INDIVIDUAL RELIGIOSITY AND DEMOCRATIC CITIZENSHIP

Stable democracies rely on citizens sharing a set of broadly defined democratic norms and participatory commitments incorporating both rights and duties (Janoski Reference Janoski1998; Lister Reference Lister2003; Dalton Reference Dalton2009). The right and duty to participate in decision-making in social, economic, cultural, and political life are central components of democratic citizenship, grounded in a commitment to the ideals of citizens’ responsibilities to others in a democratic community (Marshall Reference Marshall1950; Meade Reference Meade1986; Lister Reference Lister2003; Conover et al. Reference Conover, Searing and Crewe2004; Bolzendahl and Coffé Reference Bolzendahl and Coffé2009). Citizenship norms reflecting these values are defined as a “…shared set of expectations about the citizen's role in politics” (Dalton Reference Dalton2008, 78). These norms encompass a broad set of commitments to citizenship basics such as being law-abiding, politically engaged, socially aware, and taxpaying voters, and beyond these basics to further aspects of citizenship such as politically-informed consumption, helping others, and participating in civic associations (Stolle et al. Reference Stolle, Hooghe and Micheletti2005; Denters et al. Reference Denters, Gabriel, Torcal, Deth, Montero and Westholm2007; Dalton Reference Dalton2009).

Democratic citizenship is also a function of what citizens actually do, including public participation in a variety of political venues (Verba et al. Reference Verba, Nie and Kim1978; Verba Reference Verba1996). When scholars talk about democratic participation they often focus on formal political outcomes such as voting or party membership, but a broad array of informal and non-electoral political and civic activities are key components of strong democratic citizenship (Almond and Verba Reference Almond and Verba1989; Inglehart Reference Inglehart1997; Stolle et al. Reference Stolle, Hooghe and Micheletti2005; Burns Reference Burns2007; Dalton Reference Dalton2008; Coffé and Bolzendahl Reference Coffé and Bolzendahl2010; Bolzendahl and Coffé Reference Bolzendahl and Coffé2013).

Individual religiosity has been positively linked to certain aspects of citizenship including democratic values (Denters et al. Reference Denters, Gabriel, Torcal, Deth, Montero and Westholm2007; Bolzendahl and Coffé Reference Bolzendahl and Coffé2009; Ben-Nun Bloom and Arikan Reference Ben-Nun Bloom and Arikan2012; Vlas and Gherghina Reference Vlas and Gherghina2012; Smidt Reference Smidt2013), civic involvement (Ruiter and De Graaf Reference Ruiter and De Graaf2006; Putnam and Campbell Reference Putnam and Campbell2010; McAndrew and Voas Reference McAndrew and Voas2013; Smidt Reference Smidt2013), and political participation including voting, party membership, and protest (Norris Reference Norris2002; Driskell et al. Reference Driskell, Embry and Lyon2008; Norris and Inglehart Reference Norris and Inglehart2011; Smidt Reference Smidt2013; Arikan and Ben-Nun Bloom Reference Arikan and Ben-Nun Bloom2019). The mechanisms explaining this connection are multiple and overlapping. First, religious involvement may promote civic and community interests (Putnam Reference Putnam2001; Borgonovi Reference Borgonovi2008; Putnam and Campbell Reference Putnam and Campbell2010) by instilling norms about belonging and community (Fox Reference Fox2001; Voicu Reference Voicu2012; Smidt Reference Smidt2013). Second, according to social network explanations, those attending religious services have more opportunities to interact with other citizens, discuss politics, and engage in community activities (Ben-Nun Bloom and Arikan Reference Ben-Nun Bloom and Arikan2012). Third, religious elites can encourage their engaged communities to become involved democratic citizens (Verba et al. Reference Verba, Schlozman and Brady1995; Jelen and Wilcox Reference Jelen and Wilcox2002). Finally, religion may equip individuals with the practical knowledge and skills (human capital) they need to become engaged citizens (Smith Reference Smith1996; Sherkat and Ellison Reference Sherkat and Ellison1999; Wald et al. Reference Wald, Silverman and Fridy2005; Smidt Reference Smidt2013), as well as organizational resources to engage in political action (Morris Reference Morris1984; Brady et al. Reference Brady, Verba and Schlozman1995; Putnam Reference Putnam2004; Burns and Kniss Reference Burns, Kniss, Snow, Della Porta, Klandermans and McAdam2013).

This established and fairly considerable literature leads us to the following hypothesis:

CROSS-NATIONAL DIFFERENCES IN RELIGIOSITY AND DEMOCRATIC CITIZENSHIP

The relationship between individual-level religiosity and democratic citizenship is well documented, but does not address comparative aggregate variation in democratic citizenship. Research has shown drastic cross-national differences in average religiosity (Fox Reference Fox2006; Fox and Flores Reference Fox and Flores2009; Arikan and Ben-Nun Bloom Reference Arikan and Ben-Nun Bloom2019), frequently measured as average frequency of attendance (Ruiter and van Tubergen Reference Ruiter and van Tubergen2009). We know contextual-level religion frequently operates over and above the individuals who contribute to it, having a “sui generis” influence on the social and sexual values and behavior of both religious and secular people (Kelly and DeGraaf Reference Kelly and DeGraaf1997; Adamczyk and Pitt Reference Adamczyk and Pitt2009; Lim and MacGregor Reference Lim and MacGregor2012). Yet, we know much less about how aggregate religiosity relates to average national differences in citizenship norms and behavior.

Theoretically, aggregate religious attendance is a contextual factor that could operate indirectly, through various network and exposure effects (Regnerus Reference Regnerus2003; Smith Reference Smith2003), as well as more formal legal or cultural institutions (Finke and Adamczyk Reference Finke and Adamczyk2008). The studies that could guide our expectations about the potential relationship between aggregate societal religiosity and political norms and behavior are limited and contradictory. When considering the impact of aggregate religious attendance and volunteering, for example, some work finds a positive relationship (Ruiter and De Graaf Reference Ruiter and De Graaf2006; Reference Ruiter and De Graaf2010) while other studies do not (Van der Meer et al. Reference Van der Meer, Te Grotenhuis and Pelzer2010; Arikan and Ben-Nun Bloom Reference Arikan and Ben-Nun Bloom2019).Footnote ² Specifically, Lim and MacGregor (Reference Lim and MacGregor2012) find that religious localities in the United States have lower rates of volunteering.

Arguments for a positive relationship between societal religiosity and strong citizenship norms and behaviors are largely based on the concept of “moral communities” (Stark Reference Stark1996; Ruiter and De Graaf Reference Ruiter and De Graaf2006) in which higher average religious attendance indicates people's greater exposure to religious culture and the higher likelihood of having religious persons in one's social network (Kelly and DeGraaf Reference Kelly and DeGraaf1997; Ruiter and De Graaf Reference Ruiter and De Graaf2006). On the basis of this framework, a more religious context should, in theory, promote positive citizenship norms and civic participation. Lim and MacGregor (Reference Lim and MacGregor2012), however, argue that religiosity is too geographically or socially stratified and uneven to use aggregate religiosity as a proxy for religiosity in social networks, and their findings strongly challenge the moral communities’ framework. Lim and MacGregor provide several possible explanations for why contextual religiosity seems to have a negative impact on volunteering. For example, in more religious contexts there may be more free-riding, or that aggregate religiosity—if taken as a proxy for a national religious culture—could influence individual behaviors in ways that matter differently from, or regardless of, personal religiosity. Lacking a large literature with any consensus about how contextual religious culture impacts individual civic attitudes and behaviors, we draw on Lim and MacGregor (Reference Lim and MacGregor2012) to set forth the following hypothesis:

RELIGIOUS CONTEXTS AND DEMOCRATIC CITIZENSHIP

Earlier, we offered arguments regarding the impact of religiosity among individuals as well as between states. Both approaches neglect the importance of potential contextual variation in individual differences. Does the relationship between individual religiosity and democratic citizenship norms and behavior vary based on national religious context? Research shows that religious individuals are more committed to citizenship norms and political activities, but this relationship may vary based on whether the individual lives in nations where their average fellow citizens are more or less actively involved in religion. In the United States, the individual propensity to volunteer is depressed in more highly religious contexts, though religious context does not affect frequent attendees’ level of volunteering (Lim and MacGregor Reference Lim and MacGregor2012). Across 53 countries, Ruiter and De Graaf (Reference Ruiter and De Graaf2006) found that individual attendance promotes volunteering more in the most secular countries.

Although the specific mechanisms underlying such potential cross-level interaction are difficult to untangle (and doing so is beyond the scope of this paper), we offer a few related possibilities that inform our expectations. One possibility is that in contexts where everyone is religious, there may be general saturation of religiosity and it may be a less distinctive factor in people's civic commitment and engagement. Specifically, if everyone is religious, the choice to be religious could be less meaningful, and the religious teachings and values of religious institutions potentially less counter-cultural. Widespread religiosity may also dissuade interest in political matters (Rink Reference Rink2018), marginalize those who are more weakly tied to the church (Lim and MacGregor Reference Lim and MacGregor2012), and/or place more sociopolitical issues under clerical control (Buckley Reference Buckley2016).Footnote ³ Conversely, being religious in a more secular society may be more distinctive, require more from adherents, and/or promote a sense of “embattlement” and perceived need for activism on behalf countercultural values. Therefore, we set forth the following hypothesis:

DATA

Our data come from the 2014 International Social Survey Program (ISSP) module on Citizenship (ISSP Research Group 2016; Joye and Sapin Reference Joye and Sapin2017). The ISSP coordinates survey research across a variety of different countries to provide social scientists with a reliable, cross-national dataset. For our analysis, we focus on 28 historically-Christian nations (see discussion of individual religious attendance below).Footnote ⁴ We focus on cases with complete data for a final sample size of 36,024 individuals. Nations, sample sizes, and means/proportions for key variables are listed in Table 2. Descriptive statistics for all individual-level variables are listed in the Appendix (Table A1).

MEASURES

Independent Variables: Country Level

We measure country-level religiosity with average national religious attendance, which we consider to proxy the cultural pervasiveness of religious involvement. This measure is aggregated from the individual measure of attendance provided through the ISSP and recoded to range from 0 continuously to 1. Table 1 presents the actual variation in the untransformed variable on average and for each nation.

Table 1.

Means for key state-level variables for all 28 nations

Notes: Raw data presented here, in models below these variables are recoded to range from 0 to 1 and/or are standardized.

Source: ISSP 2014.

All models control for 2014 Gross Domestic Product (GDP) per capita, adjusted for purchasing parity power (PPP), as publicly available from the World Bank (htttp:// www.worldbank.org). A variable controlling for level of democracy was too highly correlated with GDP to be included in the analysis. We also use categories from the Inglehart-Welzel Culture Map, one of the most powerful methods of identifying the cultural schisms among global society (Inglehart and Welzel 2015). This map identifies two major dimensions of cross-cultural variation: traditional values versus secular-rational values and survival values versus self-expression values. Given that attitudes are highly correlated with the philosophical, political, and religious ideas historically dominant in particular regions, Inglehart and Welzel have been able to identify substantively meaningful cultural clusters (Welzel Reference Welzel2013). We examined these clusters one-by-one and in combination with one another, and found the Catholic European cluster is notably different from the other clusters with regard to citizenship norms, and the Protestant European cluster is significantly different with regard to political participation. We include dichotomous variables for these clusters.

Table 1 presents average attendance, citizenship norms, and political participation for each of the 28 countries considered in this study.

ANALYTIC STRATEGY

Hierarchal linear regression models were run in Stata with the “xtmixed” command, and verified in HLM7. Data used in individual-level estimates are weighted using a factor provided by the ISSP. We examined the data visually through scatterplots and tested for the presence of influential cases with diagnostic tests (e.g., DFBETAs and values for Cook's D) (Van der Meer et al. Reference Van der Meer, Te Grotenhuis and Pelzer2010). We ran all models by dropping states individually and by dropping groups of states that were potential outliers. These approaches did reveal important inconsistencies in the results for democratic norms but not political participation. Differences are discussed in more detail in the results below.Footnote ⁸ We allow the slope for individual-level religious service attendance to vary randomly in all models.

CROSS-NATIONAL PATTERNS: RELIGION, NORMS, AND PARTICIPATION

As seen in Table 1, levels of religious attendance, political norms, and political participation vary widely across nations. We begin by considering scatterplots of the country-level relationships between these measures.

Figure 1 shows the relationship between the standardized measure of citizenship norms and average attendance (ranging from 0 to 1). These variables correlate at r  =  .40, which increases to .50 when one visual outlier, Slovakia, is dropped (or, alternatively, to .44 when Spain is dropped). The scatterplot indicates a positive relationship, though with substantial variation, between attendance and citizenship norms across the nations in this sample. We will present further information on outlier testing below.

Figure 1.

Bivariate relationship between average religious service attendance (y-axis) and average citizenship norms.

We found a strong positive correlation between attendance and citizenship norms, but when turning to citizenship behavior we see a strong negative relationship. The overall correlation is r = −.54, and this increases to −.78 when two visual outliers, Russia and Hungary, are not included in the calculations. Descriptively, then, we see some evidence that more religious cultures (as measured by average attendance) may bolster the expression of normative commitments, but undermine actual participation. However, we still need to consider multilevel and multivariate models to further explore these relationships. We next turn to models that allow us to test our hypotheses about the potential impact of country-level religion independent of, and in interaction with, individual-level religion.

MODELS OF RELIGION, NORMS, AND PARTICIPATION

Individual-Level Relationships

Table 2 shows the relationship between individual religious service attendance and both citizenship norms and political participation. Both coefficients are significant and positive, indicating (as expected) strong support for Hypothesis 1: individuals who attend religious services more frequently are more committed to citizenship norms and more involved in political activities. Findings for controls are similar to those in prior work studying norms and participation (Kam and Palmer Reference Kam and Palmer2008; Bolzendahl and Coffé Reference Bolzendahl and Coffé2009; Coffé and Bolzendahl Reference Coffé and Bolzendahl2010; Marien et al. Reference Marien, Hooghe and Quintelier2010). Specifically, women are more normatively committed to democratic citizenship than their lower participation would indicate. Those who are more politically interested and engaged hold stronger norms and participate more. Finally, participation is strongly cleaved by socio-economic divides such that those who have more resources in this area participate significantly more.Footnote ⁹

Table 2.

Results for individual-level hierarchal linear regression models of religious attendance on citizenship norms and political participation, N  =  36,024

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

Understanding Aggregate Norms and Participation Cross-Nationally

Figures 1 and 2 raise a number of questions about the appropriate sample to focus on in a second-level hierarchal analysis. A combination of visual assessments, dropping nations systematically, and examining clusters of influential cases are key for ensuring robust results (Ruiter and De Graaf Reference Ruiter and De Graaf2010; Van der Meer et al. Reference Van der Meer, Te Grotenhuis and Pelzer2010). Multilevel models for both norms and participation were checked across all these factors. Assessing and adjusting for outliers did not change the overall patterns for political participation, though dropping some nations strengthen the relationship described below. With regard to norms, the aggregate positive relationship is no longer significant when controlling for or excluding the United States. Dropping other visual outliers (e.g., Slovakia) does not change the results. We present results for final models in Table 3.

Figure 2.

Bivariate relationship between average religious service attendance (y-axis) and average political participation.

Table 3.

Results from hierarchal linear regression models of average national religious attendance on citizenship norms and political participation, N  =  36,024

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

Although our descriptive plots suggested a strong positive relationship between aggregate attendance and citizenship norms, this relationship does not persist in models that control for key country-level differences. The negative residuals in Figure 1 belong to many historically-Catholic nations (e.g., Poland and Slovakia). Here we see that this group of 10 nations has significantly weaker citizenship norms. Yet, as we see in model 1 of Table 3, there is still a positive relationship with aggregate attendance. As further testing shows in model 2, however, the result was not stable and simply including a control for the United States—which seems to have an outsized impact on the pattern—makes the aggregate relationship between attendance and citizenship norms become insignificant.Footnote ¹⁰ In this case, we fail to reject the null in our second hypothesis and conclude there is no clear evidence that more religious nations have weaker commitments to citizenship norms.

In models 3 and 4, we examine political participation and find a negative relationship between average religious service attendance in a nation and that nation's political participation. Robustness checks did not indicate any important outliers and the main coefficient remained significant when dropping countries individually. Russia and Hungary were visual outliers (Figure 2), and when dropping them the relationship becomes even stronger. There is a positive relationship between Protestant European nations and participation, also reflecting patterns seen in Figure 2. These findings support Hypothesis 2, and suggest that more religious countries are more likely to eschew informal political participation. One clue for understanding this finding may be found in examining cross-level interactions, which we turn to next.

Cross-Level Interactions: Norms and Participation in Religious Context

The results presented in Table 4 consider whether the relationship between individual attendance and democratic norms and participation vary across religious contexts (in this case measured as the average level of religious service attendance in a country). Although a strong positive relationship was found (Table 2) between individual attendance frequency and citizenship norms, country-level tests (Table 3) suggested it could not be extended toward explaining variation across nations in the average importance assigned to these norms. The results in Table 4 indicate that the positive relationship between more frequent religious service attendance and norms is weaker in more religious contexts. Model 1 includes no country controls, model 2 controls for the United States, and model 3 controls for two additional outliers (Philippines and Croatia). All three nations have higher than average commitments to citizenship norms, thus controlling for all of them presents a conservative test of the cross-level interaction. The negative interaction remains significant.

Table 4.

Results from hierarchal linear regression models of the interaction between individual and national religious attendance on citizenship norms and political participation, N  =  36,024

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

In models 4 and 5, we examine the same relationship for political participation and also find that a more religious context (higher average levels of attendance) depresses the positive relationship between individual attendance and political participation. The negative main coefficient of country-level religiousness on political participation is strengthened slightly when controlling for Hungary and Russia, two nations with substantially lower than average levels of participation. Overall, we see strong support for Hypothesis 3: a more religious context appears to weaken the positive relationship between attendance and democratic values and behavior.

To more clearly illustrate these patterns, we modeled the marginal predicted values for citizenship norms and political participation at key levels of individual and national attendance in Figures 3 and 4. Figure 3 is comprised of panels A and B, which are based on models 2 and 3 in Table 4, respectively. Focusing on the most conservative model in panel B, we see that frequent attendees hold strong commitments to citizenship norms regardless of context, but as the context become more religious (higher average attendance) those who attend less frequently have much stronger normative values.

Figure 3.

Predicted marginal values for citizenship norms across levels of individual and national average religious services attendance.

Figure 4.

Predicted marginal values for political participation across levels of individual and national average religious services attendance.

In Figure 4, which illustrates model 4Footnote ¹¹ in Table 4, the pattern for political participation is different because religious attendance operates differently at the individual and country levels. As Table 3 shows, religious individuals tend to participate more, and countries with higher levels of attendance tend to have less participation overall. The negative interaction thus indicates that, although frequent attendees participate more, this relationship becomes significantly weaker as the context become more religious. Those who never attend church participate less, but their participation is not shaped as strongly by religious context. Overall, then, we find that although there are large differences in citizenship norms and behaviors by attendance in more secular countries, the attendance gap shrinks with higher average attendance. In fact, the predicted probabilities in Figures 3 and 4 indicate that citizenship norms converge at similarly high, and citizenship behaviors at similarly low, levels in the most religious countries.