Sage Journals: Discover world-class research

Abstract

French

While attending college, religious participation tends to decline among American students, but evidence of changes in religious belief is less clear. On the bases of both secularization theory and the moral communities thesis, we used multi-level modeling techniques to test whether institutional diversity predicts changes in student belief at Christian institutions. Results suggest that declines in absolutism were associated with increasing religious and political diversity, and religious diversity amplified the effects of academic tenure. Political diversity, however, explained the effects of religious diversity in combined models, suggesting that challenging political discourse may be more important for changes in religious belief than diversity of religious worldviews in the context of Christian higher education in the United States.

Keywords

diversity higher education moral communities religion secularization

Introduction

The challenge of modernity to religious faith, according to Berger (1967; 2014), results from exposure to an increasingly broad range of moral worldviews and philosophies. In his early work, he described secularization as an inevitable byproduct of pluralistic competition; the co-presence of multiple sacred orders of meaning – or ‘plausibility structures’ – undermines the legitimacy of them all (Berger, 1967). More recently, however, he has argued that the main implication of the emerging social reality is a change in ‘the “how” rather than the “what” of an individual’s faith’ (Berger, 2014: 32). Modernity, he claims, does not secularize per se. Rather, building upon the view that secularization is better characterized as the increasing recognition that belief structures, once taken-for-granted in relatively homogenous societies, are choices among many options (Taylor, 2007), he argues that modernity relativizes religious faith. Faith necessarily coexists with some degree of doubt, absolutism is decreasingly tenable, and the management of doubt is the preeminent challenge of religious individuals.

Berger’s refined theory of modernity frames the current study on student religious belief at Christian colleges and universities. The college campus constitutes a particularly interesting context in which individuals are often exposed to various types of diversity. Emerging adulthood, generally, is a unique time of socialization and identity formation (Arnett, 2000; Longest and Smith, 2011), and religious individuals attending college tend to experience this transitional phase without the watchful eyes of their parents and religious youth leaders for the first time. But while studies consistently find that religious behavior tends to decline during college (Hill, 2009; Hunsberger, 1978; Uecker et al., 2007), evidence regarding changes in religious belief is mixed, and the processes related to these changes are complex (Hill, 2011; Mayrl and Oeur, 2009; Reimer, 2010). Recent studies have investigated the role of institutional context in conditioning patterns of change in religious belief in practice. Framing colleges and universities as ‘moral communities’ (see Stark, 1996), some have found that religious homogeneity at conservative Protestant schools serves as a protective factor against declines in religiosity (Davignon and Thomson, 2015; Hill, 2009).

Here, we test Berger’s (2014) reformulated theory that the pluralism of modernity relativizes, rather than secularizes, religious faith by considering the effects of both religious and political diversity on absolutist beliefs related to salvation and the Bible among students at conservative Christian institutions of higher learning in the United States.

Religiosity at college

Student religiosity at institutions of higher education has implications for student experience and formation. Religious participation appears to enhance satisfaction with the student experience, and it reduces participation in risky behaviors such as substance use and its associated negative physical, social, and academic consequences (Bryant and Astin, 2008; Mayrl and Oeur, 2009; Nelms et al., 2007; Perkins, 1987). Questions remain, however, about the effects of college education on individual religiosity (for a summary see Mayrl and Oeur, 2009). Most American college students are religiously affiliated, but the strength of their involvement is often uneven; students tend to report praying or attending services ‘occasionally’ more often than ‘frequently’, for example (Mayrl and Oeur, 2009: 263).

Moreover, religious participation tends to decline in college, though Uecker and colleagues (2007: 1683) attribute such declines to factors that affect all emerging adults, including the ‘late-night orientation of young adult life’ and norms against appearing ‘too religious’, as well as the fact that many churches do not emphasize college-aged ministries. Whether religious belief also declines is, however, more contentious. College students are likely to be exposed to secular perspectives, including explanations of the world that may compete with religious beliefs, but the authors did not find any changes in religious belief among college students (for similar results see Hunsberger, 1978). Mayrl and Uecker (2011) also found that respondents who had attended college were not more unorthodox, naturalistic, reserved, inclusive, individualistic, or independent than those who did not.

Uecker et al. (2007) speculate that these null findings are due to the lack of engagement of most college students with the intellectual life of a college environment, especially on matters of ‘morality or meaning’. Instead, most students treat the college experience pragmatically, as a means to a career. Others may not be religiously savvy enough to recognize challenges to their religious faith. Overall, though, the authors believe American college campuses are becoming less hostile to religion, and possibly less hostile than other contexts experienced by non-collegians, such as the workplace.

However, these recent studies run counter to the received wisdom that religious faith declines as a result of attending college. Wuthnow (1988) observed that the college-educated were more likely to hold religiously liberal views, including those about the Bible. Feldman and Newcomb (1969: 326) attributed declines in religious belief to ‘declining “authoritarianism”, dogmatism, and prejudice, together with decreasingly conservative attitudes’ among students as they progressed through school. College education is also associated with declines in denominational loyalty (Roof, 1989; Smith and Sikkink, 2003; Wuthnow, 1988).

What accounts for the mixed results about changes in student religiosity across empirical studies? Increasingly, researchers are investigating institution-level processes.

Institutional context

Recent evidence suggests that institutional characteristics moderate changes in religious belief during college. Reimer (2010), for instance, found that attending religious institutions in the US mitigates the effect of exposure to secular theories and philosophies during college somewhat. Similarly, Hill (2009: 530) found that student religiosity was more robust at conservative Protestant universities than at Catholic or mainline Protestant universities because the former act as ‘moral communities’ that ‘effectively create overlapping networks of like-minded individuals that reinforce and legitimate religious life by creating a shared moral order’. Davignon and Thomson (2015) found that institutional factors at conservative Christian institutions including student perception of the availability of spiritual mentors and homogeneity of religious belief are protective of religious belief and practice.

That institutional religious diversity is inversely related to religiosity is compatible with both the ‘moral communities thesis’ (Stark, 1996) and secularization theory. According to Berger (1967), religious monopolies can effectively maintain the credibility of their sacred understanding of the world, but competition among ‘plausibility structures’ undermines the credibility of all of them. That said, secularization theories have been challenged by theorists advocating a ‘new paradigm’ positing pluralism to be positively related to individual-level religiosity because of supply-side competition in unregulated ‘religious markets’ (Stark and Finke, 2000; Warner, 1993). Even Berger (2014), citing evidence such as the persistence of high levels of religiosity in the US and the rapid growth of global Pentecostalism, has recanted his original secularization formula, suggesting that what is now needed in the sociology of religion is a theory of pluralism. Modernity does not secularize societies, he claims, but rather it relativizes them. Particular religious beliefs that may have once been taken for granted are increasingly recognized as choices among many alternatives. The challenge for the modern believer is to negotiate their commitment to faith with the experience of doubt and provide cognitive room for uncertainty.

Berger’s modified thesis is empirically justifiable. Dense religious peer networks strongly predict absolutist attitudes among American teens (Trinitapoli, 2007), and individuals who attended college have reported marginally more religious doubt than those who did not (Mayrl and Uecker, 2011). Hill (2011: 546) concludes that American college students are not abandoning Christian beliefs and identity, but they are exhibiting more skepticism ‘of the existence of super-empirical entities’. Thus, to the extent that college is an environment where diversity is encountered, previously held religious beliefs may either be re-evaluated or integrated with other forms of knowledge, including scientific knowledge or nontraditional religious beliefs.

The current study

We set out to test Berger’s (2014) reformulated theory of pluralism and apply it to changes in religious belief among students of higher education. Increasing education attainment has likely played a role in the declining absolutist belief in the US after World War II (Stroope, 2011; Wuthnow, 1988), but institutional variance in changing beliefs is not well understood. If pluralism relativizes religious belief, then institutional diversity should have an inverse relationship with student-level absolutist belief, even at Christian colleges and universities. We also expect that institution-level diversity conditions the student-level effect of academic tenure. Upperclassmen have had a longer exposure to a multiplicity of ‘plausibility structures’, so they should be less likely to hold absolutist effects than their underclassman counterparts, and the gap should be greatest at the most diverse schools. Formally, we hypothesize that:

H1: Absolutist religious belief declines with increasing institutional diversity.

H2: Absolutist religious belief declines with increasing tenure (academic rank).

H3: Institutional diversity amplifies the effect of academic rank on absolutist religious beliefs.

We consider absolutist religious beliefs related to both salvation and the Bible, as each have important social implications. Those who believe in one true path to heaven (soteriological exclusivism) tend to be highly religious, deferential to religious authority, and morally and politically conservative (Putnam and Campbell, 2010), and their worldview tends to be maintained by the cultivation of social relationships in restricted and homogenous networks (Wuthnow, 2007). The belief that every word of the Christian Bible is true and accurate (biblical literalism) relates to views about politics (Franzen and Griebel, 2013), gender (Hoffmann and Bartkowski, 2008), the environment (Sherkat and Ellison, 2007), and criminal punishment (Grasmick et al., 1992).

We also consider the effects of both religious and political diversity. The latter has been relatively unexplored, but recent trends in disaffiliation suggest the possible causal efficacy of political factors. Specifically, the proportion of Americans who report no religious affiliation has been rising substantially in recent decades, especially among political moderates and liberals (Hout and Fischer, 2002). Putnam and Campbell (2010) characterize this pattern as a reaction against the convergence of religious and political conservatism in the 1970s and 1980s among cohorts who came of age in the 1990s and 2000s due to their relatively liberal views on issues such as homosexuality and marijuana use. If political tension is spurring recent religious disaffiliation, then it is also possible that educational contexts characterized by relative diversity of political attitudes can also relativize religious belief. Conversely, schools characterized by relative homogeneity, religious or political, may be protective of religious absolutism. Religious exclusivists, according to Wuthnow (2007), tend to be have relatively restricted social networks of like-minded individuals; these types of networks can function as ‘echo chambers’ that reinforce and even strengthen belief (Putnam and Campbell, 2010).

Data and methods

The data for this study come from the CCCU Denominational Study (see Davignon et al., 2013; Glanzer et al., 2013; Rine et al., 2013), which was commissioned by the Council for Christian Colleges & Universities (CCCU). The CCCU is an association of theologically conservative Christian colleges and universities that restricts its membership to institutions that actively integrate the Christian faith with teaching and scholarship (Council for Christian Colleges & Universities Profile, 2014). Students who attend CCCU institutions often do so because of their explicit Christian identities (Davignon, 2016). While these institutions represent a niche within higher education in the US, they are ideal for testing the effects of religious pluralism because religion is a salient feature of the institutional context. They also serve as ‘moral communities’ that protect, and even nurture, religious belief and behavior (Davignon and Thomson, 2015; Hill, 2009). Any relativizing effects can be more readily attributed to religious pluralism rather than potentially secularizing influences because these institutions are overtly opposed to the latter.

The data used here come from Phase III of the study, which focused on student religious practices and attitudes (Davignon et al., 2013). Researchers invited each of the 114 CCCU member institutions to participate in the study, and 32 of these schools agreed to administer an online survey to their undergraduate student bodies via email in the fall of 2012. Of these schools, less than 30 students completed the survey at one institution, which was therefore excluded from the final analysis. The purpose of this research study is to examine how institutional characteristics at CCCU schools influence student religious belief, not to generalize to all CCCU institutions. Even so, previous research using this sample suggest that the participating institutions are fairly representative of the CCCU as a whole (Davignon and Thomson, 2015). The resulting sample consisted of 31 CCCU institutions and 6,106 student responses, with student responses per school ranging from 36 to 502. The overall response rate of the online survey was 11%, which reflects the fact that online surveys tend to have lower response rates (Couper et al., 1997). This response rate is nearly identical to a recent large-scale survey on American religious identity¹. While low response rates plague many recent surveys on religion, they are not necessarily related to non-response bias, which nevertheless can be reduced by weighting for demographic characteristics (Groves, 2006). Therefore, while our sample of college students is comparable to the population in terms of sex and race (Davignon and Thomson, 2015), we control for these measures in our analysis. Nevertheless, we acknowledge the relatively low response rate as a possibly limitation of this study.

Dependent variables

Outcomes included two dichotomous variables representing absolutist beliefs about salvation and the bible. The former (soteriological exclusivism) was indicated by the statement ‘Christ is the only source of salvation, and only believers in Christ will receive salvation’. The latter (biblical literalism) was indicated by the belief that ‘The Bible is the inspired Word of God, not mistaken in its statements and teachings, and is to be taken literally, word for word’. All other responses to respective belief questions (including ‘don’t know’) were collapsed into reference categories.

Independent variables

At the school level, the central characteristic of interest was diversity. Religious diversity was operationalized with two measures. First, religious diversity (affiliation) was constructed as a Herfindahl index, or the sum of the squares of market shares (Σ $p_{i}^{2}$ ), based on the proportions of students at each school from different traditions. Because students at CCCU schools are predominantly evangelical, Steensland et al.’s (2000) religious tradition categories did not yield much variation. Nevertheless, sectarian institutions of higher learning are increasingly admitting students who do not share the denomination of the institution (Glanzer et al., 2013; Ross, 2011), so we instead categorized affiliation by the religious ‘families’ defined by the Association of Religious Data Archives (ARDA)². Second, we also considered religious diversity (beliefs), operationalized as the proportion of students at each school that identified as both ‘born again’ and ‘Bible believing’, following Davignon and Thomson (2015).

Political diversity was also constructed as a Herfindahl index based on student indication of their political ideology on a seven-point scale (1 = ‘extremely conservative’ to 7 = ‘extremely liberal’). Given that the original measure is ordinal (in contrast with the nominal categories used to construct the religious affiliation diversity measure), the political diversity measure has a particular limitation, namely that a school with mostly conservatives can have a score similar to one that is mostly liberal. Because most students in the sample are politically conservative (about 63.7% identify as ‘leaning conservative’ to ‘extremely conservative’ while only 16.9% identify as ‘leaning liberal’ to ‘extremely liberal’), though, the diversity measure was closely correlated with a measure of average political liberalism of students by campus (r=.698, p<.001). We therefore control for any confounding effect due to the relative political ethos at a school by also controlling for campus liberalism, an aggregated measure of the average political liberalism of students at each school.

Raw Herfindahl index scores were subtracted from a value of 1, thereby ranging from 0 (representing homogeneity) to nearly 1 (where many groups have equal shares). Schools were relatively diverse, with average religious diversity (affiliation) and political diversity scores of .716 and .762, respectively (Table 1). Schools had an average of 62.3% of students who were both born again and Bible-believing, and average campus liberalism was 3.108, corresponding to ‘leaning conservative’. The two religious diversity indices were moderately, but not significantly, correlated (r=.338, p=.063). Religious diversity (affiliation) was significantly correlated to political diversity (r=.387, p=.032), but religious diversity (beliefs) was not (r=.338, p=.063).

Table 1.

Descriptive statistics.

	N	Mean	SD	Minimum	Maximum
Dependent variables
Soteriological exclusivism	6069	.533	0.499	0	1
Biblical literalism	6064	.339	0.474	0	1
School-level independent variables
Religious diversity (affiliation)	31	.716	.093	.493	.842
Religious diversity (beliefs)	31	.377	.113	.179	.656
Political diversity	31	.762	.035	.654	.811
Campus liberalism	31	3.108	.363	2.574	4.548
Student-level independent variables
Academic rank	6106	2.524	1.124	1	4
Female	6095	.690	.462	0	1
White non-Hispanic	6067	.814	.389	0	1
GPA	5415	4.381	.813	1	5
Mother’s education	6048	5.182	1.479	1	7
Father’s education	5951	5.320	1.607	1	7
Friends in denomination	5230	2.810	1.084	1	5
Christian friends	5956	4.244	.701	1	5
Religious behavior	6056	9.672	2.980	3	24
Fundamentalist	6106	.082	.274	0	1
Bible-believing and born again	6106	.633	.482	0	1
Bible-believing and born again (parental)	6106	.584	.493	0	1
Student liberalism	5391	3.139	1.407	1	7

At the student level, academic rank (i.e., years in school) was measured as a four-point ordinal scale ranging from freshman (1) to senior (4), each of which represented about a quarter of the sample. Student-level variables for political liberalism and Bible-believing and born again were included using the same measures used to construct the corresponding school-level variables. A proxy for student’s initial religiosity was controlled by using a parental version of Bible-believing and born again; the dichotomous measure was assigned a value of 1 if a student identified at least one parent by both labels (58.5% of sample). To control for the effect of social network composition (Hill, 2009; Smith and Sikkink, 2003; Uecker et al., 2007), two additional measures for the number of respondents’ (1) friends in (the same) denomination and (2) Christian friends (ranges from 1 = ‘none’ to 5 = ‘all’) were also included. Respondents tended to have more Christian friends than friends in their denomination, which provided face validity for the use of religious diversity measures; respondent social networks at college tended to include friends from a variety of Christian traditions, so religious diversity at a particular school could be a salient factor influencing absolutist religious belief.

We also controlled for student religious behavior, or the sum of responses to three items including hours during a typical week spent in (1) ‘Prayer/meditation’, (2) ‘Bible reading’, and (3) ‘Church attendance and activities’ (α=.78). Responses to each ranged from 1 = ‘none’ to 8 = ‘over 20’ (average sum of 9.672). And because fundamentalist identity is associated with the outcome variables, especially biblical literalism (Ammerman, 1982), we included a dichotomous fundamentalist measure based on self-identification (8.2% of sample).

Sociodemographic controls included gender (female=1, 69% of sample), race/ethnicity (white non-Hispanic=1, 81% of sample), and both mother’s and father’s education, each measured on a 7-point ordinal scale (1 = ‘8th grade or less’ to 7 = ‘postgraduate work/degree’). Average mother’s and father’s education was 5.2 and 5.3, respectively, suggesting students’ parents averaged nearly a bachelor’s degree in education. Grade point average (GPA) was measured as a five point ordinal scale ranging from 1 = ‘less than 2.0’ to 5=‘greater than 3.5’, with an average GPA category of 4.4 (between ‘3.0-3.49’ and ‘greater than 3.5’).

Prior to multivariate analyses, all indices and ordinal measures, including religious diversity (affiliation), religious diversity (beliefs), political diversity, campus liberalism, academic rank, political liberalism, GPA, mother’s education, father’s education, friends in denomination, Christian friends, and religious behavior, were mean-centered to reduce the potential for multicollinearity.

Analytical method

This study conceptualizes students as nested within particular school environments (see Davignon and Thomson, 2015; Hill, 2009) that are characterized by varying degrees of religious and political diversity. To assess hypotheses regarding school-, student-, and cross-level effects, multilevel regression modeling (Raudenbush and Bryk, 2002) was utilized, as it allowed for the randomization of intercept slopes for student-level characteristics across schools. Since dependent variables were dichotomous, the GLIMMIX procedure in SAS 9.2 was performed. Soteriological exclusivism (53.3% of the sample; see Table 1) and biblical literalism (33.9% of the sample) were modeled as Bernoulli distributions with a logit link function.

Null models (not reported) indicated that significant variation of the dependent variables existed across schools. The intraclass correlation coefficient (ICC), which indicates both between-school heterogeneity and within-school homogeneity, for soteriological exclusivism was:

ICC = \frac{{\hat{σ}}_{u 0}^{2}}{{\hat{σ}}_{u 0}^{2} + π^{2} / 3} = \frac{. 2921}{. 2921 + 3.289} = . 0815

And the ICC for biblical literalism was:

ICC = \frac{{\hat{σ}}_{u 0}^{2}}{{\hat{σ}}_{u 0}^{2} + π^{2} / 3} = \frac{. 3760}{. 3760 + 3.289} = . 1026

Because 8 to 10% of student variance can be attributed to school-level effects, multilevel modeling is necessary due to the potential for large Type I errors in single-level model estimation (Wang et al., 2011). For both outcome measures, the effects of each diversity measure were modeled with student-level effects separately first, then all three were combined in full models. Cross-level effects were then tested for each of the three diversity measures on both dependent variables. Notably, preliminary testing revealed that all student-level variables should remain fixed rather than be permitted to be random. Further testing was also performed to determine if any schools represent influential observations based on its Cook’s D statistic; if so, they were omitted from the sample for the modeling of a particular outcome.

Results

Estimated parameters from models predicting the effects of school diversity on soteriological exclusivism are reported in Table 2. Support for the first hypothesis was mixed. At the school-level, diversity of religious belief (-2.57) was inversely related to soteriological exclusivism in Model 2, as was both political diversity (-8.354) and campus liberalism (-.496) in Model 3. When all school-level effects were estimated simultaneously (Model 4), however, only political diversity remained significant (-9.323). At the student level, academic rank was significant in each model, as predicted in H2; the odds that respondents held soteriological exclusivist beliefs decreased by 10.3% (=1–e^-.109) for each incremental increase in academic rank. Other significant effects at the student level included friends in denomination (.102), religious behavior (.057), and being Bible-believing and born again (.387), each of which had consistently positive effects, as well as student liberalism (-.305), which had a negative effect.

Table 2.

Unstandardized coefficients from multilevel models of soteriological exclusivism.

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6		Model 7
	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE
Intercept	−.063	.138	−.046	.131	−.091	.121	−.084	.120	−.064	.138	−.049	.132	−.095	.121
Religious diversity (affiliation)	−1.725	.921					.000	.675	−1.662	.924
Religious diversity (beliefs)			−2.57***	.674			−1.054	.681			−2.456**	.681
Political diversity					−8.354**	2.703	−9.323**	2.947					−8.175**	2.737
Campus liberalism					−.496*	.230	−.238	.278					−.483*	.232
Academic rank	−.110***	.032	−.110***	.032	−.108***	.032	−.109***	.032	−.109***	.032	−.113***	.032	−.112***	.033
Academic rank X religious diversity (affiliation)									−.297	.363
Academic rank X religious diversity (beliefs)											−.773*	.322
Academic rank X political diversity													1.148	1.855
Academic rank X campus liberalism													−.346*	.157
Female	.048	.074	.054	.074	.036	.074	.039	.074	.049	.074	.054	.074	.035	.074
White non-Hispanic	−.124	.100	−.113	.100	−.124	.099	−.119	.099	−.123	.100	−.113	.100	−.124	.099
GPA	.064	.046	.064	.046	.064	.045	.064	.045	.063	.046	.065	.046	.065	.045
Mother’s education	−.006	.028	−.004	.028	−.001	.028	.000	.028	−.006	.028	−.005	.028	−.002	.028
Father’s education	.024	.026	.022	.026	.025	.026	.023	.026	.024	.026	.022	.026	.025	.026
Friends in denomination	.104**	.034	.109**	.034	.102**	.034	.102**	.034	.103**	.034	.107**	.034	.102**	.034
Christian friends	−.003	.055	−.027	.055	.005	.054	−.010	.055	−.002	.055	−.021	.055	.007	.054
Religious behavior	.059***	.013	.058***	.013	.058***	.013	.057***	.013	.059***	.013	.058***	.013	.058**	.013
Fundamentalist	.222	.130	.217	.130	.216	.130	.212	.130	.223	.130	.224	.130	.218	.130
Bible-believing and born again	.398**	.122	.383**	.122	.398**	.122	.387**	.122	.395**	.122	.376**	.122	.391**	.122
Bible-believing and born again (parental)	.178	.118	.175	.118	.174	.118	.172	.118	.180	.118	.184	.118	.185	.118
Student liberalism	−.314***	.027	−.313***	.027	−.304***	.027	−.305***	.027	−.314***	.027	−.314***	.027	−.304***	.027
Var[random int]	.415***	.070	.345***	.063	.212***	.050	.199***	.049	.415***	.070	.349***	.063	.214***	.050
-2 Log Likelihood	5077.05		5068.37		5048.49		5046.15		5076.38		5062.59		5041.20
Pearson Chi-Square / DF	.99		.99		1.00		1.00		.99		.99		.99

Notes: N (student) = 4094, N (school) = 31. Weighted at the school level. *p<.05 **p<.01 ***p<.001 (two-tailed tests).

Coefficients from models estimating cross-level interactions for the soteriological exclusivism outcome are also reported in Table 2. In support of H3, the interaction between religious diversity (beliefs) and academic rank was significant in Model 6, as were both of the main effects. With increasing religious diversity, the negative effect of academic rank was amplified (-.773). As illustrated in Figure 1, the probability of being soteriological exclusivist decreases more with academic rank in environments with diverse religious beliefs than in more homogenous environments. The difference is stark, too; while the predicted probability that seniors hold absolutist soteriological beliefs is only about 3% (=(.578-.560)/.560) lower than that of freshman at schools with religious diversity at one standard deviation below the mean, it declines by 28.9% (=(.507-.361)/.507) from the freshman rank to the senior rank among students at schools with religious diversity at one standard deviation above the mean.

Figure 1.

The interactive effects of diversity in religious belief and academic rank on soteriological exclusivism.

Further, while political diversity was not found to amplify the negative effect of academic rank contrary to H3, the political ethos of an institution did condition the effect of academic rank on soteriological exclusivism. Specifically, the negative effect of academic rank was amplified with increasing political liberalism (-.346; see Figure 2). At the most conservative schools, seniors were actually slightly more likely to hold absolutist beliefs than freshman, but at schools where average student liberalism was highest, declines in soteriological exclusivism were particularly sharp. Where campus liberalism was one standard deviation above the mean, the predicted probability that seniors held absolutist belief was about a 32.1% (=(.547-.371)/.547) lower than that of freshman.

Figure 2.

The interactive effects of campus liberalism and academic rank on soteriological exclusivism.

Models predicting biblical literalism are reported in Table 3. Religious diversity (affiliation) had a significant negative effect (-1.724) in Model 1, as did political diversity (-14.928) in Model 3. However, as with soteriological exclusivism, only political diversity was significant in the combined model (Model 4). Academic rank was again significant in every model, in support of H2. For every incremental rise in academic rank, the odds of holding biblical literalist beliefs decreased by 16.6% (=1-e^-.182) in Model 4. Student level Christian friends (-.162 in Model 4) had a surprising but consistent negative effect, as did student liberalism (-.354), while religious behavior (.085), fundamentalist (.538), and Bible-believing and born again (.366) each had consistently positive effects.

Table 3.

Unstandardized coefficients from multilevel models of biblical literalism.

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6		Model 7
	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE	Est.	SE
Intercept	−.924***	.162	−.936***	.161	−1.003***	.139	−1.014***	.139	−.933***	.163	−.937***	.161	−1.011***	.139
Religious diversity (affiliation)	−1.724***	1.297					1.132	.938	−1.548	1.302
Religious diversity (beliefs)			−1.802	1.063			−.685	.816			−1.784	1.066
Political diversity					−14.928***	3.866	−15.723***	3.888					−14.348**	3.889
Campus liberalism					−.299	.473	−.281	.526					−.318	.474
Academic rank	−.190***	.035	−.191***	.035	−.187***	.035	−.187***	.035	−.186***	.035	−.194***	.036	−.178***	.038
Academic rank X religious diversity (affiliation)									−1.251**	.430
Academic rank X religious diversity (beliefs)											−.209	.387
Academic rank X political diversity													−3.916	2.450
Academic rank X political diversity
Academic rank X campus liberalism													.246	.283
Female	.109	.082	.112	.082	.095	.082	.098	.082	.115	.082	.112	.082	.101	.082
White non–Hispanic	−.142	.107	−.138	.107	−.153	.107	−.151	.107	−.139	.107	−.138	.107	−.149	.107
GPA	−.055	.050	−.055	.050	−.051	.050	−.052	.050	−.059	.050	−.055	.050	−.055	.050
Mother’s education	−.042	.030	−.042	.030	−.039	.030	−.039	.030	−.041	.030	−.042	.030	−.038	.030
Father’s education	.020	.028	.019	.028	.020	.028	.019	.028	.019	.028	.019	.028	.020	.028
Friends in denomination	.052	.038	.055	.038	.047	.037	.053	.038	.053	.038	.054	.038	.048	.037
Christian friends	−.163**	.060	−.173**	.061	−.152*	.059	−.162**	.060	−.159**	.060	−.172**	.061	−.153*	.059
Religious behavior	.084***	.014	.085***	.014	.085***	.014	.085***	.014	.085***	.014	.084***	.014	.085***	.014
Fundamentalist	.536***	.129	.536***	.129	.537***	.128	.538***	.128	.543***	.129	.537***	.129	.525***	.129
Bible–believing and born again	.363**	.140	.357*	.140	.371**	.140	.366**	.140	.350*	.140	.355*	.140	.369**	.140
Bible–believing and born again (parental)	.101	.133	.101	.133	.089	.133	.091	.133	.110	.133	.103	.133	.087	.133
Student liberalism	−.357***	.031	−.358***	.031	−.353***	.031	−.354***	.031	−.358***	.031	−.358***	.031	−.352***	.031
Var[random int]	.533***	.087	.520***	.085	.278***	.058	.268***	.056	.535***	.087	.521***	.086	.278***	.058
−2 Log Likelihood	4335.48		4334.48		4305.18		4303.48		4327.02		4334.18		4302.09
Pearson Chi–Square / DF	.98		.98		.98		.98		.98		.98		.98

Notes: N (student) = 3795, N (school) = 29. Influential observations from schools 23 and 28 were omitted. Weighted at the school level.

p<.05 **p<.01 ***p<.001 (two-tailed tests).

Models regressing biblical literalism on cross-level effects partially supported H3. While the main effect for religious diversity (affiliation) was not significant, its interaction with academic rank was, as was the main effect of academic rank (Model 5). As with soteriological exclusivism, diversity in religious affiliation amplified the negative effect of academic rank on biblical literalism (-1.251), but the effect is much less dramatic (see Figure 3) since the main effect of religious diversity is not significant. Still, declines from freshman to senior ranks at schools with religious diversity at one standard deviation above the mean (47.7%=(.384-.201)/.384) were more than three times greater than declines at schools with religious diversity at one standard deviation below the mean (14.0%=(.305-.262)/.305).

Figure 3.

The interactive effects of diversity in religious affiliation and academic rank on biblical literalism.

Finally, contrary to expectations from H3, the interaction between political diversity and academic rank in predicting biblical literalism was not significant, though both main effects were in Model 7.

Discussion

Hypotheses related to the relativizing effect of diversity were generally supported. Using multilevel modeling techniques, results confirmed that religious and political diversity have negative effects on soteriological exclusivism and biblical literalism among students at Christian colleges and universities in the US. Interestingly, inclusion of political diversity in combined models rendered the effect of religious diversity non-significant. Consistent with moral communities literature (Davignon and Thomson, 2015; Hill, 2009), religious and (especially) political homogeneity at conservative Christian schools protects absolutist belief.

Further, academic rank also had a negative effect on soteriological exclusivism and biblical literalism, suggesting that time of exposure to a diverse environment has an accumulating effect, even controlling for parental religiosity to account for the relative level of religiosity of students as incoming freshman. These findings generally support Berger’s revised theory that pluralism arising in modernity does not secularize per se, but rather it relativizes³. The implication is that these students, while not abandoning their faith altogether, are making room for uncertainty as they must increasingly negotiate between competing truth claims and plausibility structures. While faith increasingly coexists with doubt, doubt does not replace faith altogether.

As expected, the negative effect of academic rank was amplified by religious diversity, and it was also conditioned by the political ethos of a campus; the more diverse and liberal the students at a particular school, the stronger the effect of academic rank on absolutist belief. Interestingly, though, having Christian friends at school also had an independent negative effect on biblical literalism, suggesting that simply engaging in discourse among more diverse Christian friends outside a student’s own denominational tradition could diminish commitment to a particular reading of the Bible, thus undermining perception of the Bible’s inerrancy.

Finally, evidence suggests that diversity of political attitudes may be the primary form of plurality by which commitment to absolutist religious views is softened at conservative Christian colleges and universities in the US, given that it explained the effects of religious diversity, both in affiliation and diversity, in combined models. In Bergerian terms, it seems that interactions with politically diverse peers in a college context are having the effect of undermining the certainty requisite for holding absolutist views about salvation and the Bible. Conversely, the ‘echo chambers’ of homogenous political environments (Putnam and Campbell, 2010; Wuthnow, 2007) seem to be protecting absolutist belief. Perhaps increased political diversity increases political dialogue and conflict on campus. As statistical minorities at these schools (only 16.91% are ‘leaning liberal’ to ‘extremely liberal’, and only 1.5% are ‘extremely liberal’), discursive political conflict in a predominantly conservative context may serve to crystalize a liberal subcultural identity (see Smith, 1998) that reinforces relativist religious beliefs rather than absolutist beliefs. Further investigation is warranted to determine if the ‘backlash’ of disaffiliation among recent cohorts in reaction to convergence of political and religious conservatism (Hout and Fischer, 2002; Putnam and Campbell, 2010) is manifested on college campuses by changes in other types of religious beliefs and behaviors as well.

One implication for the observed relationship between diversity and absolutist belief relates to student involvement in the sciences. While most religious Americans support institutional science and most scientists do not perceive religion and science to be in conflict (Baker, 2012; Ecklund and Park, 2009; Evans and Evans, 2008), those with an exclusive view on salvation constitute an important exception. Longest and Smith (2011) found that individuals who believe the only way to heaven is through Jesus are more likely to adopt a conflict narrative between science and religion, which could contribute to selection of exclusivists out of scientific and academic majors and career paths. These findings suggest that exposure to political and religious diversity could ameliorate this effect.

The interpretations offered here should be considered with caution, given that the data is cross-sectional rather than longitudinal. While the association between academic rank and absolutist belief is robust and in the hypothesized direction, we cannot definitively rule out cohort/period, life-course, or selection explanations. The observed phenomenon may simply be continuation of the politically-based trend in disaffiliation. And because our sample does not include a comparison group of young adults who did not attend college, declines in absolutist belief may be due to normal maturation processes. Student attraction to more or less diverse institutions also might correlate to open-mindedness.

We have attempted, however, to mitigate these limitations. First, it is unlikely that each cohort of incoming freshman is increasingly relativist due to an exogenous cultural effect given the narrow range of birth-years represented here. Further, to the degree that parental religious belief serves as a proxy for initial religiosity at the start of college, there seems to be no cohort effect on either of the outcomes considered here. Second, we controlled for cross-level interactions. Even if an aging effect is occurring, that the effect is stronger in more diverse institutions supports the relativizing thesis. Finally, the effect of academic rank remained strong and significant even when controlling for all measures of institutional diversity. Thus, students may be self-selecting into environments that are conducive to their beliefs, but to the degree that academic rank is a proxy for cumulative exposure to diversity, its effect is occurring in diverse and homogenous contexts alike.

Our findings, therefore, support the relativizing effect of education-diversity thesis, and that the sample was limited to respondents at conservative Christian institutions suggests that these effects are related to the experiences of diversity and education themselves, rather than exposure to secular faculty. But the American context is distinct, especially among post-industrial societies, both for its level of religiosity (Norris and Inglehart, 2011) and the degree to which religion influences political behavior. Many Americans, according to Froese (2014), believe God is both interested in politics and favors the conservative party. Political diversity may thus be more potent in affecting religious belief in the US than in contexts where culture and politics are more secular.

And while our sample is particularly well-suited for testing the relativizing effect of diversity, it is not ideal for observing secularization because conservative Christian institutions of higher learning are special cases that function to preserve religious faith (Hill, 2009). It is possible that the effects of diversity are amplified in contexts where conservative students are less able to develop tight social networks of like-minded individuals. The effects could be further compounded by the disproportionately non-religious faculty at elite institutions (Ecklund and Scheitle, 2007). It is thus premature to generally conclude that diversity on college campuses does not secularize religious belief. On the other hand, some have suggested that religious students at Christian schools struggle more with their faith because they are encouraged to work through difficult questions related to the truth claims of their tradition (Bryant and Astin, 2008). Whether greater religious diversity at public schools serves to reinforce absolutist beliefs by providing a context of increased conflict with negative reference groups (c.f. Hammond and Hunter, 1984; Hill, 2009; Smith, 1998) also remains to be studied. Additional research should expand the scope of this study beyond conservative Christian universities in the US to consider the effects of diversity in more liberal, and in cross-national, contexts.

Footnotes

Appendix

Appendix A.

Included denominations and traditions by religious family.

Baptist ( N =987, 16.2%)	Non-denominational Christian ( N =1989, 32.6%)
American Baptist	Non-denominational Christian
Church of Christ
Disciples of Christ	Pentecostal/Charismatic ( N =647, 10.6%)
General Conference/Converge Worldwide	Assemblies of God
Independent Baptist	Church of God - Cleveland
National Baptist	Hope Chapel
Southern Baptist	Four Square
State Baptist Conventions	Vineyard
Other Baptist	Other Pentecostal/Charismatic

Catholic ( N =159, 2.6%)	Reformed/Presbyterian ( N =359, 5.9%)
Catholic/Roman Catholic	Christian Reformed
	Presbyterain Church (U.S.A.)
European Free Church ( N =204, 3.3%)	Presbyterian Church in America
Brethren in Christ	Reformed Church in America
Fellowship of Grace	Reformed Presbyterian Church in America
Mennonite Brethren	Other Presbyterian
Mennonite Church U.S.A.	Other Reformed
Quaker/Friends
Other Anabaptist	Other Christian ( N =480, 7.9%)
Other Brethren	Anglican/Episcopalian^a
	Bible Church
Holiness ( N =643, 10.5%)	Congregational^a
Christian & Missionary Alliance	Eastern Orthodox
Church of God - Anderson	Seventh Day Adventist^a
Church of the Nazarene
Salvation Army	Other Religion, Non-Christian ( N =31, .5%)
Wesleyan	Buddhism
	Hinduism
Lutheran ( N =104, 1.7%)	Islam
Evangelical Lutheran Church in America	Other
Lutheran Church - Missouri Synod
Wisconsin Evangelical Lutheran Synod	Atheist/Agnostic ( N =52, .9%)
Other Lutheran	Agnostic
	Atheist
Methodist/Piety ( N =451, 7.4%)
Evangelical Covenant Church
Evangelical Free Church
Free Methodist Church
United Brethren
United Methodist Church
Other Evangelical
Other Methodist

Note: Total N=6106. ^aIncluded with ‘Other Christian’, despite classification in distinct families by the ARDA (http://www.thearda.com/denoms/families/groups.asp), due to relatively low representation in the total sample.

Acknowledgements

The authors thank Holly Collins for assistance with French translation.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

Author biographies

Robert A THOMSON, Jr. is a Postdoctoral Research Fellow in the Religion and Public Life Program and the Department of Sociology at Rice University.

Address: P.O. Box 1892, Houston, Texas 77251, USA.

Email: bob_thomson@baylor.edu

Phil DAVIGNON is an Assistant Professor of Sociology at Union University. He has published research in the Review of Religious Research, Religion and Education, and Christian Higher Education.

Address: 1050 Union University Drive, Jackson, TN 38305, USA.

Email: pdavignon@uu.edu

References

Ammerman

(1982) Operationalizing Evangelicalism: An amendment. Sociological Analysis 43(2): 170–171.

Arnett

(2000) Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist 55(5): 469–480.

Baker

(2012) Perceptions of science and American secularism. Sociological Perspectives 55(1): 167–188.

Berger

(1967) The sacred canopy: Elements of a sociological theory of religion. New York: Anchor Books, Doubleday.

Berger

(2014) The many altars of modernity: Toward a paradigm for religion in a pluralist age. Boston, MA: Walter de Gruyter.

Bryant

Astin

(2008) The correlates of spiritual struggle during the college years. The Journal of Higher Education 79(1): 1–27.

Council for Christian Colleges and Universities Profile (2014). Available at: http://www.cccu.org/about/profile (accessed 15 January 2014).

Couper

Blair

Triplett

(1997) A comparison of mail and e-mail for a survey of employees in federal statistical agencies. Paper presented at the Annual Meeting of the American Association for Public Opinion Research, Norfolk, VA.

Davignon

(2016) Factors Influencing college choice and satisfaction among students at Christian colleges and universities. Religion and Education 43(1): 77–94.

10.

Davignon

Thomson

Jr (2015) Christian colleges and universities as moral communities: The effects of institutional characteristics on student religiosity. Review of Religious Research 57(4): 531–554.

11.

Davignon

Glanzer

Rine

(2013) Assessing the denominational identity of American Evangelical colleges and universities. Part III: The student experience. Christian Higher Education 12(5): 315–330.

12.

Ecklund

Park

(2009) Conflict between religion and science among academic scientists? Journal for the Scientific Study of Religion 48(2): 276–292.

13.

Ecklund

Scheitle

(2007) Religion among academic scientists: Distinctions, disciplines, and demographics. Social Problems 54(2): 289–307.

14.

Evans

(2008) Religion and science: Beyond the epistemological conflict narrative. Annual Review of Sociology 34: 87–105.

15.

Feldman

Newcomb

(1969) The impact of college on students. Piscataway, NJ: Transaction Publishers.

16.

Franzen

Griebel

(2013) Understanding a cultural identity: The confluence of education, politics, and religion within the American concept of biblical literalism. Sociology of Religion 74(4): 521–543.

17.

Froese

(2014) Religion and American politics from a global perspective. Religions 5(3): 648–662.

18.

Glanzer

Rine

Davignon

(2013) Assessing the denominational identity of American evangelical colleges and universities, Part I: Denominational patronage and institutional policy. Christian Higher Education 12(3): 181–202.

19.

Grasmick

Davenport

Chamlin

Bursik

(1992) Protestant fundamentalism and the retributive doctrine of punishment. Criminology 30(1): 21–46.

20.

Groves

(2006) Nonresponse rates and nonresponse bias in household surveys. Public Opinion Quarterly 70(5): 646–675.

21.

Hammond

Hunter

(1984) On maintaining plausibility: The worldview of evangelical college students. Journal for the Scientific Study of Religion 23(3): 221.

22.

Hill

(2009) Higher education as moral community: Institutional influences on religious participation during college. Journal for the Scientific Study of Religion 48(3): 515–534.

23.

Hill

(2011) Faith and understanding: Specifying the impact of higher education on religious belief. Journal for the Scientific Study of Religion 50(3): 533–551.

24.

Hoffmann

Bartkowski

(2008) Gender, religious tradition, and biblical literalism. Social Forces 86(3): 1245–1272.

25.

Hout

Fischer

(2002) Why more Americans have no religious preference: Politics and generations. American Sociological Review 67(2): 165–190.

26.

Hunsberger

(1978) The religiosity of college students: Stability and change over years at university. Journal for the Scientific Study of Religion 17(2): 159–164.

27.

Longest

Smith

(2011) Conflicting or compatible: Beliefs about religion and science among emerging adults in the United States. Sociological Forum 26(4): 846–869.

28.

Mayrl

Oeur

(2009) Religion and higher education: Current knowledge and directions for future research. Journal for the Scientific Study of Religion 48(2): 260–275.

29.

Mayrl

Uecker

(2011) Higher education and religious liberalization among young adults. Social Forces 90(1): 181–208.

30.

Nelms

Hutchins

Pursley

(2007) Spirituality and the health of college students. Journal of Religion and Health 46(2): 249–265.

31.

Norris

Inglehart

(2011) Sacred and secular: Religion and politics worldwide (second edition). Cambridge: Cambridge University Press.

32.

Perkins

(1987) Parental religion and alcohol use problems as intergenerational predictors of problem drinking among college youth. Journal for the Scientific Study of Religion 26(3): 340–357.

33.

Putnam

Campbell

(2010) American grace: How religion divides and unites us. New York: Simon & Schuster.

34.

Raudenbush

Bryk

(2002) Hierarchical linear models: Applications and data analysis methods. London: SAGE.

35.

Reimer

(2010) Higher education and theological liberalism: Revisiting the old issue. Sociology of Religion 71(4): 393–408.

36.

Rine

Glanzer

Davignon

(2013) Assessing the denominational identity of American evangelical colleges and universities, Part II: Faculty perspectives and practices. Christian Higher Education 12(4): 243–265.

37.

Roof

(1989) Multiple religious switching: A research note. Journal for the Scientific Study of Religion 28(4): 530–535.

38.

Ross

(2011) Generic Christian U. ChristianityToday.com . Available at: http://www.christianitytoday.com/ct/2011/january/22.14.html (accessed 24 July 2014).

39.

Sherkat

Ellison

(2007) Structuring the religion-environment connection: Identifying religious influences on environmental concern and activism. Journal for the Scientific Study of Religion 46(1): 71–85.

40.

Smith

Sikkink

(2003) Social predictors of retention in and switching from the religious faith of family of origin: Another look using religious tradition self-identification. Review of Religious Research 45(2): 188–206.

41.

Smith

(1998) American Evangelicalism: Embattled and thriving. Chicago: University of Chicago Press.

42.

Stark

(1996) Religion as context: Hellfire and delinquency one more time. Sociology of Religion 57(2): 163–173.

43.

Stark

Finke

(2000) Acts of faith: Explaining the human side of religion. Berkeley: University of California Press.

44.

Steensland

Robinson

Bradford Wilcox

Park

Regnerus

Woodberry

(2000) The measure of American religion: Toward improving the state of the art. Social Forces 79(1): 291–318.

45.

Stroope

(2011) Education and religion: Individual, congregational, and cross-level interaction effects on biblical literalism. Social Science Research 40(6): 1478–1493.

46.

Taylor

(2007) A secular age. Cambridge, MA: Harvard University Press.

47.

Trinitapoli

(2007) ‘I know this isn’t PC but…’: Religious exclusivism among US adolescents. The Sociological Quarterly 48(3): 451–483.

48.

Uecker

Regnerus

Vaaler

(2007) Losing my religion: The social sources of religious decline in early adulthood. Social Forces 85(4): 1667–1692.

49.

Wang

Xie

Fisher

(2011) Multilevel models: Applications using SAS®. Boston, MA: Walter de Gruyter.

50.

Warner

(1993) Work in progress toward a new paradigm for the sociological study of religion in the United States. American Journal of Sociology 98(5): 1044–1093.

51.

Wuthnow

(1988) The restructuring of American religion: Society and faith since World War II. Princeton, NJ: Princeton University Press.

52.

Wuthnow

(2007) America and the challenges of religious diversity. Princeton, NJ: Princeton University Press.

Religious belief in Christian higher education: Is religious and political diversity relativizing?

Abstract

Keywords

Introduction

Religiosity at college

Institutional context

The current study

Data and methods

Dependent variables

Independent variables

Analytical method

Results

Discussion

Footnotes

Appendix

Acknowledgements

Funding

Notes

Author biographies

References