment in modern America is explained by the survival of particular types of moral communities, and by the integration of individuals into those communities” (1981, p. 91). The processes by which families and religious communities influence the development of religious belief and commitment and encourage religious behavior are an eminently sociological issue deserving further study.

Marie Cornwall is professor of sociology at Brigham Young University. This article was originally published in Social Forces 68:572–92; reprinted with permission.


Acock, Alan C. 1984. “Parents and Their Children: The Study of Inter-Generation Influence.” Sociology and Social Research 68:151–71.

Albrecht, Stan L. and Tim B. Heaton. 1984. “Secularization, Higher Education, and Religiosity.” Review of Religious Research 26:43–58. Also in this volume.

Alston, Jon P. and William A. McIntosh. 1979. “An Assessment of the Determinants of Religious Participation.” The Sociological Quarterly 20:49–62.

Berger, Peter L. 1967. The Sacred Canopy: Elements of a Sociological Theory of Religion, edited by David G. Bromley. Garden City, NY: Doubleday.

Boissevain, J. 1974. Friends of Friends. Oxford: Blackwell.

Campbell, Colin. 1971. Toward a Sociology of Irreligion. London: Macmillan.

Cornwall, Marie. 1985. “Personal Communities: The Social and Normative Bases of Religion.” University of Minnesota, Minneapolis. Unpublished manuscript.

———. 1987. “The Social Bases of Religion: A Study of Factors Influencing Religious Belief and Commitment.” Review of Religious Research 29:44–56.

———. 1988. “The Influence of Three Agents of Religious Socialization: Family, Church, and Peers.” Pp. 207–31 in The Religion and Family Connection: Social Science Perspectives, edited by Darwin L. Thomas. Provo, UT: Religious Studies Center, Brigham Young University.

———. 1989. “Faith Development of Men and Women over the Life Span.” Pp. 115–39 in Aging and the Family, edited by Stephen J. Bahr and Evan T. Peterson. Lexington, MA: Lexington.

Cornwall, Marie, Stan L. Albrecht, Perry H. Cunningham, and Brian L. Pitcher. 1986. “The Dimensions of Religiosity: A Conceptual Model with an Empirical Test.” Review of Religious Research 27:226–44. Also in this volume.

Craven, Paul and Barry Wellman. 1973. “The Network City.” Sociological Inquiry 43:57–88.

Davidson, James D. 1977. “Socio-Economic Status and Ten Dimensions of Religious Commitment.” Sociology and Social Research 61:462–85.

Davidson, James D. and Dean D. Knudsen. 1977. “A New Approach to Religious Commitment.” Sociological Focus 10:151–73.

Demerath, Neil J. III. 1965. Social Class in American Protestantism. Chicago: Rand McNally.

DeVaus, David and Ian McAllister. 1987. “Gender Differences in Religion: A Test of the Structural Location Theory.” American Sociological Review 52:472–81.

Dillman, Don A. 1978. Mail and Telephone Surveys: The Total Design Method. New York: Wiley.

Durkheim, Emile. 1915. The Elementary Forms of the Religious Life. Translated by Joseph Ward Swain. New York: Free Press.

Erikson, Erik. 1950. Childhood and Society. New York: Norton.

Fee, Joan L., Andrew M. Greeley, William C. McCready, and Teresa Sullivan. 1981. Young Catholics in the United States and Canada. Los Angeles: Sadlier.

Fenn, Richard K. 1978. Toward a Theory of Secularization. Storrs, CT: Society for the Scientific Study of Religion, University of Connecticut.

Finney, John M. 1978. “A Theory of Religious Commitment.” Sociological Analysis 39:19–35.

Fischer, Claude S. 1982. To Dwell among Friends: Personal Networks in Town and City. Chicago: University of Chicago Press.

Fischer, Claude S., Robert Max Jackson, C. Ann Stueve, Kathleen Gerson, Lynn M. Jones, and Mark Baldassare. 1977. Networks and Places: Social Relations in the Urban Setting. New York: Free Press.

Freud, Sigmund. 1933. New Introductory Lectures in Psychoanalysis. New York: Norton.

Gaede, Stanley D. 1976. “A Causal Model of Belief-Orthodoxy: Proposal and Empirical Test.” Sociological Analysis 37:205–17.

———. 1977. “Religious Participation, Socioeconomic Status, and Belief Orthodoxy.” Journal for the Scientific Study of Religion 16:245–62.

Glass, Jennifer, Vern L. Bengtson, and Charlotte Chorn Dunham. 1986. “Attitude Similarity in Three-Generation Families: Socialization, Status Inheritance, or Reciprocal Influence?” American Sociological Review 51:685–98.

Glock, Charles Y. and Rodney Stark. 1965. Religion and Society in Tension. Chicago: Rand McNally.

Glock, Charles Y., Benjamin Ringer, and Earl Babbie. 1967. To Comfort and to Challenge. Berkeley: University of California Press.

Greeley, Andrew M. and Peter Rossi. 1966. The Education of Catholic America. New York: Aldine de Gruyter.

Greeley, Andrew M., William McGready, and Kathleen McCourt. 1976. Catholic Schools in a Declining Church. Kansas City, MO: Sheed and Ward.

Himmelfarb, Harold S. 1979. “Agents of Religious Socialization among American Jews.” The Sociological Quarterly 20:447–94.

Hoge, Dean R. and Jackson W. Carroll. 1978. “Determinants of Commitment and Participation in Suburban Protestant Churches.” Journal for the Scientific Study of Religion 17:107–27.

Hoge, Dean R. and David T. Polk. 1980. “A Test of Theories of Protestant Church Participation and Commitment.” Review of Religious Research 21:315–29.

Hougland, James G. and James R. Wood. 1980. “Correlates of Participation in Local Churches.” Sociological Focus 13:343–58.

Jones, Lynne M. and Claude S. Fischer. 1978. “Studying Egocentric Networks by Mass Survey.” Working paper No. 284, Social Density and Mental Health. Unpublished manuscript.

Kanter, Rosabeth Moss. 1972. Commitment and Community. Princeton, NJ: Harvard University Press.

King, Morton B. and Richard A. Hunt. 1972. Measuring Religious Dimensions: Studies in Congregational Involvement. Dallas: Southern Methodist University.

Lenski, Gerhard E. 1963. The Religious Factor: A Sociologist’s Inquiry. New York: Anchor.

Lofland, John and Rodney Stark. 1965. “Becoming a World-Saver: A Theory of Conversion to a Deviant Perspective.” American Sociological Review 30:862–75.

Lyon, David. 1985. “Rethinking Secularization: Retrospect and Prospect.” Review of Religious Research 26:228–43.

McCallister (Jones), Lynne and Claude S. Fischer. 1978. “A Procedure for Surveying Personal Networks.” Sociological Methods and Research 7:131–48.

Martin, David. 1978. A General Theory of Secularization. Oxford: Blackwell.

Marty, Martin E. 1985. “Transpositions: American Religion in the 1980s.” The Annals 480:11–23.

Mueller, Charles W. and Weldon T. Johnson. 1975. “Socioeconomic Status and Religious Participation.” American Sociological Review 40:785–800.

Nelsen, Hart M. and Raymond H. Potvin. 1980. “Toward Disestablishment: New Patterns of Social Class, Denomination and Religiosity among Youth?” Review of Religious Research 22:137–54.

Niebuhr, H. Richard. 1929. The Social Sources of Denominationalism. New York: Meridian.

Pope, Liston. 1942. Millhands and Preachers. New Haven, CT: Yale University Press.

Roberts, Michael K. and James D. Davidson. 1984. “The Nature and Sources of Religious Involvement.” Review of Religious Research 25:334–50.

Shiner, Larry. 1967. “The Concept of Secularization in Empirical Research.” Journal for the Scientific Study of Religion 6:207–20.

Stark, Rodney and William Sims Bainbridge. 1985. The Future of Religion: Secularization, Revival, and Cult Formation. Berkeley: University of California Press.

Stark, Rodney and Charles Y. Glock. 1968. American Piety: The Nature of Religious Commitment. Berkeley: University of California Press.

Troeltsch, Ernest. 1931. The Social Teaching of the Christian Churches. Translated by Olive Wyon. New York: Macmillan.

Welch, Kevin W. 1981. “An Interpersonal Influence Model of Traditional Commitment.” The Sociological Quarterly 22:81–92.

Wellman, Barry. 1979. “The Community Question: The Intimate Networks of East Yorkers.” American Journal of Sociology 84:1201–31.

White, Richard H. 1968. “Toward a Theory of Religious Influence.” Pacific Sociological Review 11:23–28.

Wilson, Bryan. 1966. Religion in Secular Society. Harmondsworth, UK: Penguin.


[1] Data collection was funded by The Church of Jesus Christ of Latter-day Saints. This paper has benefited from the helpful comments of Darwin L. Thomas and anonymous referees. Direct correspondence to the author, Department of Sociology, Brigham Young University, Provo, UT 84602.

[2] Obviously, there are feedback loops within the model. For example, while personal community relationships influence religious beliefs and commitment, belief and commitment are also likely to influence personal community relationships. Unfortunately current empirical analysis does not allow us to fully test the reciprocal effects. While a LISREL model would seem most appropriate, the inclusion of all nonrecursive paths produces a non-identified model.

[3] We have chosen to use family as a dummy variable for two reasons. First, in earlier research we made a distinction between two-parent families where both parents were LDS, two-parent families where only one parent was LDS, and single-parent families. We found that single-parent and interfaith families, no matter how religious, were less able to channel their children into religious activities which would help them develop and maintain a religious identity (see Cornwall 1988). It may be that single parents and member parents in an interfaith marriage are less able to provide strong religious socialization within Mormonism because they themselves are not adequately integrated into the social system. This may be a finding unique to Mormonism where there is a strong emphasis on traditional family life.

[4] Path coefficients in the model vary slightly from the betas presented in table 11.1 because some variables were deleted from the model for the sake of simplicity.