Social Influence and Religious Pluralism

by William Sims Bainbridge

Advances in Group Processes, 1995, Volume 12, pages 1-18.

ABSTRACT

Sociologists of religion currently debate whether the proportion of the population affiliated with religious groups is increased or decreased by the diversity of denominations available in the community. Both viewpoints can be rooted in a general theory of religion that sees faith as the result of individual needs and processes of social exchange. Probably, both denominational diversity and denominational monopoly have effects that can increase religious mobilization, dependent upon the context of other factors, and the balance of these opposing forces will shift from one data set to the next, and from one set of control variables to another. Rather than causing sociologists to despair at the possibility of conducting conclusive research, this complex situation should inspire us to undertake fresh research projects directly examining the social and cultural dynamics of religious groups.

Social Influence and Religious Pluralism

Contemporary theories and much recent research in the sociology of religion are strongly oriented toward group processes, social networks, and interpersonal exchange. However, the considerable accomplishments in this subdiscipline have as yet had little impact on sociology and social psychology generally. Furthermore, researchers on group processes who lack backgrounds in the sociology of religion seem unaware of the opportunities to test or develop general theories via empirical studies of churches, religious movements, and the eminently interactional processes of recruitment and defection. Thus an analysis of a major controversy in the sociology of religion may not be out of place in a volume on group processes.

A current debate centers on two opposing arguments about the relationship between religious pluralism and commitment (Warner 1993). In a series of publications, Roger Finke and Rodney Stark (1988,1989a, 1989b, 1992; Finke, 1989) have pictured religion as a market economy in which denominations compete with each other for members. Different individuals and groups in society have different needs, cultures, and non-religious affiliations, so therefore religious pluralism should increase commitment by offering each person the style of religion that suits him or her best. In their empirical work, Finke and Stark have tried to show that rates of church membership are higher where there are more denominations in the religious marketplace.

In contrast, other researchers have argued that religious pluralism has a negative effect on church membership (Breault 1989a, 1989b; Land, Deane, and Blau 1991; Blau, Land, and Redding 1992; Blau, Redding, and Land 1993; cf. Christiano 1987). Religious monopoly might be associated with higher rates of religious involvement, if individual affiliations are chiefly the result of social influence, and if social influence is most effective when it is monolithic.

Thus, the narrow debate over denominational diversity and religious mobilization bears directly upon two distinctive general models of group process. The diversity-mobilization argument conceptualizes group affiliation in terms of individual choices among competing suppliers, with individuals maximizing their satisfaction by selecting the suppliers that best meet their personal needs. The monopoly-mobilization argument sees affiliation in terms of the net power of social influences operating within a diffuse social network, wherein persons are more strongly impelled to join a group the greater the proportion of their consociates who are members.

In the vigorous debate over proper methods of empirical research that has raged in the journals over this issue, the theoretical underpinnings have tended to become obscured. It is widely assumed that the diversity-mobilization hypothesis is rooted in the Stark-Bainbridge theory of religion (Stark and Bainbridge 1985, 1987; cf. Finke and Stark 1988), which laid the groundwork for a market analysis of religion and sought to establish the sociology of religion on a more rigorous footing (Simpson 1990; Collins 1993). However, in truth the monopoly-mobilization hypothesis also can be derived from the Stark- Bainbridge theory, and a close examination of how both of the competing hypotheses blend into a unified theory of group processes will reveal much about the limitations and prospects of contemporary sociology.

A THEORY OF RELIGION

The Stark-Bainbridge theory seeks to derive a long series of formal propositions about human exchange and religion, from a very short list of "axioms," by means of logical deduction and a set of strict definitions. Admittedly, the arguments it provides are only the outlines of possible proofs rather than fully stated rigorous deductions, and given the scope of the theory considerable effort would be required to fill all the gaps. In general, the theory accepts the ideas of George C. Homans (1974, 1984) about what a theory should do and how it should be constructed. A formal theory is a system of propositions, that logically derives relatively specific statements from relatively general ones, thereby explaining empirical observations in terms of general laws.

The theoretical system consists of a logical structure connecting three kinds of statements: axioms, definitions, and propositions. There are seven axioms, all to be quoted subsequently, comparable to the traditional axioms of plane geometry. Definitions associate each term to be defined with a particular set of properties or conditions that must be satisfied for something to be declared to be an instance of that term. Propositions, equivalent to the theorems of plane geometry, are relatively general statements that can be derived, directly or indirectly, from axioms and definitions. For convenience, the appendix of A Theory of Religion lists the axioms, numbered Al through A7, the most important definitions, Def. 1 through Def. 104, and a swarm of propositions, PI through P344. Many of the propositions do not concern religion, per se, but are statements about human interaction that were needed as stepping stones to propositions that were narrowly focused on religion.

The theory begins with a statement so obvious that social scientists hardly ever notice its importance: "Al Human perception and action take place through time, from the past into the future." Behaviorism and its modern sociological equivalent, social learning theory, assert that human behavior is conditioned by past events. Rational choice theory argues that behavior is oriented toward expected future consequences of actions. There is no disagreement here, only a difference of emphasis.

Past experiences, especially the contingencies of prior reinforcement, give people conceptual frameworks for anticipating the value of future events and selecting among actions that will influence the future. "A2 Humans seek what they perceive to be rewards and avoid what they perceive to be costs." Although some sensations are directly rewarding, most human experience concerns things and events that are merely stages on the way to palpable rewards, what are sometimes called instrumental rewards. Thus, much of the time, humans seek things and situations that they believe will be rewarding, but they can never be sure.

Beyond the distinction between palpable (or primary) rewards and instrumental (or secondary... tertiary... etc.) rewards, many other categories and dimensions could be identified. "A3 Rewards vary in kind, value, and generality." Because rewards vary in kind, humans can profit from exchanges with each other, and there is no one best way of obtaining all rewards. Among the instrumental rewards are various pieces of information that guide us in obtaining other rewards. Rewards are general to the extent that they include other (less general) rewards.

Given the variety of rewards that may be sought, and our good fortune to be scions of a rather successful history of biological evolution, humans are mentally equipped to follow complex plans for the attainment of various goals. "A4 Human action is directed by a complex but finite information-processing system that functions to identify problems and attempt solutions to them." This system, which consists of the "hardware" of the brain as well as the "software" of culture, is commonly called the mind.

At this point in the development of the theory, it became possible to sketch the derivation of propositions. Especially important are propositions that introduce the concept of explanation: "Def. 10 Explanations are statements about how and why rewards may be obtained and costs are incurred." Thus, explanations are instructions, recipes, plans, indeed theories but ones with practical significance. Among the many strange qualities of sociology, perhaps the least defensible is the discipline's general failure to pay attention to skills and practical knowledge. The "sociology of knowledge" is really the sociology of political ideology, as if there were nothing worth knowing in this life but the "fact" that the bourgeoisie exploits the proletariat. At root, the human mind is neither political nor philosophical but practical. "P3 In solving problems, the human mind must seek explanations." "P4 Explanations are rewards of some level of generality."

Thus, the theory of religion is basically utilitarian, but it is also tragic. "A5 Some desired rewards are limited in supply, including some that simply do not exist." "A6 Most rewards sought by humans are destroyed when they are used." "A7 Individual and social attributes which determine power are unequally distributed among persons and groups in any society." These three axioms mean that some pressing human desires will always be frustrated, that satisfaction quickly turns to dissatisfaction, and that some people are even worse off than others in what is inescapably a disappointing world.

The ray of sunshine in this gloom is social exchange. "P6 In pursuit of desired rewards, humans will exchange rewards with other humans." Because rewards are consumed, people will seek the same one repeatedly, thus placing a high priority on gaining a good explanation on how to do so reliably and efficiently. The wide variation in the distribution of rewards means that some people are ready and willing suppliers of a particular reward, accepting a different reward in exchange for it through recurring interactions with the same other individuals. Thus arise relationships and social networks.

Explanations are among the rewards most commonly obtained through social exchange. Unfortunately, but consistent with the tragedy of the human condition, explanations on how to obtain scarce and nonexistent but intensely desired rewards are often very difficult to test empirically. Therefore, they spread relatively unchecked through networks of communication. "PI 4 In the absence of a desired reward, explanations often will be accepted which posit attainment of the reward in the distant future or in some other non-verifiable context." "Def. 18 Compensators are postulations of reward according to explanations that are not readily susceptible to unambiguous evaluation."

The greatest unmet needs, such as everlasting life with good health and unlimited love, bring people to religion. "P22 The most general compensators can be supported only by supernatural explanations." "Def. 20 Compensators which substitute for a cluster of many rewards and for rewards of great scope are called general compensators." "Def. 22 Religion refers to systems of general compensators based on supernatural assumptions."

THE MONOPOLY-MOBILIZATION THEORY

The idea that consensus strengthens religion arises in a discussion of the historical emergence and social role of religious specialists (Stark and Bainbridge 1987, pp. 89-96). Why should people give material and social rewards to priests who promise to mediate on their behalf with the gods? Why do they not simply talk to the gods directly? The answers are somewhat involved, but a few key points can be stated briefly.

People seldom have the opportunity to evaluate fully an explanation they receive through social exchange. Yes, if action based on it manifestly delivers the desired reward, it is a good explanation. But the person does not know whether it is the best one, because some other set of actions might deliver the same reward more cheaply. The collective experience of many people often provides a better judgment than the limited knowledge of just one person, especially with rewards that are seldom unambiguously obtained through simple actions. Thus, even in the most practical realms of human life, people often rely upon others to evaluate explanations for them. This is especially so when the explanations are very difficult to test through personal experience, as is the case with religious compensators.

In their interaction with other individuals, people learn which exchange partners to trust, and they will develop mental lists of the people to consult about particular problems. An individual's exchange partners will have similar lists, and by communicating about their problems people develop a rough consensus concerning whom to exchange with for a given reward. Thus, professional specialists will arise in the historical development of human culture, as soon as the particular field of endeavor had developed a plausible subculture. Because it was not constrained by the harsh empirical failures that could discredit professions rooted in physical technologies, religion was one of the very first professions to arise. In a sense it bootstrapped itself into existence, as priests conspired to convince the laity that their particular brand of religion was the only valuable one, often in a monopolizing alliance with the state.

A string of theoretical propositions outlines the argument: "PI 3 The more valued or general a reward, the more difficult will be evaluation of explanations about how to obtain it." "P62 No human being can personally evaluate all the explanations he uses, including verifiable ones." "P63 The value an individual places on an explanation is often set by the values placed on it by others and communicated to him through exchanges." "P64 In the absence of a more compelling standard, the value an individual places on a reward is set by the market value of that reward established through exchanges by other persons. ""P65 The value an individual places on a general compensator is set through exchanges with other persons." "P66 When there is disagreement over the value of an explanation, the individual will tend to set a value that is a direct averaging function of the values set by others and communicated to him through exchanges, weighted by the value placed on such exchanges with each partner."

This chain of very general observations suddenly focuses again on religion. "P67 The more cosmopolitan a society with respect to religious culture, the lower the market value of any given general compensator." "Def. 36 Cosmopolitan refers to the existence of plural cultures within a society." Competing denominations are plural cultures, and thus diversity is equivalent to religious cosmopolitanism. To the extent that denominations differ in doctrine or practice, a person socially connected to two or more will experience lowered faith, compared to someone living in a religiously monolithic society where there are no religious disagreements.

THE DIVERSITY-MOBILIZATION THEORY

The pluralism argument arises from a discussion of the origins of sects (Stark and Bainbridge 1987, pp. 141-149). The hallmark of religion is that it provides general compensators that substitute for the set of major rewards, such as everlasting life, of which all human beings are deprived. Thus, one would think that a single Church Universal could provide the same comfort to everyone.

But this ignores the fact that some highly desired rewards are actually possessed by some persons while denied to others. Health and wealth are good examples. The wealthy may tend to be healthy, compared with the very poor, but ill health is found in all social groups. In contrast, poverty and wealth tend to be concentrated in very different corners of a community's social network. This means that deep social cleavages may arise separating the rich from the poor, whereas the healthy and unhealthy are found in every group and most families. Thus it is not surprising that compensators associated with social stratification have often been the basis for schisms that turn a unified church into two daughter organizations, a church and a sect.

For example, the members of a church congregation are often people with relatively high social status in their secular communities, whereas the members of a sect congregation are of lower secular status. To some extent, the church sanctifies the existing social status of its members. The sect, in contrast, provides compensatory status. To the sect member, possession of wealth and high education are not badges of honor; rather, faith in God and membership in the sect are the most noble qualities, conferring subjective status. In the extreme, the sect will define as sinful the consumption habits of members of the church congregation, while the church defines as presumptuous the status claims of sect members.

In terms of the theory's definitions and propositions, the argument proceeds according to the following outline: "Def. 57 A sect movement is a deviant religious organization with traditional beliefs and practices. ""Def. 59 Deviance is departure from the norms of a culture in such a way as to incur the imposition of extraordinary costs from those who maintain the culture. ""Def. 60 A schism is the division of the social structure of an organization into two or more independent parts." "P140 Consumers of scarce rewards and consumers of the corresponding compensators tend to avoid relationships with each other." "P142 Tension with the surrounding sociocultural environment is equivalent to subcultural deviance, marked by difference, antagonism, and separation." "PI 43 To the extent that religious groups are involved with compensators for scarce rewards, they are in tension with the sociocultural environment. ""PI 85 The more socially closed a group, the more readily it can generate and sustain compensators, through exchanges among members."

The underlying logic of this argument is that monolithic religious denominations cannot simultaneously meet the contrasting needs of people who possess highly desired scarce rewards and people who lack them. Given a modicum of religious freedom, the result is schism and the eruption of high- tension sect movements. Because these tend to be socially encapsulated and begin life with relatively small memberships, a society will tend to have many of them. But once several sects exist, they will collectively serve a substantial fraction of the relatively deprived classes in the society. The denominations they broke out of will be even better positioned to serve the needs of relatively advantaged persons, because they will feel fewer demands to serve the disadvantaged. The result of sect formation, therefore, will be increased mobilization: "PI 53 In a culture with a dominant religious tradition, the emergence of sect movements increases the proportion of the population affiliated with religious organizations."

The conditional with which this proposition begins is important in two ways. First, if the society has a dominant religious tradition, shared by denominations and sects alike, then the doctrines and practices of low-tension and high-tension groups will differ only in degree, not kind. That is, both liberal Protestant denominations and radical Protestant sects agree on the divinity of Jesus, the efficacy of prayer, and many other important points. Competing Protestant groups do not contradict each other completely, but only in circumscribed areas such as the proper attributes of social status. A Protestant society, therefore, has many of the advantages of religious monopoly despite possessing religious diversity.

Second, if the society lacks a dominant religious tradition, then religious movements are much more free to undertake radical religious innovation and become cults. "Def. 58 A cult movement is a deviant religious organization with novel beliefs and practices." "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." "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." Religious innovation may lay the groundwork for entirely new religious traditions that eventually gain substantial power in society, as was the case with Christianity in the Roman Empire and Mormonism today.

Healing cults, especially, can serve needs for relatively specific compensators that cut across social classes. Thus, a society like contemporary America that possesses many sects and cults would appear to have a highly fruitful form of diversity that by serving many different patterns of need should achieve a high level of religious mobilization. However, the memberships of cults remain quite small, and cults seldom appear in the measures of diversity that have been employed by published empirical studies concerning the mobilization theories. And one would think that very active cultic subcultures would undercut the plausibility of any particular set of compensators, because they disagree with respect to the most general compensators as well as specific ones, thereby greatly reducing mobilization. So there may be good reason to limit the diversity-mobilization theory to sects and denominations within a dominant tradition, as the original Stark-Bainbridge theory does.

THE EQUILIBRIUM PROBLEM

The fact that the Stark-Bainbridge theory leads to apparently contradictory predictions is not the flaw it might seem to be. Theories in the natural sciences frequently describe situations of opposing forces in which the resultant can be calculated only upon very precise quantitative knowledge of all the vectors. A space probe, moving above the earth, may be in a stable orbit, or it may not. Gravity pulls it toward the center of the earth, and its velocity may aim it in quite another direction. Qualitatively, one could say it might crash, or it might escape. A single equation can combine the opposing factors of gravity and momentum, and if accurate numbers can be put in, it provides a definite prediction. Similarly, with religious mobilization, the relative balance of competing forces determines the results.

Under conditions of religious freedom, in which the state does not suppress religious groups of the kinds that are popular in a free market of denominations, a community is likely to have the number of denominations desired by its population-to have reached an equilibrium in its religious market. Factors may work against this principle, notably the excessive cost of supporting a large number of different churches in a community with a small absolute population or the time lag in adjusting to major social changes. But if each community has the number of denominations it deserves, perhaps ultimately reflecting its socioeconomic and ethnic diversity, then diversity may not correlate with mobilization across communities, even if the diversity- mobilization theory is correct. Only if the system is in disequilibrium will the effect appear in ecological data.

Not all forms of disequilibrium will reveal a particular real effect that is concealed by equilibrium. If the need for religious diversity drops, for instance because of ethnic assimilation, then the number of denominations will remain too high, perhaps for a long time. This factor alone would distort the diversity- mobilization correlations. And if both the diversity-mobilization and monopoly-mobilization theories are true, then the result will be a reduction of the proportion of the population involved in churches and apparent evidence only for the monopoly theory.

One would expect ethnic diversity to be a prime determinant of the number of religious denominations in a community, although the evidence in favor of this apparently self-evident idea is weak at best (Greeley 1972; Blau, Land, and Redding 1992). It is possible that ethnic divisions in a population are so nearly coterminous with other divisions that a distinctive ethnic factor cannot be discerned. For example, if ethnic groups are concentrated in different geographic parts of the community, their different needs may be met by several neighborhood churches of a single denomination, each with a distinctive style. This has certainly often been the case with Catholicism in the United States, where separate Irish and Italian congregations exist in different neighborhoods of the same city. If a society is stratified by ethnicity, then separating the population into different denominations by social class simultaneously separates them by ethnicity. Ethnic assimilation, under these conditions, might have no effect on the number of denominations in a community.

The diversity-mobilization theory states that religious monopoly fails to meet the religious needs of some groups in society, and that some degree of religious pluralism will achieve a higher rate of religious involvement. But what is the mathematical function here? Is it possible that two denominations is a sufficient number to meet the varying needs of socioeconomic groups in an ethnically homogeneous society? Or will three groups achieve a much higher degree of mobilization than two? Clearly, there is the logical possibility that a positive mathematical function exists between the number of denominations and mobilization, but it achieves its maximum very quickly. If so, the real amount of variation among communities may lie largely beyond the range where the diversity-mobilization effect manifests itself. Another way of stating this troublesome possibility is to say that current levels of diversity may be so much above true equilibrium levels in many communities that variations in diversity have no implications for mobilization.

Disequilibrium in the United States and Canada is greatly the result of geographical migration and migration rates have strong, robust negative correlations with rates of church membership (Bainbridge 1990). Areas into which large numbers of people are moving tend to have relatively low church-member rates. To some extent, these places may be poor in denominations, as well, but it is also possible that they have substantial numbers of denominations, many of which are poor in clergy and church buildings relative to the population of the community and are thus unprepared to absorb the rush of in-migrants. Few metropolitan areas experience actual population declines, so out-migration may actually be compatible with equilibrium. However, geographical movement of different ethnic groups and social classes within a city often strands older churches where an appropriate clientele is lacking, thus giving a false impression of denominational diversity.

These observations raise the issue of denomination policy. To some extent, denominational diversity is the result of decisions made by the leaders of major denominations, and it is difficult to predict what social factors related to key variables in the mobilization theories might influence those decisions. Despite all the publications in the past fifteen years employing estimates of church membership, one very seldom sees analyses based on a readily-accessible variable from the U.S. Census: the proportion of employed persons who are clergy. In my own experience, this variable does not seem to correlate with anything and I have never discovered anything worth publishing about it. One hesitates to state any conclusions in the absence of serious research studies, but one implication is that many areas of low church membership are not low in clergy and actually may be oversupplied in churches, whatever the number of denominations may be.

Some large liberal Protestant denominations are reluctant to ordain clergy and to send them into an area unless there are congregations waiting for them. This might leave opportunities for smaller and more evangelical denominations, who will rush into areas that are poorly served by mainline denominations. The result may be a high level of religious diversity, even as the failure of some major denominations to recruit actively leaves the church- member rate low.

The existence of one or two highly successful religious movements in a community, ones that accomplish very high levels of mobilization for idiosyncratic reasons having nothing to do with the hypotheses under consideration here, may produce a high church-member rate while driving weaker denominations out of the local market. Thus, diversity may increase mobilization of the general population, while mobilization by one or two particular denominations reduces diversity. The net effect could be a statistical correlation of zero, despite the fact that it results from two highly significant effects.

People may not be especially free to adopt the religious affiliations that would suit them best as individuals, because they are socially tied to other individuals with somewhat different needs. Therefore, to the extent that the social network of a community is highly interconnected or cohesive, the diversity-mobilization effect would be muted. One might also think that cohesion and network interconnectivity strengthen social influence. The relative balance between the diversity and monopoly effects will vary as a function of the relative social cohesion versus social disorganization of the community, and thus variations in the stability and extensiveness of social relationships across time and space may significantly distort empirical research on the sources of religious mobilization.

Perhaps the greatest current debate in the sociology of religion, the secularization controversy, can be conceptualized in terms of diversity and mobilization. Is religion getting stronger or weaker? If it is changing in strength, in response to other social and cultural changes, then one would think that the rate of change is uneven across communities. It is hard to predict how differential secularization might distort empirical tests of the mobilization theories. To some extent, however, standard secularization theories dovetail with monopoly-mobilization theory. One reason people may lose faith in their religion is through exposure to many very different creeds that appear to contradict each other, and "widespread loss of faith" is how many social scientists might define secularization.

The Stark-Bainbridge theory, in contrast, argues that secularization is not confined to contemporary society, but is a universal process found in all societies and eras. The central, low-tension religious organizations of a society are forever being co-opted by secular powers and are constantly suffering erosion of faith. But this weakening of low-tension denominations is offset by the emergence of sect movements that revive their shared religious tradition. With the passage of time, many successful sects moderate and become prey to secularization themselves, but this merely sets the stage for yet more sects to emerge in further schisms. Over very long periods of time, entire traditions can become secularized and the result is stimulation of religious innovation. Out of the myriad of cults that arise in the ruins of an old religion, a very few will thrive, and one or two will establish entirely new traditions, thus starting the cycle all over again.

Where the United States stands in this long-term historical evolution cannot be determined with confidence (Bainbridge 1989, 1993). In their separate work, Stark has tended to stress revival of existing traditions, while Bainbridge has stressed innovation. But in their collaborative work they leave entirely open the question of whether the end of the twentieth century is a period of equilibrium sustained by revival or the beginning of a period of disequilibrium in which old traditions are weakening yet innovation has not yet established a successor tradition. Whether equilibrium or disequilibrium would be better for the researcher is a moot point, not only because we cannot confidently decide which kind of period our society has entered, but also because each brings its distinctive theoretical and methodological problems.

METHODOLOGICAL CHALLENGES

Sociological research is a difficult undertaking, our successes are never complete, and we often must make do with relatively weak standards of inference in using data to test theories (Collins 1989; Blalock 1989; Lieberson 1992). Although we dare not say so too loudly, this means that good empirical studies are almost never conclusive, unless multiple studies performed with diverse methodologies by different teams agree with each other to a high level of consistency. A single good empirical study is at best a plausibility argument, illustrating a theory and suggesting that it is worth taking seriously. Unfortunately, one senses that a necessary (but by no means sufficient) quality for publication in a scientific journal is the expression of great confidence in one's results. Ambiguous studies whose researchers are very dubious about their own results are almost never published.

Some years ago, Laurie Russell Hatch and I did a study testing the hypothesis that the strength of religion in a community, operationalized as the church- member or adherent rates, would correlate negatively with the proportion female in the elite professions of banker, lawyer, physician, and industrial manager (Bainbridge and Hatch 1982). Our three datasets employed church-membership rates based on different data sources: the 1926 census of U.S. religious bodies (Bureau of the Census 1930), a nongovernmental survey of American denominations carried out in 1971 (Johnson, Picard, and Quinn 1974), and the 1971 census of Canada (Statistics Canada 1974). To our consternation, the results were ambiguous. Somewhat lamely, we suggested that the much stronger support for the hypothesis in the Canadian data reflected the more conservative role of religion in that country. Coefficients wavered across professions and years in a most perplexing fashion, both with and without plausible statistical controls. At a scholarly conference, just after the article was published, a colleague expressed both surprise and pleasure that a journal editor would publish such unclear results.

To some extent, of course, each occupation in the dataset had its own unique characteristics, possessing its own distinctive career paths, barriers to female employment, and routes around those barriers. All these may vary geographically across the communities and some of those variations may correlate with other key variables or with their errors. In addition, all these effects may be powerfully shaped by the histories, recruitment pools, and social networks of particular denominations (cf. Pescosolido and Georgianna 1989). The analogy for mobilization research is that denominations may need to be analyzed separately, because each has a unique social base, history, recruitment strategy, and sensitivity to diversity or monopoly (Blau, Redding, and Land 1993).

Questions are frequently raised about the validity of the surveys of denominational membership done in 1971, 1980, and 1990, on which some of the mobilization studies rest. Hadaway, Marler, and Chaves (1993) have raised the possibility that actual church attendance has dropped in recent decades, even as self reports of attendance have remained roughly constant. Their own evidence concerns a marked discrepancy between claims of church attendance by a random sample of Protestants in one Ohio county with counts of actual attendance at the same churches, all data coming from 1992, so the researchers really do not have evidence of a decline in church membership. But their suggestion is worth taking seriously, and if the gap between the survey data and reality has changed markedly over time, one would think the errors vary also across geographic distance and across various community-level socioeconomic and cultural variables as well.

All those who have worked with the early twentieth-century censuses of religious bodies have been impressed by the apparent quality of the data. Unlike the recent studies that asked central denominational offices to provide county-level membership estimates, the old religious censuses sent a questionnaire to each local church in the nation. Few modern data collection efforts in the sociology of religion have been done with such great care and none have been on so large a scale. However, errors undoubtedly exist and it is very difficult to estimate them.

The church membership for a given city or county frequently includes some unknown number of persons who live outside the geographical unit, perhaps just across the county line, and thus will be counted in the numerator but not the denominator when calculating the church-member rate. The census reports were quite explicit that denominations differed greatly in whether they counted children as members or restricted membership to confirmed adults, and corrections for this problem are undoubtedly less than perfect (Bainbridge and Stark 1981). The rate of geographical migration, in and out of the community, presumably correlates with the error in the membership estimates. All of these errors may vary systematically across the units of analysis. For example, the tendency of a single church or denomination to miss-estimate its membership will cause less relative error in communities with many churches and denominations.

Courageously, Finke and Stark have taken the data series far back to the time of the American Revolution and it is greatly to their credit that they have done so. However, in some of the research quantitative sociology has climbed far out on a limb that may not be strong enough to sustain some of the statistical analyses that are now being done there. The full trove of data employed by Finke and Stark is sufficiently varied, that I tend to accept their model of steadily increasing church membership over the full sweep of American history, and a number of their analyses of particular denominations seem rather convincing to me, as well. But the more estimating and interpolating there is, and the greater the use of control variables, the more uncertainty I feel in the results.

Beginning in 1850, the U.S. censuses began collecting information directly from each local church on the number of seats for parishioners in their houses of worship and some sociologists have mistaken these for membership counts (Desroche 1971; Whitworth 1975). In 1890, data were collected on both parishioners and seats-counting both heads and tails, as it were-and this permitted Finke and Stark (1986) to develop a very clever statistical technique for estimating membership on the basis of seats. Blau, Land, and Redding (1992) adapted this method, in a sophisticated statistical analysis of the expansion of religious affiliation from 1850 to 1930 that made extensive use of interpolation. My own rather thorough familiarity with the data leaves me quite undecided about the appropriateness of concatenating so many assumptions and estimation techniques, and perhaps merely as a matter of personal timidity I have not dared to join the empirical battles described here.

Many sociologists decry all such research on the grounds that quantitative studies of geographically-based rates suffer from the ecological fallacy (Robinson 1950; Hannan 1971). That is, whereas the empirical data have all been at the most macro level, the theory really concerns group processes that mediate between the individual and the community. To a great extent, this so-called fallacy is just a variant of spuriousness and can be handled by a combination of appropriate control variables and caution when interpreting robust correlations (Bainbridge 1992). Correlations in the religious mobilization literature, however, appear far from robust and application of different control variables and statistical procedures appears to give wildly different results.

Few sociological theories are stated in quantitative terms and those that are expressed through well-defined functions tend to be very simple. Thus it is not surprising that the diversity and monopoly theories are not stated with sufficient precision that they can readily be combined to make definitive predictions. Computer simulation is a tool that may help improve the rigor with which we are able to derive and connect theoretical propositions (Bainbridge 1987, 1995; cf. Bainbridge and Stark 1984), but this works best for very narrowly framed theories and for the foreseeable future we cannot expect great rigor of theories that have considerable scope. The alternative is to extract the parameters of a general theory from careful measurement of variables in empirical studies, but the host of problems with existing church membership data leaves one with little confidence that this can be accomplished without substantial new data collection efforts.

CONCLUSION

The pluralism-mobilization theory is not new within American sociology. In 1879 and 1880, my great-grandfather, William Folwell Bainbridge, completed a tour of Protestant missions throughout Asia and the Middle East, to develop what he called a "science of missions," essentially a sociology of the factors that determined relative success of attempts at religious conversion. He noted a movement among some missionaries to divide up the world into spheres of influence by single denominations. "But I have observed that, as a rule, those mission stations of whatever church or denomination, which are left entirely by themselves, both for the present and the prospective future, do not show that activity and develop that strength, which are manifested in those mission fields where the presence or imminence of emulation has been felt. It was evident in Yokohama that Presbyterians and Methodists were prompting each other to a larger measure of evangelizing enterprise than either would have commanded with all the responsibility in the hands of a single mission, even though reinforced to the full extent of the other denomination's resources of men and means" (Bainbridge 1882, p. 270). He then devoted several pages to supportive evidence and a consideration of various social processes that might give rise to this diversity-mobilization effect.

Unfortunately, the sociology of religion, like sociology more generally, has tended to ignore nineteenth-century empirical American social science and to sink its roots instead deep into European social thought. One result has been the tendency to invest very little systematic research effort into close examination of the social dynamics that take place within and around religious groups, at the level of single congregations. The existence of aggregate-level data on church membership, available essentially free to any social scientist willing to expend energy coding and computerizing them, has quite appropriately attracted many sociologists of religion. In some cases, results have been robust and easy to interpret. But when they are far from robust and alternate interpretations abound, as in the mobilization literature, one can only conclude that very different kinds of data will be needed.

In terms of group processes, the diversity-monopoly argument has classic dimensions. On the one hand, a free market offers individuals a range of choices, thus maximizing potential satisfaction of demand and inducing the greatest number of people to be involved in the market. On the other hand, to the extent that religious faith is sustained by unanimity of opinion among a person's exchange partners, religious pluralism erodes faith, thus reducing the perceived value of religious goods and minimizing involvement in the market. These rational-choice and social-influence elements are merged in the original Stark-Bainbridge theory, which derived religion from a set of propositions about human needs, the group processes that produce particular faiths, and the formal organizations that stabilize and sustain them.

Let me state an outrageous but appealing scientific law: Every theory that informed and intelligent social scientists believe to be true, is true. Thus there are only two questions to be answered through empirical research. First, what is the domain of the theory; that is, when and to what phenomena does it apply? As Walker and Cohen (1985) note, a theory ideally should contain clear statements of the scope or boundary conditions of its propositions, but in fact theories seldom do. Second, what precise mathematical function describes the strength of the effect predicted by the theory? When two theories that competent sociologists believe to be true contradict each other within some portion of their domains, real effects will cancel each other out to some extent, giving an ambiguous or false picture of the truth value of one or both theories.

Thus, we can suggest two valid responses to the present debacle in macro-level research on religious mobilization. First, the problem cries out for collection of new data in studies designed to test the rival hypotheses, and the best approaches would examine the effects simultaneously at the level of individuals, small groups, and the religious economies of communities. Such studies will be costly, but the scientific gains from investment in them will extend far beyond the narrow limits of religious research, because the competing theories are so solidly based upon general models of group process. Second, researchers should recognize that contextual effects can powerfully shape the macro consequences of micro phenomena. If macro-level variables in fact shift the balance between rival diversity and monopoly effects, then the meaning of small group processes depends powerfully upon factors at a higher level of aggregation. Taken together, these two points argue for the importance of real-world research on small groups and for the realization that group processes can be powerfully affected by the larger social conditions surrounding them.

ACKNOWLEDGMENT

The views expressed in this article do not necessarily represent the views of the National Science Foundation or the United States.

REFERENCES

Bainbridge, W. F. 1882. Around the World Tour of Christian Missions: A Universal Survey. New York: C. R. Blackall.

Bainbridge, W. S. 1987. Sociology Laboratory. Belmont, CA: Wadsworth.

----- . 1989. "Religious Ecology of Deviance." American Sociological Review 54: 288-295.

----- . 1990. "Explaining the Church Member Rate." Social Forces 68: 1287-1296.

----- . 1992. Social Research Methods and Statistics. Belmont, CA: Wadsworth.

----- . 1993. "New Religions, Science, and Secularization." Pp. 277-292 in The Handbook on Cults and Sects in America, edited by D. G. Bromley and J. K. Hadden. Greenwich, CT: JAI.

----- . 1995. "Neural Network Models of Religious Belief." Sociological Perspectives

----- . in press. Bainbridge, W. S. and L. R. Hatch. 1982. "Women's Access to Elite Careers: In Search of a Religion Effect." Journal for the Scientific Study of Religion 21: 242-255.

Bainbridge, W. S. and R. Stark. 1981. "Suicide, Homicide and Religion: Durkheim Reassessed." Annual Review of the Social Sciences of Religion 5: 33-56.

----- . 1984. "Formal Explanation of Religion: A Progress Report." Sociological Analysis 45: 145-158.

Blalock, H. M. 1989. "The Real and Unrealized Contributions of Quantitative Sociology." American Sociological Review 54: 447-460.

Blau, J. R., K. C. Land, and K. Redding. 1992. "The Expansion of Religious Affiliation: An Explanation of the Growth of Church Participation in the United States, 1850-1930." Social Science Research 21: 329-352.

Blau, J. R., K. Redding, and K. C. Land. 1993. "Ethnocultural Cleavages and the Growth of Church Membership in the United States, 1860-1930." Sociological Forum 8: 609-637.

Breault, K. D. 1989a. "New Evidence on Religious Pluralism, Urbanism, and Religious Participation." American Sociological Review 54: 1048-1053.

----- . 1989b. "A Reexamination of the Relationship between Religious Diversity and Religious Adherents." American Sociological Review 54: 1056-1059.

Bureau of the Census. 1930. Religious Bodies: 1926. Washington, DC: U.S. Government Printing Office.

Christiano, K. J. 1987. Religious Diversity and Social Change. Cambridge: Cambridge University Press.

Collins, R. 1989. "Sociology: Prescience or Antiscience?" American Sociological Review 54: 124-139.

----- . 1993. "A Theory of Religion." Journal for the Scientific Study of Religion 32: 402-406.

Desroche, H. 1971. The American Shakers. Amherst: University of Massachusetts Press.

Finke, R. 1989. "Demographics of Religious Participation: An Ecological Approach, 1850-1980." Journal for the Scientific Study of Religion 28: 45-58.

Finke, R. and R. Stark. 1986. "Turning Pews Into People: Estimating 19th Century Church Membership." Journal for the Scientific Study of Religion 25: 180-192.

----- . 1988. "Religious Economies and Sacred Canopies: Religious Mobilization in American Cities, 1906." American Sociological Review 53: 41-49.

----- . 1989a. "How the Upstart Sects Won America: 1776-1850." Journal for the Scientific Study of Religion 28: 27-44.

----- . 1989b. "Evaluating the Evidence: Religious Economies and Sacred Canopies." American Sociological Review 54: 1054-1056.

----- . 1992. The Churching of America: 1776-1990, New Brunswick, NY: Rutgers University Press.

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

Hadaway, C. K., P. L. Marler, and M. Chaves. 1993. "What the Polls Don't Show: A Closer Look at U.S. Church Attendance." American Sociological Review 58: 741-752.

Hannan, M. T. 1971. Aggregation and Disaggregation in Sociology. Lexington, MA: Lexington.

Homans, G. C. 1974. Social Behavior: Its Elementary Forms. New York: Harcourt Brace Jovanovich.

----- . 1984. Coming to My Senses. New Brunswick, NJ: Transaction.

Johnson, D. W., P. R. Picard, and B. Quinn. 1974. Churches and Church Membership in the United States. Washington, DC: Glenmary Research Center.

Land, K. C., G. Deane, and J. R. Blau. 1991. "Religious Pluralism and Church Membership: A Spatial Diffusion Model." American Sociological Review 56: 237-249.

Lieberson, S. 1992. "Einstein, Renoir, and Greeley: Some Thoughts about Evidence in Sociology." American Sociological Review 57: 1-15.

Pescosolido, B.A. and S. Georgianna. 1989. "Durkheim, Suicide, and Religion: Toward a Network Theory of Suicide." American Sociological Review 54: 43-48.

Robinson, W. S. 1950. "Ecological Correlation and the Behavior of Individuals." American Sociological Review 15: 351-357.

Simpson, J. H. 1990. "The Stark-Bainbridge Theory of Religion." Journal for Scientific Study of Religion 29: 367-371.

Stark, R. and W. S. Bainbridge. 1985. The Future of Religion. Berkeley: University of California Press.

----- . 1987. A Theory of Religion New York: Toronto/Lang.

Statistics Canada. 1974. 1971 Census of Canada. Ottawa: Queen's Printer.

Walker, H. A. and B. P. Cohen. 1985. "Scope Statements: Imperatives for Evaluating Theory." American Sociological Review 50: 288-301.

Warner, R. S. 1993. "Work in Progress toward a New Paradigm for the Sociological Study of Religion in the United States." American Journal of Sociology 98: 1044-1093.

Whitworth, J. M. 1975. God's Blueprints. London: Routledge and Kegan Paul.