Since the 1960s every decade has experienced a great cult scare. From the Manson family to the Moonies, the Waco sect to the Jonestown massacre, the activities of cults prompt great public concern. Interest groups such as the American Family Foundation and the Cult Awareness Network devote their resources to monitoring the activities of deviant religious groups. The media continually warn of the dangers of cultic brainwashing strategies and the FBI continues to monitor closely religious groups they perceive to be potential dangers. Despite the great public and governmental attention afforded to cults and sects, however, researchers still know relatively little about why individuals join such groups.
This lack of cumulative knowledge about cults and sect affiliation stems from the absence of a testable theory of religious membership. Psychiatrists have tended to examine cult and sect affiliation from a victimization perspective, assuming that cult leaders somehow overcame an individual's free will. Sociological studies have rejected the brainwashing model, focusing instead on the interactions between potential and established cult members. Nevertheless, the absence of a coherent frame work and confusion about terms has hampered such studies.
A potential answer to the lack of a theoretical framework lies with the Stark-Bainbridge theory of religion. In what is certainly the most ambitious attempt at religious theorizing to date, Stark and Bainbridge attempt to explain everything from the nature of God to the process of secularization, using clearly defined terms and carefully outlined propositions. Given the potential importance of the Stark-Bainbridge theory, sociologists need to meet Stark and Bainbridge's challenge to test it (Stark and Bainbridge 1987: 13). In this paper we will test Stark and Bainbridge's propositions about one area of particular interest to sociologists, psychiatrists, and the general public - conversion to sects and cults.
PREVIOUS RESEARCH ON CULT AND SECT AFFILIATION
Previous research exploring the reasons why people join religious cults and sects has been sharply divided along professional lines. Psychiatrists and therapists have compiled a vast literature on deviant religious movements, much of which argues that people join cults because they have been "brainwashed." Sociologists in general have been critical of the brainwashing model and have attempted to explain conversion to cults and sects via sociological concepts such as role theory and organizational theory.
The psychological literature is rife with discussions of brainwashing and its supposed cure, "deprogramming" (see for example: Edwards 1981; Goldberg 1983; Keiser and Keiser 1987; and Levine 1980). Popular books on religious cults focus almost exclusively on the brainwashing model. Sargant (1957), for example, argues that cult leaders change their victims' belief systems via "sensory overload," while Appel (1983) discusses in great detail the process by which cult leaders isolate potential converts, break their wills and, finally, force a new belief system upon them. The brainwashing model assumes that cults somehow suppress potential converts' ability to reason, thus ensnaring them into the group. Cults supposedly facilitate brainwashing by using such methods as sensory deprivation, physical threats, unbalanced diets, repetitive chanting, and childish games (Beckford 1985: 95). Under the brainwashing model, converts to cults and sects are viewed as victims. Cult and sect leaders are villainous manipulators whose "ultimate objective is to take control of the minds and lives of their members" (Levine 1980: 35).
While psychiatrists tend to view conversion to cults and sects as the result of brainwashing, sociologists have tended to focus on the interactive processes within small religious groups. Numerous researchers have conducted participant observations and ethnographies of cults and sects. For example, Festinger, Riecken, and Schachter (1956) recorded the interactive processes in a millennial UFO cult, while Bainbridge (1978) studied The Process, a satanic cult (see also Balch 1976; Balch 1980; Damrell 1977; Scott 1980; Stillson 1974; Wallis 1975a, 1975b). While focusing on the interactive processes within cults and sects, sociologists have also been openly critical of the brainwashing model. Kirkpatrick (1988) and Ungerleider and Wellisch (1979) found no evidence that cult membership has a negative effect on members. Richardson and Kilbourne (1983) found numerous flaws in the brainwashing model, including the use of overgeneralized stereotypes that assume deception on the part of all cult leaders; the frequent comparison between cults and totalitarian governments, which does not take into account the ability of totalitarian governments to confine dissidents; the assumption that persons subjected to certain persuasion techniques lack free will; and the inherent sampling bias created by an over-reliance upon the accounts of former, possibly disgruntled converts (30) Other researchers have argued that claims of brainwashing are a form of repression whereby dominant religious traditions demonize smaller, less powerful groups (Kelley 1983; Robbins and Anthony 1978; Robbins, Anthony, and McCarthy 1983).
Sociologists have proposed several alternative explanations for cult membership. Balch (1980) argues that role theory provides an adequate explanation for cult membership. In his study of the Bo and Peep UFO cult, Balch found that members learned how to act as believers over a period of time, rather than undergoing a sudden conversion. Bromley and Shupe (1979), in their study of the Unification Church, also investigated conversion from the perspective of role theory. Pilarzyk (1978) examined conversion from a phenomenological perspective, concluding that conversions to sects and cults differ "in the degree of subjective change in world view, lifestyle and biography undergone by group members" (401). Suchman (1992) examined conversion using organizational theory, concluding that intermarriage, education, and age have a positive effect on religious mobility.
The abundance of research into cult and sect conversion, unfortunately, has failed to yield a coherent body of data. As Robbins (1988) argues:
Unfortunately, the research and theory in the area of conversion is redolent of multiple confusions related to the divergent promises, conceptual frameworks, nomenclature and behavioral referents which different researchers have employed. In consequence many studies are not strictly comparable (64).
Theoretical confusion extends even to the concepts of sect and cult. A variety of different definitions of sect and cult have been proposed by researchers. Many researchers define cults and sects by their organization - cults are fairly disorganized religious groups that may develop into a more organized form, the sect. Richardson (1979) argued that the Jesus Movement group had developed from a cult into a sect. Wallis (1975) argued that Scientology has developed from a loosely structured cult, into a more organized sect with a central authority. Other researchers have defined cults based on the dedication of their members. Goldberg (1983) defines a cult as a minority religious group with a "greater or excessive devotion or dedication to some person, idea or thing" (170). Yet another definition for a cult centers around the type of beliefs members hold. Several researchers, such as Campbell (177) and Jackson and Jobling (1968) define cults as religious groups holding "mystical" or occult beliefs. Given the widely varied conceptualizations of sects and cults, it is not surprising that researchers have been unable to yield coherent data.
THE STARK-BAINBRIDGE THEORY OF RELIGION
Sociologists Rodney Stark and William Sims Bainbridge have recognized the lack of a theoretical framework in religious research. In 1987, they published A Theory of Religion, which they hope will provide such a framework. Grounded in the classical traditions of exchange theory and control theory, the Stark-Bainbridge theory consists of 344 separate, testable propositions which seek to explain everything from secularization to affiliation with sects and cults. In fact, the Stark-Bainbridge model is the "only extensive formal deductive theory of religion" (Simpson 1990: 367).
Despite the obvious and potential importance for the sociology of religion, the Stark-Bainbridge theory has been virtually ignored by researchers. Other than critiques by Nock (1989) and Wallis (1986), the theory's propositions have yet to be tested. Therefore, this paper will test one component of the Stark-Bainbridge model - those propositions concerned with why people join sects and cults. A complete explication of the Stark-Bainbridge theory lies beyond the scope of this paper. Therefore, only those propositions necessary for understanding the theory's predictions about sect and cult membership will be discussed.
STARK AND BAINBRIDGE ON SECT AND CULT AFFILIATION
The Stark and Bainbridge theory provides yet another set of definitions for cult and sect. They conceptualize a church as a conventional religious organization, that is, an organization that supports the norms of its surrounding society. A sect movement is a "deviant religious organization with traditional beliefs and practices" (124). A cult movement is a "deviant religious organization with novel beliefs and practices" (124). Thus, sects are movements that split off from churches because of a dispute over beliefs and practices. Should members of a church disagree about how members should behave or what they should believe, those who lose the argument may form a sect. Cults, on the other hand, are new religious movements - movements that have not split from another religious organization. Cults may appear via two means, innovation and importation. Innovation occurs when an individual starts a new cult within a society, usually because he or she had what is believed to be a revelation. Importation occurs when a group that is well- established in one society is brought into another society (126-27). Thus, Stark and Bainbridge would consider Catholicism a cult, if missionaries brought it to a society that had yet to experience it.
The Stark-Bainbridge model views religion in utilitarian terms. In fact, the theory is based on the classical notions of exchange theory - "Humans seek what they perceive to be rewards and avoid what they perceive to be costs" (1987: 27). Some of the desired rewards, however, are limited in supply or are simply unavailable. When people greatly desire a reward that they can not acquire, argue Stark and Bainbridge, they will accept compensators for these rewards in the form of "explanations that are not readily susceptible to unambiguous evaluation" (36).
The concept of compensators is the key to the Stark-Bainbridge model. People are drawn to religion, they argue, because it provides compensation for a lack of some reward. Thus, for those in the lower levels of the stratification system, religion can provide a comforting explanation for a lack of riches by teaching that the poor will rule in heaven. Religious compensation is not limited to the poor, for there are some rewards, such as eternal life, which all people desire but are unable to obtain. However, proposition 18 of the theory reveals an important qualification - "Humans prefer rewards to compensators and attempt to exchange compensators for rewards" (37). In other words, if an individual somehow obtains riches, he or she will prefer those earthly riches to promises of riches in the afterlife.
As a religious movement develops, argue Stark and Bainbridge, there will be increasing tension within the group between those with high levels of power and those with low levels of power:
P132: The greater the degree of power inequality in a religious organization the greater will be the potential for group conflict over the distribution of rewards and the emphasis on compensators (136).
Those high in power will tend to have the most control over the movement, and per proposition 18, the powerful members will tend to prefer rewards to compensators. Therefore, the powerful will attempt to change the movement's beliefs in order to allow them their rewards. In other words, a person who can afford an expensive sports car will want a religion that allows him or her that sports car. As the powerful change a religious movement, those with less power will become disenchanted with that movement. Since they cannot attain the desired rewards, they will prefer the movement to focus on compensators. Should this tension continue, these disenchanted members will tend to split away from the movement.
Given this brief description of Stark and Bainbridge's theory, the following are specific propositions they make about who should be likely to join sects and cults:
P204: Persons with low stakes in conformity tend to have less favorable evaluations of conventional [religious] explanations than do persons with higher stakes in conformity (191).
P219: Persons who desire limited rewards that exist, but who lack the social power to obtain them, will tend to affiliate with sects, to the extent that their society possesses a dominant religious tradition supported by the elite (211).
P220: Persons who desire limited rewards that exist, but who lack the social power to obtain them, will tend to affiliate with cults, to the extent that their society does not possess a dominant religious tradition supported by the elite (212-13).
P221: People who repeatedly experience turning points in their lives are more likely than other people to become chronic religious seekers (229).
Thus, three variables may be drawn from these propositions. Proposition 204 indicate the importance of the individual's stake in conformity. Propositions 219 and 220 indicate that affiliation with a cult or sect will depend upon the strength of conventional religious organizations. Proposition 221 notes the importance of turning points in determining whether an individual will be drawn to cults or sects, as Stark and Bainbridge argue that "religious seekers" are more likely than others to find their way into such groups. Stark and Bainbridge's use of stake in conformity closely follows Hirschi's (1969) conceptualization of a social bond composed of "attachments, investments, involvements, and beliefs," which deters individuals from engaging in deviant behavior (Stark and Bainbridge 1987: 190).
The type of deviant religious movements with which people with low stakes in conformity will affiliate depends upon the nature of the religious organizations in their area. If their region has a dominant and conventional religious tradition, disenfranchised individuals will tend to join a sect, which maintains most of the tenets of that dominant religious tradition while changing certain tenets to reflect increased tension with the surrounding society. Should individuals reside in an area without a dominant, conventional religious tradition, they will tend to join cults - deviant religious groups that are innovated or imported.
In addition to the individual's stake in conformity and the strength of conventional religious organizations, the number of turning points will also effect the probability of religious deviance. A turning point, as defined by Stark and Bainbridge, is a "period of markedly decreased attachment, investment, involvement and belief, taken singly or in any combination" (1987: 229). Thus the loss of a job, the failure of a marriage, and the end of an important attachment, would all be "turning points" in an individual's life. Stark and Bainbridge propose that people who have experienced many such turning points will be less likely to become attached to any particular religious organization. In fact, such people may become "chronic religious seekers," moving from religious group to religious group and eventually landing in cults and sects:
In societies where many sects and cults are available, this intermittently explosive lifestyle may lead to a sequence of short term affiliations with religious movements, or at least oscillations in participation in a single group over time (Stark and Bainbridge 1987: 229).
Finally, a fourth variable of interest, education, appears in an earlier formulation of the Stark-Bainbridge theory (1985). Stark and Bainbridge argue that an individual's educational level influences the type of religious group he or she will be attracted to. Cults involve new ideas and therefore "necessitate the willingness and the ability to entertain the new and unfamiliar" (Stark and Bainbridge 1985: 405). Thus, Stark and Bainbridge propose that the more educated will be drawn to cults, while the less educated will be drawn to the relatively more conventional ideas espoused by sects and churches. Studies have consistently found a positive relationship between education and cult membership (Barker 1981; Nordquist 1978; Church of Scientology 1978; Stark 1972, 1984; Wuthnow 1978).
The propositions Stark and Bainbridge provide easily lend themselves to a model of religious affiliation with the following independent variables: stake in conformity, education, strength of conventional religious organizations, and several items measuring turning points, including the subjects" number of marriages, number of moves, number of jobs, and number of authority figures they lived with while growing up. The dependent variable in this case is the form of religious affiliation - church, sect, cult, or no religious preference. Each of the independent variables is expected to have a direct effect on the level of religious affiliation. The direct effects of the independent variables should differ depending on the type of religious organization, church, cult, or sect. Therefore, three different models have been created.
Figure 1 provides the model for church membership with the expected relationships among the variables. Since Stark and Bainbridge define churches as religious organizations that support the conventional order and norms of society (Stark and Bainbridge 1987: 124), a person's stake in conformity should be positively related to church membership. The direct effect for education should be negative, since Stark and Bainbridge argue that the more educated tend to prefer novel religious ideas. Naturally, the strength of conventional religious organizations should have a positive effect on individuals joining such an organization. Finally, the four "turning points" items should be negatively related to church membership.
Although Stark and Bainbridge do not hypothesize about the effects of the independent variables on one another, expected indirect effects have been built into the three models. In addition to its direct effect on the type of religious organization the subject belongs to, education is expected to have a positive effect on stake in conformity. As an individual's education increases, the stake in conformity will likely increase, due to the increased investments a higher education provides.
Furthermore, each of the turning points items is expected to have a negative effect on stake in conformity. Since Stark and Bainbridge argue that the turning points an individual has experienced could lead that person to become a chronic seeker, the authors hypothesized that such chronic seekership should also negatively effect the chances of successful employment, marriage, and investments, and, hence, the current stake in conformity. The number of marriages and number of moves are expected to be positively correlated, since a failed marriage often includes a change in households. The number of moves and number of jobs are also expected to be positively correlated, since a change of location may require a change in jobs and vice versa.
Figure 2 provides the model for sect membership with the expected relationships among the variables. Since Stark and Bainbridge define sects as deviant religious groups, the individual's stake in conformity should be negatively related to participation in such a group. The direct effect for education in the sect model should be negative, since Stark and Bainbridge argue that the more educated should be attracted to cults. The four "turning points" measures should have a positive effect on joining a sect. Finally, the strength of conventional religious organizations should have a positive effect on sect membership. Per proposition 219 of the Stark-Bainbridge theory, the disenfranchised should tend to affiliate with sects if their region has a strong, conventional religious tradition.
Figure 3 provides the model for cult membership, with the expected relationships among the variables. As discussed previously, Stark and Bainbridge argue that education should be positively related to cult membership. The relationship between stake in conformity and cult membership should be negative, as in the sect model - the higher an individual's stake in conformity, the less likely that individual should be to join a deviant religious group. The relationship between the number of turning points an individual has experienced and cult membership should be positive, as in the sect model. Contrary to the sect model, however, the strength of conventional religious organizations should be negatively related to cult membership, as Stark and Bainbridge argue that individuals will tend to join such groups when conventional organizations are weak.
The data used in this analysis are from the National Survey of Families and Households (NSFH), a national, multistage area probability sample of 17,000 housing units, which resulted in 13, 017 subjects. Eligible respondents were over age 18, with two exceptions: currently married persons under age 19 were considered eligible, as were respondents under age 19 who lived in a household without older members. The pool of subjects includes a double sampling of blacks, Puerto Ricans, Mexican Americans, single-parent families, families with stepchildren, cohabiting couples, and recently married couples.
Funded by the Center of Population Research of the National Institute of Child Health and Human Development, the NSFH interviews gathered information about the respondent's family living arrangements during childhood, marital and cohabitation experience, the experience of leaving the parental home, and education, fertility, and employment histories (Center for Demography and Ecology 1990: 7). For each household, one adult was randomly selected as the main respondent and given both face-to-face and self-administered interviews. The respondent's spouse or cohabiting partner was also given a self- administered questionnaire.
The NSFH study was chosen for the purposes of this paper because of the manner in which it asked respondents to indicate their religious preference. While many surveys, such as the General Social Survey, simply ask respondents to choose from a set of major religious categories (e.g., Protestant, Catholic, Jewish, or other), the NSFH asked respondents to provide the specific name of their religious organization, if any. The religion codes were categorized using a typology of religious organizations created by J. Gordon Melton (1977), resulting in 64 different religious groups. Testing the Stark and Bainbridge theory is impossible without such a specific categorization.
The manner in which items were coded allowed inclusion of 12,415 subjects in this study. There are seven independent variables: education, stake in conformity, strength of conventional religious organizations, number of marriages, number of moves, number of jobs, and number of authority figures. The dependent variable is the type of religious affiliation, that is, church, sect, or cult. A religious preference of "None" served as the contrast category. In addition to the direct effect for the independent variables, indirect effects of education and the four turning points items, through stake in conformity, were proposed. Education was operationalized as the number of years of formal education the respondent reported. The average respondent had not received further education after graduation from high school (mean = 12.4 years).
Stake in Conformity. Hirschi conceptualized four elements of the "social bond" that keep individuals tied to conventional society. An individual's current stake in conformity is the strength of these elements. Attachments are important relationships that keep individuals from deviating. In other words, an individual who has strong attachments will, theoretically, not want to risk the disapproval of these significant others by engaging in deviant behavior. Investments of money, energy, or time made by a person in order to obtain a certain life-style, compose another part of the stake in conformity. Someone who has worked hard and achieved material success will, theoretically, avoid deviant behavior, since it might threaten the investments they have made in a certain life-style. Involvement refers to the extent to which individuals engage in conventional activities. Hirschi argues that an individual who is frequently involved in activities such as church groups or social clubs will be "too busy doing conventional things to find time to engage in deviant behavior" (188). The final element of the social bond, beliefs, assumes that individuals will be less likely to engage in deviant behavior if they have strong, conventional beliefs. The stake in conformity construct used in this study contained twelve items thought to represent each of the four elements of the social bond.
The strength of the subjects' attachments was measured by three variables. First, it was determined whether the subject was currently married or had a live-in partner. The second variable was the number of children the subject had. Finally several items representing support received from family members, friends, neighbors, and co-workers were combined to create a social support measure. The NSFH asked respondents to indicate whether they had received help from friends, neighbors, co-workers, sons or daughters, parents, brothers or sisters and other relatives in several areas, including child care, transportation, home or car repairs, housework, and, finally, advice, encouragement, and/or moral support. Thus, a high score on the support measure indicates that the respondent has a strong network of attachments from which he or she can receive necessary support.
The strength of each subject's investments was measured by six variables - the subject's employment status, household income, and whether or not the subject owned a home, real estate, a business or farm, and/or a car, a boat, or recreational vehicle.
Involvement was measured by determining the number of activities in which the subject participates. Activities were divided into two separate measures: The frequency of social events with friends, neighbors, and family members, and the frequency of participation in different types of groups. Respondents were asked how often they had social contact with friends, neighbors, and family on a scale ranging from "Never" to "Several times a week." These items were added together to create an overall score for socializing. The NSFH also asked respondents how often they participated in fourteen different types of groups, including fraternal groups, service clubs, veterans' groups, political groups, labor unions, sports groups, youth groups, school-related groups, hobby/garden clubs, school fraternities/sororities, nationality groups, farm organizations, literary/art study groups, and professional or academic societies. Respondents indicated the frequency of each activity on a five-point scale ranging from "Never" to "Several times a week." The fourteen items were added together to create the overall group score. Thus, for each measure, higher scores indicate a greater degree of participation in social activities.
The belief dimension of the stake in conformity was measured by creating a "conservative beliefs" score. The conservatism variable consisted of the sum of responses to six questions regarding values and morals, measured on a Likert-type scale. The items were as follows: 1) Marriage is a lifetime relationship and should never be ended except under extreme circumstance; 2) Preschool children are likely to suffer if their mother is employed; 3) Parents ought to provide financial help to their adult children when the children are having financial difficulty; 4) It is all right for an unmarried couple to live together even if they have no interest in considering marriage; 5) Parents should encourage just as much independence in their daughters as in their sons; and 6) If a husband and wife both work full time, they should share household tasks equally. The six items were coded such that higher scores indicated higher levels of conservative family values. Therefore, items 4, 5, and 6 were reverse coded. The items were then added to create a conservative beliefs measure. The higher the score, the more conservative the respondent's beliefs.
The twelve variables discussed above were summed to create the stake in conformity scale. All of the variables included in the scale, were standardized after missing values were set to the mean. A high score on the scale indicates a high stake in conformity. The scale has a reliability coefficient of .53.
Strength of Conventional Religious Organizations. As noted previously, Stark and Bainbridge hypothesize that the strength of the conventional organizations in an individual's area of residence will, in part, determine what type of deviant religious groups he or she will be likely to join. In an earlier formulation of their theory of religion, Stark and Bainbridge (1985) used the Glenmary Research Center's census of denominations to compute church membership rates per 1,000 in the population for the different regions of the United States (76). In general, church rates decrease as one moves from east to west across the United States. The Northeast has the highest church membership rate, followed by the North Central region, the South, and finally, the West. The NSFH provides the region of the country for each respondent. Thus each respondent was given a strength of conventional religious organizations score of 1 to 4, depending on their region of residence. The higher the score, the greater the strength of conventional religious organizations in the respondent's area.
Turning Points. Stark and Bainbridge's idea of turning points proved difficult to operationalize. An ideal measure of turning points for any particular individual, per Stark and Bainbridge's definition, would include the number of times an individual has experienced a decreased attachment, investment, involvement, or belief. Two measures of decreased attachments, one measure of decreased investments, and one measure thought to tap decreased attachments and involvement's are used in this study to represent turning points. The authors realize that an ideal measure of turning points, per Stark and Bainbridge's definition, should include decreased beliefs as well.
One measure of decreased attachments is the respondent's number of marriages - a high number of marriages equaling a high number of failed attachments. The other measure combines several questions from the dataset to determine how many different sets of authority figures the respondent was made to live with during childhood. A series of questions asked respondents whether they had lived with stepparents, adoptive parents, and/or grandparents or other relatives for four months or more while growing up. Theoretically, each time the respondents were made to live with a new set of authority figures, they experienced a "turning point" per their attachments; that is, they lost an attachment to one set of authority figures and were forced to accept another. Approximately 79% of the respondents lived with only their parents while growing up. Approximately 21% of the sample lived with at least one other set of authority figures. The resulting variable was dichotomous, with a zero indicating that the respondent had only lived with his or her parents while growing up.
The number of jobs the respondents had had was used as a measure of decreased investments. The final variable used to create the turning points score, the number of moves the respondent has experienced, arguably taps the attachment, investment, and involvement dimensions of the stake in conformity. Depending on the distance involved in a move, it may result in lost attachments and involvements.
Type of Religious Affiliation. The dependent variable in this analysis, the type of religious affiliation, involved coding the 64 religious groups represented in the NSFH as either churches, sects, or cults, per the definitions in the Stark and Bainbridge theory. Religious groups were coded using three sources, Stark and Bainbridge (1985), Melton and Geisendorger (1977), and Melton (1989). Those groups explicitly mentioned as either churches, sects, or cults in Stark and Bainbridge's work were coded as such. Approximately 32% of the religious organizations could be coded in this manner.
For those groups not mentioned in the Stark and Bainbridge work, two works by religious scholar J. Gordon Melton were consulted. Melton's works provide detailed information about the history, leadership, and beliefs for religious organizations in the United States. Melton organizes the religious groups into two general groups - those within the Christian tradition and those outside the Christian tradition. Each group is further divided into families and subfamilies that have shared heritage, beliefs, and life-styles. Groups within the dominant, Christian tradition were coded as churches, unless there was evidence that the group had beliefs conflicting with societal norms, in which case they were coded as sects. For example, groups in the Holiness family were categorized as sects, as Melton's description indicates that members are supposed to separate themselves from worldly things in their search for perfection (1977: 35). Members of the European Free Church family were also coded as sects, as they practice a "social ethic tending toward isolation from social structures" (1979: 243).
Groups which Melton indicated as being outside the dominant religious traditions were coded as cults, per the Stark and Bainbridge definition. The Islamic, Hindu, Buddhist, and Shinto/Taoism subfamilies were coded as cults, as they represent an imported religious group with deviant religious beliefs. Approximately 48% of the religious groups were coded using Melton's works.
Approximately 19% (12) of the groups had to be eliminated from the analysis for several reasons. First, groups had to be eliminated if the respondent did not provide enough information to code their religion as church, sect, cult, or no religious preference. For example, some people labeled themselves with the generic term "Christian," and others simply indicated that they practiced their religion in private. Some categories were too broad to code as a church or sect. For example, "All other members of the Pietist family" could not be coded because the Pietist family includes both groups that might be considered churches and groups that might be considered sects.
Mormonism and its offshoots added a third area of difficulty. Mormonism could be defined as a sect of Christianity, since it utilizes aspects of the Christian tradition in its beliefs. However, Mormonism adds so much new material to the Christian tradition, in the form of an entire new "testament, " The Book of Mormon, that it should be considered a new tradition, and, hence, a cult. Furthermore, in certain areas of the country, such as Utah, Mormonism is the dominant religious tradition. The problems categorizing Mormonism led Stark and Bainbridge to eliminate Mormon groups from their 1985 study (245). For this study, three categories were removed from the analysis due to this classification difficulty - Mormonism, the Reorganized Church of Latter Day Saints, and "All other members of the Latter Day Saint family."
Removing respondents who fell into the above-mentioned religious groups or categories left a sample of 12, 415 subjects. One thousand thirty-one (1,031) of the respondents indicated that they had no religious preference. Of the remaining 52 religious groups, 17 were classified as churches, 26 as sects, and 8 as cults. Church members account for 10,366 (83.5%) of the subjects. Sect members account for another 883 subjects (7.1%). The remaining 135 subjects (1.1%) are classified as cult members. A full list of religious groups and their coding is available from the senior author upon request.
The model contains both a categorical outcome variable (religious affiliation) and an interval-level intervening variable (stake in conformity). We therefore have had to employ both OLS and polytomous logistic regression in order to estimate all direct and indirect paths in the model. OLS was used for the estimation of all direct paths from the exogenous variables to stake in conformity. Polytomous logistic regression (Aldrich and Nelson 1984; DeMaris 1992; Hosmer and Lemeshow 1989) was employed for the estimation of both direct and indirect paths from all independent variables to religious affiliation. Indirect paths were estimated via the hierarchical regression technique advocated for path models by Alwin and Hauser (1975).(1)
Although OLS regression is familiar to most data analysts, polytomous logistic regression is less well known. This technique is similar to OLS in that a continuous dependent variable is modeled as a linear function of a parameter set, except that the response is the natural logarithm, or "log," of the odds of membership in one category of a categorical variable, versus some "baseline category." The baseline chosen here is having a religious affiliation of "no preference." Three odds are of interest here: the odds of church affiliation versus no affiliation; the odds of sect affiliation versus no affiliation; and the odds of cult affiliation versus no affiliation. Each log odds is modeled as a linear function of the predictor set; hence, there are three equations to be estimated for the dependent variable, church affiliation. SPPS has no procedure for polytomous logistic regression at present. However, the analysis can be accomplished by using dichotomous logistic regression to estimate each log odds separately, as outlined by Begg and Gray (1984). The only drawback to this is that the estimates are obtained by maximizing three separate likelihood functions rather than just one. Although this involves some loss of efficiency in parameter estimation, the problem is not particularly severe when the sample size is large, as it is here (Begg and Gray 1984; Hosmer and Lemeshow 1989).
Logistic regression coefficients can be interpreted in terms of the log odds or the odds. The coefficients have the same interpretation for the log odds as coefficients in OLS. That is, [b.sub.k] represents the change in the log odds for a unit increase in the kth predictor, controlling for the other effects in the model. On the other hand, exp([b.sub.k]) represents the multiplicative change in the odds for a unit increase in the kth predictor, controlling for the other effects in the model. For example, the estimate of the effect of stake in conformity on the log odds of cult affiliation [ILLUSTRATION FOR FIGURE 6 OMITTED] is -0.68. In terms of odds, this means that each unit increase in stake in conformity multiplies the odds of cult affiliation by a factor of exp(-0.68) = .934. Hence, each unit increase in stake in conformity slightly reduces the odds of being in a cult, as opposed to having no religious affiliation. The other coefficients are similarly interpreted.
The polytomous logistic regression yielded interesting results for church affiliation. The predictor set was highly significant ([[Chi].sup.2] = 193.19, df = 7, p [less than] .0001). Figure 5 reproduces the results for church affiliation in path form, displaying only the significant paths. Curiously, the stake in conformity variable did not have a significant effect on the log odds of church membership. Thus a key component of the Stark-Bainbridge theory was not supported by our results. However, as predicted by the theory, the strength of conventional religious organizations in the subject's area has a significant (p [less than] .01) and positive effect on the log odds of church membership. A unit increase in the strength of conventional religious organizations results in a .2831 increase in the log odds of church membership. Education also had the expected effect on church affiliation (p [less than] .01). Each additional year of education decreases the odds of church Affiliation by 5.2%.(2)
The effects of the turning points variables were mixed. All four of the turning points variables had significant effects on church affiliation, but only two of these effects conformed to the model's predictions. The number of authority figures the subject experienced had the predicted negative effect on church affiliation (b = -.1434, p [less than] .01), as did the number of times the subject has moved (b = -.1493, p [less than] .01). The remaining two turning points variables had effects contrary to predictions. The number of marriages actually increased the log odds of church affiliation (b = .0841, p [less than] .05), as did the number of jobs the subject has held (b = . 1243, p [less than] .01).
Finally, education had the predicted positive effect on the intervening variable, stake in conformity (b = .6053, p [less than] .01). The subject's number of marriages continued to act contrary to prediction, with a significant (p [less than] .01) and positive effect on stake in conformity. The number of moves had the predicted negative effect on stake in conformity, but the number of jobs and number of authority figure did not have significant effects.
The predictor set for the model of cult affiliation was also quite significant ([[Chi].sup.2] = 36.431, df = 7, p [less than] .0001). Only two of the variables, however, had a significant effect on cult affiliation. Figure 6 reproduce the model for cult membership, displaying only the significant paths.
None of the turning points variables had a significant effect upon cult affiliation. Curiously, as with church affiliation, the subject's number of marriages appeared to increase their current stake in conformity (b = 1.006, p [less than] .05). The strength of conventional religious organizations also did not have a significant effect upon cult affiliation. However, both education and stake in conformity had significant effects (p [less than] .01) upon cult affiliation which conformed to the model's predictions. Each unit increase in stake in conformity decreased the log odds of cult affiliation by -.068. In other words, each unit increase in stake in conformity decreases the log odds of cult affiliation by 7.04%. Furthermore, each additional year of schooling increased the log odds of cult affiliation by nearly 20% (19.94).
1 Due to space considerations, the authors did not include breakdowns of all direct and indirect effects. A set of tables depicting all direct and indirect effects is available from the senior author upon request.
2 The percentage change in the odds for a unit increase in [X.sub.i] is computed via the formula 100[exp([b.sub.j]) - 1] (Demaris 1992: 47).
Aldrich, J. H. and F. D. Nelson. 1984. Linear probability, logit, and probit models. Thousand Oaks, CA: Sage.
Alwin, D. F. and R. M. Hauser. 1975. The decomposition of effects in path analysis. American Sociological Review 40: 37-47.
Appel, W. 1983. Cults in America. New York: Holt, Rinehart and Winston.
Bainbridge, W. S. 1978. Satan's power: A deviant psychotherapy cult. Los Angeles: University of California Press.
Balch, R. W. 1976. Salvation in a UFO. Psychology Today October: 58-66.
-----. 1980. Looking behind the scenes in a religious cult: Implications for the study of conversion. Sociological Analysis 41(2): 137-43.
Barker, E. 1981. Who'd be a Moonie? In The social impact of new religious movements, edited by B. Wilson, 59-96. New York: Rose of Sharon Press.
Beckford, J. A. 1985. Cult controversies: The societal response to new religious movements. New York: Tavistock Publications.
Begg, C. B. and R. Gray. 1984. Calculations of polytomous logistic regression parameters using individualized regressions. Biometrika 71: 11-18.
Bromley, D. G. and A. D. Shupe, Jr. 1979. Just a few years seem like a lifetime: A role theory approach to participation in religious movements, In Research in social movements, conflicts and change, vol. 2, edited by Louis Kriesberg, 159-86. Greenwich, CT: JAI Press.
Campbell, C. B. 1977. Clarifying the cult. British Journal of Sociology 28: 375-88.
Center for Demography and Ecology. 1990. A national survey of families and households: Codebook. Madison, WI: University of Wisconsin-Madison.
Church of Scientology. 1978. What is Scientology? Los Angeles: Church of Scientology.
Damrell, J. 1977. Seeking spiritual meaning: The world of Vedanta. Sociological Observations, no. 2. Beverly Hills: Sage.
DeMaris, A. 1992. Logit modeling: Practical applications. Newbury Park, CA: Sage Publications.
Edwards, C. 1981. The dynamics of mass conversion. Marriage and Family Review 4(3): 31-40.
Festinger, L., H. W. Riecken, and S. Schachter. 1956. When prophecy fails. New York: Harper Torchbooks.
Goldberg, C. 1983. Courage and fanaticism: The charismatic leader and modern religious cults. In Psychodynamic perspectives on religion, sect, and cult, edited by D. A. Halperin, 56-79. Boston: John Wright.
Hirschi, T. 1979. A control theory of delinquency. In Classics of criminology, edited by J. E. Jacoby, 185-92. Prospect Heights, IL: Waveland Press.
Hosmer, D. W. and S. Lemeshow. 1989. Applied logistic regression. New York: Wiley.
Jackson, J. A. and R. Jobling. 1968. Towards an analysis of contemporary cults. In A sociological yearbook of religion in Britain, edited by D. Martin, 94-105. London: SCM Press.
Keiser, T. W. and J. L. Keiser. 1987. The anatomy of illusion: Religious cults and destructive persuasion. Springfield, IL: Charles C. Thomas.
Kelley, D. M. 1983. Deprogramming and religious liberty. In The brainwashing/deprogramming controversy, edited by D. G. Bromley and J. T. Richardson. New York: Edward Mellen Press.
Kirkpatrick, L. A. 1988. The Conway-Siegelman data on religious cults: Kilbourne's analysis reassessed (again). Journal for the Scientific Study of Religion 27(1): 117-21.
Levine, E. M. 1980. Deprogramming without tears. Society March/April: 34-38.
Melton, J. G. 1989. The encyclopedia of American religions. 3rd edition. Detroit, MI: Gale Research.
Melton, J. G. and J. V. Geisendorfer. 1977. A directory of religious bodies in the United States. New York: Garland Publishing Company.
Nelson, G. K. 1969. The spiritualist movement and the need for a redefinition of cult. Journal for the Scientific Study of Religion 8: 152-60.
Nock, D. 1989. Differential ecological receptivity of conversionist and revolutionists sects: A reconsideration of Stark and Bainbridge. Sociological Analysis 50(3): 229-46.
Nordquist, T. 1978. Ananda cooperative village. Uppsala, Sweden: Borgstroms.
Pilarzyk, T. 1978. Conversion and alternation processes in the youth culture. Pacific Sociological Review 21: 379-405.
Richardson, J. T. 1979. From cult to sect: Creative eclecticism in new religious movements. Pacific Sociological Review 22: 136-66.
Richardson, J. T. and B. Kilbourne. 1983. Classical and contemporary applications of brainwashing models: A comparison and critique. In The brainwashing/deprogramming controversy, edited by D. G. Bromley and J. T. Richardson. New York: Edward Mellen Press.
Robbins, T. and D. Anthony. 1978. New religions, families, and "brainwashing." Society June: 77-83.
Robbins, T., D. Anthony, and J. McCarthy. 1983. Legitimating repression. In The brainwashing/deprogramming controversy, edited by D. G. Bromley and J. T. Richardson. New York: Edward Mellen Press.
Robbins, T. 1988. Cults, converts and charisma: The sociology of new religious movements. Current Sociology 36: 1-25.
Sargant, W. 1957. Battle for the mind: A physiology of conversion and brainwashing. New York: Doubleday.
Scott, G. G. 1980. Cult and countercult: A study of a spiritual growth group and a witchcraft order. In Contributions in Sociology, 38. Westport, CT: Greenwood Press.
Sherman, L. W., et al. 1992. Crime, punishment, and stake in conformity: Legal and informal control of domestic violence. American Sociological Review 57:680-90.
Simpson, J. H. 1990. The Stark-Bainbridge theory of religion. Journal for the Scientific Study of Religion 29: 387-94.
Stark, R. 1972. The economics of piety: Religious commitment and social class. In Issues in social inequality, edited by G. Thielbar and S. Feldman, 483-503. Boston: Little, Brown.
-----. 1984. The rise of a new faith. Review of Religious Research 26(1): 18-27.
Stark, R. and W. S. Bainbridge. 1985. The future of religion: Secularization, revival and cult formation. Berkeley: University of California Press.
-----. 1987. A theory of religion. New York: David Lang.
Stark, R., W. S. Bainbridge, and D. P. Doyle. 1979. Cults of America: A reconnaissance in space and time. Sociological Analysis 40: 347-459.
Stillson, J. J. 1974. Hare Krishna and the counterculture. New York: John Wiley and Sons.
Suchman, M. C. 1992. Analyzing the determinants of everyday conversion. Sociological Analysis 53(S): S15-S33.
Ungerleider, J. T. and D. K. Wellisch. 1979. Coercive persuasion (brainwashing), religious cults, and deprogramming. American Journal of Psychiatry 136: 279-82.
Wallis, R. 1975a. Scientology: Therapeutic cult to religious sect. Sociology 9: 89-100.
-----. 1975b. The Aetherius Society: A case study in the formation of a mystagogic congregation. In Sectarianism: Analyses of religious and non- religious sects, edited by R. Wallis, London: Peter Owen.
-----. 1986. Figuring out cult receptivity. Journal for the Scientific Study of Religion 25: 494-503.
Wright, S. A. 1991. Reconceptualizing cult coercion and withdrawal: A comparative analysis or divorce and apostasy. Social Forces 70: 125-45.
Wuthnow, R. 1978. Experimentation in American religion. Berkeley: University of California Press.
Chris Bader is a doctoral candidate at the University of Washington, Seattle, WA 98195.
Alfred DeMaris is a professor of sociology at Bowling Green State University, Bowling Green, OH 43402.