Religious pluralism and US church membership: a reassessment

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

Just as pluralism may diminish the proportion of one's social ties that reinforce one's religious identity, the next four independent variables shown in Table 1 are consistent with the claim that a reduction in the number of close ties can reduce religious involvement by reducing the number of persons well placed to observe and reinforce religious involvement. Results in Table 1 confirm that this may happen through geographic mobility or lack of the close ties that accompany marriage. Olson and Hadaway (1998) obtain very similar, and even more striking, results using the more than five thousand cities and towns in the 1991 Canadian Census (which includes a question on religious affiliation).

TABLE 1

OLS Standardized Regression Betas Predicting County Adherence Rates
1990

Independent Variables                    Standardized Betas

Religious Pluralism 1990                      -.30(***)

Variables Affecting

Number of Close Social Ties

Log Pop. Growth, prior 10 years               -.16(***)
Percent Born in Same State                     .14(***)
Percent of Adults Divorced                    -.28(***)
Percent of Adults Never Married               -.08(***)
Other Control Variables

Percent Urban                                  .21(***)
Percent Male                                  -.07(***)
Median Age                                     .14(***)
Median Family Income                           .10(***)
Percent with 4 yr. College Degree             -.01
Percent Hispanic                               .11(***)
Percent Black                                  .21(***)

Region

New England Region                            -.05(***)
Mid Atlantic Region                           -.07(***)
East North Central Region                     -.01
West North Central Region                      .21(***)

South Atlantic Region

East South Central Region                      .12(***)
West South Central Region                      .18(***)
Mountain Region                                .01
Pacific Region                                -.09(***)
R-squared                                      .561

N = 3,100 counties

* = p [less than] .05, ** = p [less than] .01,
*** = p [less than] .001

The large positive beta for percent of the county that is defined as urban by the census bureau supports Finke and Stark's (1988) and Finke et al.'s (1996) claim that urban areas are more religious. The beta for percent male is also consistent with previous results (e.g., Blau et al. 1992; Finke et al. 1996). The very strong positive beta for percent black is at least partially, if not mostly, explained by the fact that the Glenmary researchers (see Bradley et al. 1992:451ff.) estimated the numbers of Black Baptists in a county using the percentage of the county population that is African American. The dummy variables for US census regions show that regions vary considerably in their church adherence rates even after controlling for pluralism and other factors (cf., Stark and Bainbridge 1985; Kosmin 1993). The dummy variable for the South Atlantic region is excluded as the reference category since its mean adherence rate is close to the national average. For reasons explained below, Table 1 includes no controls for percent Catholic even though adding such controls has little effect when all counties are considered.