Religious Switching: Preference Development, Maintenance, and Change

Religious Switching: Preference Development, Maintenance, and Change MATTHEW T. LOVELAND Up to one-third of Americans switch religions at some time during their lives. What are the predictors of this religious mobility? This article addresses this question using a modified rational choice framework to explain the development, maintenance, and change of religious preferences. Although classical rational choice theory assumes that preferences are stable, this article suggests that preferences are variable and that social interaction works to maintain or change an individual s preferred religious choice. A cultural theory of preference formation is applied to allow for the social constraint rational choice theory often ignores. Findings suggest that childhood socialization does not cement religious preference, that formally joining a church while growing up acts to stabilize preferences, that lapsing in practice increases the likelihood of switching, and that distinctive denominations generate religious preferences that reduce individual switching. INTRODUCTION Although most people remain adherents of one religious faith throughout their lifetime, nearly one-third of U.S. adults have switched religions at least once (Roof 1989). In this article, I ask the following question: Why do people switch 1 their religious identification? I draw on recent rational choice literature and on earlier work on this question. Rational choice theorists have developed a more contextualized conception of the actor, and recent work has embedded switchers in a social network. The task is not complete, however, and I use Wildavsky s (1987, 1994) work in cultural theory to demonstrate that individuals preferences are dependent on decisions they make. REVIEW OF RESEARCH Studies of religious switching are greatly influenced by Stark and Glock s 1968 book American Piety: The Nature of Religious Commitment, in which they argue that the tendency of religious switchers is to move from theologically conservative to theologically liberal denominations in search of the higher social status liberal denominations confer on members. This proposition has been questioned in the literature. Kelley argued nearly the opposite in Why Conservative Churches are Growing (1972), theorizing that stricter churches better serve the individual s quest for meaning and for this reason would grow while more liberal churches would decline. Iannaccone (1994) argues that strict churches will be more likely to grow because they are capable of offering more benefits to members than are churches that cultivate less commitment. While strict churches retain members, liberal churches will not. According to these analysts, it is attitudes toward doctrinal strictness that cause individuals to switch religions. Some propose that focusing on the ideological nature of religion is too simple and that denominational switching occurs with little reference to a denomination s position on a liberal to conservative spectrum (Mueller 1971). Roof and Hadaway (1979) found much diversity in religious switching, including shifts of the upwardly mobile to high-status churches, and others out of religion all together. They introduced a mainline-fringe institutional typology to conceptualize the nature of religious mobility. A church s position relative to the culture at large is key to determining the mainline or fringe status. Particularly important are the moral beliefs Matthew T. Loveland, University of Notre Dame, Department of Sociology, 810 Flanner Hall, Notre Dame, IN 46556. E-mail: loveland.1@nd.edu Journal for the Scientific Study of Religion 42:1 (2003) 147 157

148 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION and behavioral styles of a religious group, and how the group is perceived by others: Mainline if regarded generally as congruent with the culture at large; fringe if not, especially if the group fosters distinctively different beliefs and behaviors (Roof and Hadaway 1979:370). Losses among mainline churches are shown to be greater than losses among the fringe groups, and much of mainline switching is between several mainline denominations. This suggests that the boundaries between the mainline groups are more permeable than are the boundaries around the fringe groups. The most important contribution of these studies is the recognition that religious switching is a complex phenomenon with many determinants. A much-cited determinant is socialization into a religious faith (Sullins 1993; Hadaway and Marler 1993), as it has been argued that those who are more religiously active as youth are less likely to switch. Sherkat and Wilson (1995) suggest that parents successful transmission of religious beliefs and practices plays a significant role in determining the likelihood of switching religions because those who hold weaker beliefs as youth have a tendency to break religious ties. Sherkat (1991), however, tested the hypothesis that Sunday School attendance or other formal religious training as a child would reduce the likelihood of switching and found that it did not. Thus, the notion that childhood socialization predicts religious switching is open to debate. Marriage is an oft-cited determinant of religious switching (Roof 1989; Sherkat 1991; Hadaway and Marler 1993; Newport 1979; Musick and Wilson 1995). Individuals who marry outside their religion of origin (religious exogamy) are believed to be more likely to switch. Hadaway and Marler (1993) theorize that individuals in exogamous marriages switch in order to harmonize the marriage. Iannaccone (1990), a rational choice advocate, argues that couples sharing a religious preference would have an easier time producing religious capital, and for this reason one of the partners in the exogamous marriage would be likely to switch to the religion of his or her spouse. Sherkat (1991) includes denomination of origin, lapses in religious practice, and geographic mobility among the most important determinants of religious switching. This research suggests that stricter religions have lower rates of switching than less strict religions. Although this claim is reminiscent of Kelley s, Sherkat (1991) indicates simply that switching out of strict religions will be less likely and not that strict religions will necessarily grow. Individuals who lapse in religious practice are thought to use this time away from religion to become aware of alternatives in religious practice. Sherkat hypothesizes that geographic mobility breaks ties individuals have with their original faith and creates situations in which ties to a new religion can form. Related to this proposition is the idea that those who frequently socialize with relatives are less likely to break ties and thus are less likely to switch. A number of other factors have been cited as possible determinants of religious switching. Sandomirsky and Wilson (1990) suggest that males are more likely to be religious switchers than females. Roof summarized findings based on the 1988 GSS, finding that switchers tend to be male and well educated (Roof 1989). Ellison and Sherkat (1990) claim that social embeddedness, cultural solidarity, and personal spirituality are more important to African Americans than are status-seeking and sociopolitical concerns so that members of this community are less likely to switch religions. KEY CONCEPTS OF RATIONAL CHOICE THEORY The sociology of religion recently experienced what one author called a paradigm shift (Warner 1993) to the use of rational choice theory to explain both institutional and individual religious behavior. The major proponents of this theoretical frame argue that the theory explains a wide range of religious activity based on a few powerful assumptions and propositions (Becker 1976; Iannaccone 1990, 1995). According to Becker (1976), The combined assumptions of

RELIGIOUS SWITCHING 149 maximizing behavior, market equilibrium, and stable preferences, used relentlessly and unflinchingly, form the heart of the economic approach (Becker 1976:5). The maximizing assumption that human behavior results from individuals seeking to maximize benefits and avoid costs is the most well known of the rational choice assumptions. The second assumption of rational choice theory is that of stable preferences. The desires of individuals are expected to remain fairly constant through time so that the individual simply adapts choices to the environment based on steady tastes. Following the assumption of stable preferences, the rational choice theorist is almost never content to explain such changes with reference to changed tastes, norms, or beliefs (Iannaccone 1995:77). PREFERENCES IN SOCIAL CONTEXT Sherkat and Wilson (1995) provide the most complete rational choice analysis of religious switching to date. The study examines the development of cultural preferences as a result of social status and prior religious consumption. The possibility of normative constraints and social relationships hindering individual religious choice is also introduced. Traditional rational choice theory posits a fully informed teleological actor who makes decisions with regard only to personal preferences, and as a result disregards the role social relationships play in influencing the decisions of actors. Sherkat and Wilson move beyond this limitation by suggesting that choices may be made to please others or avoid their wrath (1995:998). Using the concept of adaptive preferences, they state: individuals come to desire that with which they are familiar. Familiarity comes from proximity to others who also have these preferences and from the occupancy of roles and statuses with which those preferences are associated (Sherkat and Wilson 1995:996). Socialization through childhood religious practice is argued to create preferences upon which later religious choices are made. With the introduction of adaptive preferences, they begin to analyze what rational choice theorists have left as an assumption, making the claim that individual tastes should not be considered beyond the scope of scientific enterprise (Sherkat and Wilson 1995:994). Wildavsky (1987, 1994) has examined political preference formation and expression within an institutional and cultural context. Wildavsky argues that the rational choice assumption of maximizing behavior, or self-interest, is misleading when it assumes that individuals are free to adopt any interest-cum-preference they want without taking others into account (1994:138). As noted above, the individual is seen by classical rational choice theory as free to satisfy his or her preferences, with self-gratification as the only influential factor. Cultural theory, however, situates the actor within a social realm, where individual behavior will be influenced and constrained by the setting. Wildavsky continues by stating that selves are socially constructed, and the interest of the self depends on the cultural context within which the individual is situated (1994:140). He further concludes that actors make sense of their world through their social relations (Wildavsky 1994:151). It is apparent from this perspective that preferences are formed through social relations and institutional arrangements (Wildavsky 1987:4). Cultural theory...is based on the premise that preferences are endogenous...so that they emerge from social interaction in defending or opposing different ways of life (Wildavsky 1987:5). Preferences, then, are not beyond the scope of sociological inquiry. If preferences emerge from social interaction within a cultural context, we would expect that an individual s preferences will not be shaped only through childhood socialization into a religious faith (Sherkat and Wilson 1995), but that the individual s preferences may change when social relationships are altered throughout life. As underlying preferences are maintained or altered, so too will be religious choices, including religious identification. The following study seeks to examine further the development and change of religious preferences within a social context and, by so doing, refine sociological explanations of the rational choices actors make.

150 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION HYPOTHESES AND CONTROLS From the rational choice perspective, childhood socialization represents a period of preference formation when the child becomes familiar with religious belief and practice, and people come to prefer that with which they are familiar. As these preferences are assumed to be stable, changes should be unlikely. This suggests Hypothesis 1: Greater levels of childhood religious socialization will tend to decrease the likelihood of switching. Hypothesis 2 will test Wildavsky s theory of preference formation through the support of an institution. The act of formally joining a religion while growing up is measured with the 1988 GSS variable JOIN16. Respondents were asked, Did you ever join a church (synagogue) when you were growing up, that is, become a member by confirmation or such? Individuals who join a church are seen as supporting the institution through agreement with its values and in this way solidifying their religious preferences. Thus, Hypothesis 2 is stated: Individuals who join a church while growing up will tend to be less likely to switch religions. In contrast to supporting an institution, opposition creates preferences at odds with the institution. Opposition to a religious institution, such as a church, may be expressed in avoiding what is essentially voluntary attendance. An individual who chooses not to attend religious services may be implicitly opposing the values of the institution, and this may generate preferences that conflict with the religious products provided by this institution. In opposing the institution, the individual develops new preferences, which may encourage switching. Hypothesis 3 can be stated: Individuals who lapse in religious practice are more likely to switch than those who do not lapse. The fourth hypothesis concerns denominational differences in religious switching: Individuals raised in distinct religious denominations are less likely to switch than those raised in denominations that are more mainstream. Following Roof and Hadaway (1979), distinctive denominations are those that are culturally or stylistically unique from mainstream American religion or culture. Individuals raised in these distinctive social contexts are expected to be less likely to switch because they will have religious preferences that are difficult to satisfy outside of the denomination of origin. Roof and McKinney categorize the Mormons, Jehovah s Witnesses, Christian Scientists, and the Unitarian-Universalists as having distinctive religious styles, and separate these bodies from other Protestant denominations (Roof and McKinney 1987:97). Jewish adherents are also a distinctive religious body in contrast to America s Christian heritage (Roof and McKinney 1987). Those classified as others in the GSS are also considered distinctive. Although Sherkat (1999) raises the issue of Christians being improperly categorized as others in the GSS coding, it can be assumed that over half of these respondents are non-christian and are therefore distinct in the American religious sphere. 2 Roof and McKinney (1987) place Catholicism in the American cultural mainstream. However, the Catholic faith remains distinct from the Protestant religious culture in stylistic ways. An example of this stylistic difference can be found in the contrast between the celibate priesthood of Catholicism and the married ministers of Protestantism. Morris explains this Catholic distinction in the following way, the priest is a man apart and above, the consecrated, sacred minister (1997:407). This understanding of the priest as mediator, as above laity, is but one way Catholicism is stylistically distinct from the dominant Protestant culture. A final hypothesis examines whether frequency of contact with relatives stabilizes an individual s religious preference. As preferences are presumably formed through childhood socialization, maintaining contact with one s family should help to cement an individual s preferences. Conversely, individuals who spend less time with family are freer to experience new interactions, which may allow them to form new preferences. Accordingly, Hypothesis 5 is stated: Individuals who spend less time with their relatives are more likely to switch religions. I control for several predictors identified in previous research. Moving away from a childhood hometown will likely involve encountering new people and create new social connections.

RELIGIOUS SWITCHING 151 If preferences are socially constructed through the interactions of an individual, it can be expected that new social relationships may alter old preferences. Therefore, I control for geographic mobility with a measure representing those who have moved away from their childhood hometown. Preferences may also be vulnerable to new information about alternatives (Sherkat and Wilson 1995). As the individual becomes aware of religious possibilities in the environment, he or she may come to prefer the religious goods of competing religious institutions. As a way to capture this variation, I control for educational attainment. I also control for the following identified determinants of switching: religious exogamy, age, sex, race, and conservative Protestantism. Lastly, the model includes a control variable for individuals who were raised with no religious preference. DATA AND METHOD The data used in the following analysis come from the 1988 General Social Survey (GSS). In 1988 the GSS included a module on religion containing several questions about religious switching. Therefore, the data provide a well-designed instrument to answer the research question. Beyond the usual questions about the respondent s childhood religion and current religious preference, the 1988 module included a question allowing direct analysis of religious switching: Have you ever had another religious preference besides being (respondent s answer to question about religious preference)? This question allows for a detailed analysis of religious switching. 3 With missing cases removed, the sample size is 1,425 respondents. The dependent variable is dichotomous, coded 1 for individuals who have switched religion at the age of 16 or older. Switches before the age of 16 are excluded because it is assumed these switches result from changes imposed by the respondent s family rather than choices made by the individual. 4 Thirty-three percent of the respondents have switched religions since the age of 16 (see Table 1). The childhood socialization hypothesis will be tested with measures of religious school attendance, Sunday School attendance, and family prayer, predicting that greater levels of each TABLE 1 DESCRIPTIVE STATISTICS N Mean SD Min. Max. Switched religion 1,425 0.33 0.47 0 1 Years of religious school 1,425 1.34 3.19 0 15 Regular Sunday School attendance 1,425 0.55 0.50 0 1 Family regularly said grace 1,425 0.67 0.47 0 1 Formally joined a church while growing up 1,425 0.68 0.47 0 1 Lapsed in religious practice 1,425 0.55 0.50 0 1 Geographic mobility 1,425 0.59 0.49 0 1 Years of education 1,425 12.54 3.12 0 20 Exogamy 1,425 0.15 0.35 0 1 Socialize with relatives rarely 947 0.31 0.46 0 1 Age 1,425 45.20 18.20 18 89 Male 1,425 0.43 0.50 0 1 African American 1,425 0.12 0.33 0 1 Protestant 1,425 0.45 0.50 0 1 Raised conservative Protestant 1,425 0.16 0.37 0 1 Raised Catholic 1,425 0.28 0.45 0 1 Raised in distinctive denomination 1,425 0.07 0.25 0 1 Raised no religions preference 1,425 0.04 0.19 0 1

152 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION produce more particular preferences and reduce the likelihood of a religious switch. Respondents were asked, How many years did you attend religious schools? Sunday School attendance is measured by the question, When you were growing up, did you attend Sunday school or religious instruction classes regularly, most of the time, some of the time, or never? Respondents who attended regularly or attended parochial school are coded 1; all others are coded 0. Family prayer is measured by the GSS variable GRACE16, which asked, When you were growing up, did anyone usually say grace or give thanks to God aloud before meals? Respondents who said yes are coded 1 (0 = no). To measure the defense of an institution for Hypothesis 2, responses to the question, Did you ever join a church (synagogue) when you were growing up, that is, become a member by confirmation or such? are coded dichotomously. Individuals who answered yes are coded 1; all others are coded 0. Opposition to an institution through lapsing is measured by the GSS variable LAPSED, which asked, Since you were 16 years old, has there ever been a period of 2 years or more when you did not attend religious services? Those who have lapsed are coded 1 and those who have not are coded 0. The denominational hypotheses are measured by the GSS variables, RELIG16, DENOM16, and OTH16, which asked respondents to indicate the religion in which they were raised. Childhood denomination (RELIG16 or DENOM16) is used to capture the period of preference formation. RELIG16 classifies respondents as Protestant, Catholic, Jewish, none, or other. Protestants are then asked to specify a particular denomination in DENOM16 and OTH16. Dichotomous variables are created for Catholics and those in the distinctive denominations. The reference group for these categories consists of all Protestant respondents not classified in either the distinctive group or conservative Protestant category. The final hypothesis is tested with the GSS measure SOCREL, which asks respondents, How often do you spend a social evening with relatives? Possible responses to this question range from almost every day to never. Individuals who spend an evening with relatives less than once a month are coded 1; those who socialize with relatives monthly or more often are coded 0. To control for geographic mobility, respondents are asked, When you were 16 years old, were you living in this same (city/town/country)? Respondents who no longer live in the same town as at age 16 are coded 1; those who live in the same town are coded 0. Years of education are measured by the GSS variable EDUC. The exogamy variable is measured by comparing a respondent s childhood religion to the respondent s spouse s childhood religion (Sherkat 1991). A dichotomous variable is coded 1 if the respondent s childhood religion differs from the respondent s spouse s childhood religion (0 = other). 5 Race is a dichotomous variable coded 1 for African American and 0 for others. A squared-age term is included in the model to account for the fact that most switching occurs in young adulthood and then decreases with age (Sherkat 1991). A dichotomous variable for conservative Protestantism is included to control for the strict church perspective (Kelley 1972; Iannaccone 1994; Sherkat 2001), and the measure of no religious preference is also dichotomous (1 = no religious preference, 0 = others). Two models will be estimated with logistic regression, predicting the log-odds of religious switching at the individual level. The first model includes all predictors except for the measure of time spent with relatives. The second model includes this measure, and is run on a smaller sample size because the question was asked of only a limited number of respondents. RESULTS The childhood socialization hypothesis receives no support in Model 1. None of the measures attendance at a religious school, Sunday School attendance, or saying grace before meals is a significant deterrent of religious switching. Hypothesis 2, concerning formally joining a church, is supported. Respondents who joined a church during childhood are 38 percent less likely to switch later in life than those who did not join a church. Hypothesis 3 is also supported,

RELIGIOUS SWITCHING 153 TABLE 2 PREDICTORS OF RELIGIOUS SWITCHING Model 1 Model 2 Socialization Years of religious school 0.007 0.024 (0.025) (0.030) Regular Sunday School attendance 0.173 0.091 (0.131) (0.163) Family regularly said grace 0.075 0.138 (0.140) (0.175) Institutional Formally joined a church while growing up 0.473 0.553 (0.139) (0.172) Lapsed in religious practice 0.455 0.470 (0.127) (0.157) Denominational Raised Catholic 1.133 1.348 (0.186) (0.231) Raised in distinctive denomination 0.924 0.859 (0.261) (0.321) Controls Geographic mobility 0.400 0.214 (0.128) (0.161) Age 0.050 0.033 (0.021) (0.025) Age squared 0.043 0.028 (0.021) (0.025) Education 0.058 0.063 (0.022) (0.027) Male 0.025 0.078 (0.123) (0.152) African American 0.440 0.423 (0.195) (0.235) Exogamy 0.488 0.580 (0.171) (0.209) Raised conservative Protestant 0.479 0.649 (0.173) (0.218) Raised no religious preference 0.244 0.570 (0.320) (0.382) Continued Socialization Socialize with relatives rarely 0.528 (0.161) Constant 2.409 2.078 (0.543) (0.661) Observations 1,425 947 Chi square (prob > Chi2) 164.89 (0.000) 127.09 (0.000) Note. Standard errors in parentheses; p 0.05; p 0.01.

154 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION as lapsing in religious practice significantly increases the log-odds of switching. Individuals who have lapsed are 58 percent more likely to switch than are individuals who have not lapsed. Hypothesis 4 concerns denomination of origin and tests whether individuals raised in distinctive denominations are less likely to switch religions. This hypothesis is supported for both Catholics and the members of distinctive bodies. Individuals raised in distinctive groups are 60 percent less likely to switch religions than are individuals raised in mainstream Protestant bodies. Catholics are also less likely to switch, changing religious identification 68 percent less than the Protestant reference group. The final hypothesis, concerning socializing with relatives, is supported by Model 2, which was run on a subsample of 947 respondents. Those who socialize with relatives less than once a month are 70 percent more likely to switch religions than are those who more frequently spend time with their relatives. The control variables all behave as expected. Geographic mobility is a significant predictor of switching. Individuals who move from the town they lived in at age 16 have a 0.40 probability of switching. Education significantly increases the likelihood of switching, with each additional year of schooling increasing the log-odds of switching by 6 percent. Marrying outside the origin faith increases the likelihood of switching by 63 percent. Age, race, and conservative Protestantism are all significant predictors of religious switching. The first-order age term is significant and positive, and the squared-age term is significant and negative, revealing that, as expected, the likelihood of switching increases in early adulthood and then decreases as people get older. In line with previous literature, African Americans were shown to be 36 percent less likely than others to switch religions. Conservative Protestants are also less likely to switch than are the more mainstream Protestant denominations; being raised conservative Protestant reduces the likelihood of switching by 38 percent. The male and no religion variables were not significant. DISCUSSION AND CONCLUSIONS Childhood socialization does not appear to produce lasting religious preferences. Those for whom childhood was a time of religious involvement and education are no less likely to switch than are those who spent less time learning religious ways of life. When social behaviors that would influence preferences later in life are allowed to predict switching, childhood socialization is not an influential determinant of religious choice. 6 Wildavsky s insights about the endogenous nature of preferences makes it necessary to include the individual s ongoing relationship with religious institutions later in life if we are to understand the nature of religious choice. The hypotheses tested presented preference formation as the result of defending or opposing institutions, suggesting that formally joining a church implies a statement of agreement with, and a defense of, the church s values. Thus, it can be argued that those who joined churches while growing up are less likely to switch religions later in life because their preferences have stabilized in defending the institution through the formal act of joining. Conversely, lapsing in religious practice can be seen as an implicit act of opposition to the institution, which generates preferences at odds with the religious products of the institution and leads individuals away from their faith of origin. Those who spend little time with relatives are indeed more likely to switch. Again, this suggests the importance of social relationships throughout life for the maintenance of religious preferences. Preferences are formed and sustained by the social relationships people maintain or discard. The rational choice concept of preference must include this understanding to portray completely the active decisionmaker within a social setting and account for Wildavsky s claim that preferences are the result of living with other people (1987:3). Finally, the denominational hypotheses show that Catholics and members of distinctive denominations are less likely to switch. When religious bodies create preferences for unique religious goods, it will be difficult for individuals to find satisfactory religious products elsewhere. Research in voluntary association membership uses the concept of niche overlap to describe organizations

RELIGIOUS SWITCHING 155 that compete for similar members (McPherson 1983; Baum and Singh 1994; Popielarz and McPherson 1995). Niche overlap refers to a situation in which organizations compete for members from the same population. Mainline Protestant churches face the negative consequences of niche overlap as members of these churches leave for similar denominations. For the individual who belongs to one of the overlapping churches, many options are available when searching to satisfy religious preferences. For those raised in the distinctive groups (denominations filling a unique niche), the possibility of finding a new religious choice that will satisfy developed preferences is smaller. Sherkat (2001) uses this approach to show how ecumenical denominations face competition not only from other religious bodies, but also from secular organizations. He reports that the Episcopal church, while liberal, has a higher retention rate than do other ecumenical denominations, noting that the institutionalization of liturgy and ritual in the Book of Common Prayer and the performance of the Eucharist allow it to retain particular and distinctively religious elements (Sherkat 2001). This study s similar findings about the Catholic Church and the other distinctive bodies support the introduction of the concept of niche overlap into the switching literature. Using a rational choice approach with a dynamic notion of preferences provided several insights about religious switching. It has been argued, contrary to much of the existing literature, that childhood socialization does not create lasting preferences. Preferences are not cemented during childhood religious socialization, but are vulnerable to influence later in life. This is demonstrated in the finding that socializing with family throughout one s lifetime is inversely related to the likelihood of switching religions. Also, defense of, and opposition to, religious institutions were shown to be powerful preference creators. Importantly, the present study reinforces the finding that individuals raised in distinctive denominations, which create and supply satisfaction of unique preferences, are less likely to switch to new religions. Rational choice is a powerful theory when applied to religious switching because it constructs the individual as an active agent making choices between religious options. However, it is equally important to consider the social context of preference formation, and to understand preferences as open to the continuing influence of social life. This article continues the work of Sherkat and Wilson (1995) in formulating a sociologically grounded theory of preference formation. The concept of adaptive preferences is useful in understanding preference formation, but perhaps less well equipped to detect preference change. Wildavsky s (1987, 1994) approach, with its emphasis on social interaction, provides a useful way to analyze both the development and change of preferences. An emphasis on the role of social interaction will allow rational choice theorists to better understand the dynamics of religious switching. ACKNOWLEDGMENTS I would like to thank David Sikkink, Kevin Christiano, David Yamane, Michael Welch, Jamie Przybysz, David Ortiz, and two anonymous reviewers for helpful comments about earlier drafts of this paper. NOTES 1. Stark and Finke (2000) make a distinction between conversion and reaffiliation, such that conversions are shifts across religious traditions, and reaffiliation denotes shifts within religious traditions. For the purposes of this study, each type of shift will be considered, so as to capture the complete spectrum of religious life in America. Thus, a switch simply denotes a change in religious affiliation (Sherkat and Wilson 1995). 2. Sherkat (1999) reports that after reclassifying respondents who reported other religion in the GSS surveys from 1982 1998, 1.1 percent of all respondents are non-christian. Using this percentage to determine the composition of the others in the 1988 survey (0.011 1425), reveals that approximately 16 of the 30 others in the final sample are non-christian.

156 JOURNAL FOR THE SCIENTIFIC STUDY OF RELIGION 3. It should be noted that the goal of this research is not to estimate current population parameters, but to speak to general social processes of affiliation and disaffiliation, processes that can be assumed to have remained stable over the last 15 years. 4. Several scholars have pointed out that childhood religious life is vulnerable to parental influence, and that measures of childhood religiosity largely reflect conformity to parents wishes (Welch, Tittle, and Petee 1991). 5. This measure has some limitations. It is conceivable that a respondent may have switched prior to marriage, so that his or her religion during childhood was different than his or her spouse s childhood religion but the same at the time of marriage. Despite this problem, exogamy has been shown to be a powerful predictor of switching and this construction of the variable has been used previously in the literature. 6. There are some difficulties with these measures of socialization. The data show, in an unreported analysis, that Catholic youth spend more time in religious schools than do members of other faiths, an average of nearly four years compared to an average of less than one-third of one year for others. Related to this, Sherkat (1991) suggests that the concept of Sunday School may be meaningless to many people. Better measures of socialization, which capture more fully childhood religious experience, may reveal different patterns. In a separate model, mother s religious attendance was used as an additional socialization measure; this produced no changes in the results. The measures chosen reflect an interest in the religious behavior of the respondents. REFERENCES Baum, J. A. C. and J. V. Singh. 1994. Organizational niches and the dynamics of organizational mortality. American Journal of Sociology 100:346 80. Becker, G. S. 1976. The economic approach to human behavior. Chicago, IL: University of Chicago Press. Ellison, C. G. and D. E. Sherkat. 1990. Patterns of religious mobility among black Americans. Sociological Quarterly 31:551 68. Hadaway, C. K. and P. L. Marler. 1993. All in the family: Religious mobility in America. Review of Religious Research 35:97 116. Iannaccone, L. R. 1990. Religious practice: A human capital approach. Journal for the Scientific Study of Religion 29:279 314.. 1994. Why strict churches are strong. American Journal of Sociology 99:1180 1211.. 1995. Voodoo economics? Reviewing the rational choice approach to religion. Journal for the Scientific Study of Religion 34:76 88. Kelley, D. M. 1972. Why conservative churches are growing: A study in the sociology of religion. New York: Harper and Row. McPherson, M. 1983. An ecology of affiliation. American Sociological Review 48:519 32. Morris, C. R. 1997. American Catholic: The saints and sinners who built America s most powerful church. New York: Random House. Mueller, S. A. 1971. Dimensions of interdenominational mobility in the United States. Journal for the Scientific Study of Religion 10:76 84. Musick, M. and J. Wilson. 1995. Religious switching for marriage reasons. Sociology of Religion 56:257 70. Newport, F. 1979. The religious switcher in the United States. American Sociological Review 44:528 52. Popielarz, P. A. and J. M. McPherson. 1995. On the edge or in between: Niche position, overlap, and the duration of voluntary association memberships. American Journal of Sociology 101:698 720. Roof, W. C. 1989. Multiple religious switching: A research note. Journal for the Scientific Study of Religion 28:530 35. Roof, W. C. and C. K. Hadaway. 1979. Denominational switching in the seventies: Going beyond Stark and Glock. Journal for the Scientific Study of Religion 18:363 77. Roof, W. C. and W. McKinney. 1987. American mainline religion: Its changing shape and future. New Brunswick, NJ: Rutgers University Press. Sandomirsky, S. and J. Wilson. 1990. Processes of disaffiliation: Religious mobility among men and women. Social Forces 68:1211 29. Sherkat, D. E. 1991. Leaving the faith: Testing theories of religious switching. Social Science Research 20:171 87.. 1999. Tracking the other : Dynamics and composition of other religions in the General Social Survey, 1973 1996. Journal for the Scientific Study of Religion 38:551 60.. 2001. Tracking the restructuring of American religion: Religious affiliation and patterns of religious mobility, 1973 1998. Social Forces 79:1459 93. Sherkat, D. E. and J. Wilson. 1995. Preferences, constraints, and choices in religious markets: An examination of religious switching and apostasy. Social Forces 73:993 1026. Stark, R. and R. Finke. 2000. Acts of faith. Berkeley, CA: University of California Press. Stark, R. and C. Glock. 1968. American piety: The nature of religious commitment. Berkeley, CA: University of California Press.

RELIGIOUS SWITCHING 157 Sullins, P. D. 1993. Switching close to home: Volatility or coherence in Protestant affiliation patterns? Social Forces 72:399 419. Warner, S. R. 1993. Work in progress toward a new paradigm for the sociological study of religion in the United States. American Journal of Sociology 98:1044 93. Welch, M. R., C. R. Tittle, and T. Petee. 1991. Religion and deviance among adult Catholics: A test of the moral communities hypothesis. Journal for the Scientific Study of Religion 30:159 72. Wildavsky, A. 1987. Choosing preferences by constructing institutions: A cultural theory of preference formation. American Political Science Review 81:3 22.. 1994. Why self-interest means less outside of a social context. Journal of Theoretical Politics 6:131 59.