Religious pluralism and US church membership: a reassessment

Sociology of Religion, Summer, 1999 by Daniel V.A. Olson

A key debate in the sociology of religion concerns whether religious pluralism enhances or diminishes religious involvement. Some such as Peter Berger (1967) argue that the increasing pluralism of religion in modern societies is contributing to secularization. Religious pluralism undermines the social networks that reinforce the plausibility of belief. When the state, public institutions, and one's day-to-day social contacts no longer reinforce the truth of one's religious belief and, further, when these same social contacts expose one to a pluralism of beliefs, religion loses its quality as taken-for-granted truth, acceptance of beliefs and religious involvement decline.

Finke and Stark (1988) argue that pluralism has just the opposite effect. They suggest that religious monopolies become "lazy" but religious pluralism fosters competition which makes each religious group work harder to meet the religious needs of the populace, thus involving more people in religion. Moreover, increased religious choices raise the chances that any individual will become involved in some religious group.

SUMMARY OF ARGUMENT

Contrary to Finke and Stark, I find that church membership rates are lower in contemporary US counties having greater religious pluralism and that this negative correlation is unlikely to be spurious. In multiple regression no other variable has so strong an impact on church membership rates. While others report similar results using both contemporary and historical data from the US and Great Britain (Breault 1989a, 1989b; Land, Deane, and Blau 1991; Blau, Land, and Redding 1992; Blau, Redding, and Land 1993; Bruce 1992), Finke and Stark (1989) and Finke, Guest, and Stark (1996) raise important methodological objections to these analyses.

I respond to these objections and explain why Finke and Stark's interpretation of several key results showing a positive relationship is incorrect. Breault (1989a, 1989b) has already shown that the positive regression betas in several of Finke and Stark's analyses (Finke and Stark 1988, 1989; Finke 1992) are due to multicollinearity problems created by including statistical controls for the percent of the population that is Catholic and/or Mormon. However, Finke and Stark argue that these statistical controls are necessary to account for substantive characteristics of Catholicism and Mormonism and that without these controls, one cannot observe the actual, positive impact of pluralism on religious involvement.

I show that the multicollinearity problems identified by Breault do not result from substantive characteristics of Catholicism requiring statistical controls. Instead, they are a mathematical artifact of the way pluralism is calculated and the cases Finke and Stark select for analysis. I show that the same multicollinearity problems arise when controlling for the percent of the population in any denomination (even an artificially constructed "denomination" composed of Jews, Mormons, and all moderate Protestants) when samples are limited to cases where this denomination predominates. As used by Finke and Stark, the controls for percent Catholic distort the apparent effects of pluralism without controlling for the substantive factors Finke and Stark hope to account for.

Using more appropriate methods to control for Catholic presence, I show that large Catholic presence in present day counties has little effect on the church adherence rates of non-Catholics and that large Mormon presence depresses church adherence rates among non-Mormons. More importantly, religious pluralism has a negative effect on church membership even when these, more appropriate, controls are included in regressions.

Taken together with analyses of Canadian cities and counties in the 1991 Canadian census (Olson and Hadaway 1998), these analyses strongly suggest that religious pluralism is associated with lower religious involvement in contemporary North American settings and in most, but not all, historical US contexts.

RELIGIOUS ECONOMIES

The claim that religious pluralism increases religious involvement is central to much of Finke and Stark's recent work (e.g., Finke and Stark 1988, 1989, 1992; Finke 1990, 1992, 1997; Finke et al. 1996; Stark, Finke, and Iannaccone 1995; Stark 1997). While the results presented here run contrary to this claim, they do not contradict many of Finke and Stark's other, closely related, claims concerning the effects of religious environments or "economies" on religious involvement. I distinguish these claims dealing with the religious environment, or "religious economy," from other claims made by Finke, Stark, and others (e.g., Iannaccone 1994; Stark and Bainbridge 1985), claims that focus on characteristics of religious organizations themselves and/or their beliefs (e.g., their strictness or otherworldliness) rather than on the religious environment. These other claims are not the subject of this paper.

One can identify at least four important variables in the religious environment that Finke and Stark view as affecting religious involvement either directly or indirectly: regulation, market share, pluralism, and competition. While this study focuses on pluralism, the four variables are closely interrelated and need to be separately identified since some of the variables may have the effects that Finke and Stark describe even if pluralism does not.