Economic Development Leads to Stronger Support among Religious Individuals for Clerical Influence in Politics

Abstract

The secularization paradigm suggests that economic development should lead to the decline of clerical influence in politics. However, this suggestion is challenged by the fact that there is widespread support for clerics who try to transform politics according to religious precepts in many developed societies. In order to address this question, the present study examines how national economic development moderates the relationship between individual-level religiosity and attitudes towards clerical influence in politics, using the fourth and fifth waves of the World Values Survey data from 54 national samples. Multilevel regression models show that, first, religious individuals are more likely to support clerics’ political influence than nonreligious individuals. Moreover, the effects of individual religiosity in this regard depend on the economic context. In less developed societies, there is no significant difference between religious and nonreligious individuals regarding their attitudes toward clerical influence in politics. However, religious individuals become more supportive of clerical influence in politics in more developed countries.

1. Introduction

The secularization paradigm suggests that economic development should lead to the decline of clerical influence in politics. Early secularization theorists have argued that religion will withdraw from public affairs and retreat into the private sphere in modern societies (e.g., Luckmann 1967). Casanova (1994) has challenged the privatization thesis by analyzing several religion-related political movements, such as Iran Islamic Revolution and Christian Right in the U.S. However, it is unclear whether economic development may influence the degree of religion privatization. Recently, Norris and Inglehart (2011) have proposed a revised secularization theory, arguing that economic development improves people’s existential security and thus undermines religion’s influences over individuals’ attitudes and actions. Norris and Inglehart’s research suggests that the public may shift to a secular attitude towards politics due to economic development. The debate around the privatization thesis raises essential questions: Does economic development reduce the mass support for the clerics who attempt to influence politics? Alternatively, does such support receive motivation during the process of economic change?

In order to address these questions, the present study examines how national economic development moderates the relationship between individual-level religiosity and attitudes towards clerical influence in politics, using the fourth and fifth waves of the World Values Survey (WVS, hereafter) data from 54 national samples. Previous research has touched upon but not sufficiently analyzed this issue. Case studies about religion-related political movements have revealed how these movements attract followers by analyzing their strategies, discourses, or organizational characteristics (e.g., Casanova 1994; Nasr 2001; Wickham 2011; Wolf 2017; Woodberry and Smith 1998). However, this line of research falls short of explaining the cross-national variation in people’s attitudes toward religious politics. In addition, a small group of studies has emphasized the effects of national contexts, pointing out that people in some societies are more supportive of religious politics than other societies (Buckley 2015; Carlson and Listhaug 2006; Karakoç and Başkan 2012). These studies focus on economic inequality and aggregate national religiosity while not directly testing the moderate potential of economic development.

In the following sections, I elaborate on prior theories and research and set out a series of empirical expectations. I test the hypotheses with large-scale, cross-national data. Finally, I discuss the findings in light of theories of secularization, the religion-politics relationship, and religion-related political movements.

2. Literature Review

The secularization paradigm has three different but related connotations: the decline of religious beliefs and practices, privatization of religion, and differentiation of the secular spheres (Casanova 2006). Scholars have yet to find concrete evidence to argue that religion will disappear in modern societies (e.g., Stark and Finke 2000). Regarding the privatization thesis, some secularization theorists have argued that religion loses its traditional public functions and only plays a marginalized and privatized role in modern societies (e.g., Luckmann 1967). Casanova (1994) has opposed the privatization thesis by showing that religion-motivated political movements can shape political landscapes in multiple countries. These political movements, such as Iran Islamic Revolution and Christian Right in the U.S., were led by religious leaders and won support from a large segment of society. However, although Casanova has provided case studies to illustrate the limitations of the privatization thesis, his research has yet to examine how broader societal forces, such as economic development, may influence the role of religion in public life.

Norris and Inglehart (2011) have recently provided an influential theory, which explains the decline of religious beliefs and practice from the perspective of economic development. They have argued that religion is essential in traditional societies because it satisfies the psychological need for the security of people threatened by various uncontrollable risks, ranging from floods and disease epidemics to human rights violations and poverty. The economic progress in modern societies has often lifted people out of poverty and reduced other daily risks to survival, thus improving people’s sense of security and diminishing the social importance of religion. Therefore, people in developed societies are increasingly indifferent to religion. Using the WVS data, Norris and Inglehart’s research has examined the effects of economic development on individual religiosity indicators such as religious service attendance, prayer frequency, and the importance of religious values. They have found that people living in more developed societies are less religious. However, their research left a question unanswered, do religious people become less supportive of cleric influence in politics in developed countries?

According to Norris and Inglehart’s theory, one may expect that religious people are less likely to support clerical role in politics in more developed countries. Individual religiosity often influences the attitudes toward religion’s role in the political arena (e.g., Esposito 1998; Hout and Fischer 2002; Inglehart and Baker 2000; Tessler 2015; Wilkins-Laflamme 2016; Woodberry and Smith 1998). Generally, these studies reason that high-religiosity individuals regularly interact with clerics and thus are exposed to clerics’ political teachings. Furthermore, highly religious people attach considerable importance to religious values and may support clerics to spread religious values by political means. By contrast, people weakly committed to religion often withdraw from religious activities and have a secular way of life, leading to their disapproval of clerics’ political influences. For example, studies on current religious dynamics in the U.S. have noted such attitudinal differences between high- and low-religiosity individuals. The Christian Right, which attempts to influence politics with a conservative agenda, mobilizes followers mainly from devout religious adherents. However, people weakly attached to religion often object to this religion-based political movement (Hout and Fischer 2002; Woodberry and Smith 1998). Similarly, past works on Islamism have portrayed a political division between individuals who hold Islamic values and embrace a secular lifestyle. According to the literature, Islamist clerics and movements are particularly attractive to pious Muslims, while secular individuals favor the separation between Islam and politics (Tessler 2015; Wolf 2017).

Economic development may moderate the relationship between individual religiosity and the attitude toward cleric influence in politics. As pointed out by secularization scholars, a fundamental principle of the modern political system is the separation of religious and political authorities (Berger 2014). In this situation, religion’s influences are limited within the private sphere, and it withdraws from public life. Most developed countries have established the separation of religious and political authorities. Such secular institutional contexts may shape individuals’ values and behavior. Although they are highly committed to religious faith, religious individuals may accept the modern political reality that religious leaders should not interfere in politics. Therefore, one may expect that religious and nonreligious individuals in developed societies do not have significant differences in their attitudes toward cleric influence in politics. According to this line of reasoning, I hypothesize as follows:

Hypothesis 1 (H1).

The positive effects of individual religiosity on the support for cleric influence in politics are weaker in more developed countries.

However, another perspective predicts a contrasting pattern that economic development may strengthen the effects of individual religiosity. Economic development often changes social norms, with a growing number of people accepting secular or liberal values on many social issues, from gender hierarchy to religious education, from political legitimacy to family relations (Inglehart and Baker 2000). This liberal trend contradicts the traditional value orientation that many religious adherents tend to hold (Liefbroer and Rijken 2019). Religion may provide certainty and peace during drastic societal change, and religion’s traditional value orientation may remain attractive to some people (Wickham 2011; Yang 2005). Due to the correlation between individual religiosity and value orientation, many developed societies are culturally divided between the religious-conservative and the secular-liberal groups (Halman and van Ingen 2015; Wilkins-Laflamme 2016). This cultural division may further generate antagonism and conflicts as each side views the other as a moral threat to its lifestyles and worldviews (Jacoby 2014).

Moreover, economic development may empower religious elites to mobilize laypeople. In more prosperous societies, religious elites can accumulate more resources and invest more in religious networks and activities. Religious elites may have stronger mobilization capacity in more developed societies. Therefore, cultural and value conflicts, plus religious elites’ strong mobilization capacity, may lead to greater support for clerical influence in politics. Many religious elites are trained and experienced in organizing group action and constructing collective identity and can inspire followers’ commitment by referring to divine will or eternal truth (Smith 1996). Due to the diversity in religious traditions and beliefs, not all religious elites are dedicated to political activities. However, scholars have noted that many religious elites are active in the political arena in different societies (e.g., Gorski 2017; Wickham 2011). Laypeople may view these religious elites’ political involvement as helpful in upholding traditional social norms. For example, the Christian Right movement has attracted many religious individuals who disagree with the legitimation of abortion and same-sex marriage in the U.S. By contrast, the social norms in developing societies largely remain traditional and conform to most religions’ social teachings (Inglehart and Baker 2000). Religious individuals in developing societies may not perceive their way of life as threatened as their developed world counterparts. They may be less motivated to support clerical influences in politics. In conclusion, I expect that:

Hypothesis 2 (H2).

The positive effects of individual religiosity on the support for cleric influence in politics are stronger in more developed countries.

Although economic development is a crucial driver of cultural and value changes, developing societies are diverse in socio-religious landscapes. One relevant factor is government regulation of religion. In societies with heavy religious regulations, religious people are deterred from religious activities (Fox and Tabory 2008). Developing societies with different degrees of religious regulation may be diverse in socio-religious values. Moreover, communist heritage influences religious beliefs and practices (Minarik 2014). Post-communist societies have shown distinctive trajectories of cultural and value change. Therefore, political and historical contexts may lead to heterogeneity in developing societies’ socio-religious landscapes.

3. Data and Method

The primary data source of the present study is the fourth (1999–2002) and fifth (2005–2006) waves of the World Values Surveys (Inglehart et al. 2014). Only the two waves contained the questions related to the attitudes toward religious politics. The analytic sample combined the fourth and fifth waves. For those countries surveyed in both waves, I kept their fifth-wave samples. Therefore, the analytic sample includes 19 countries surveyed in the fourth and 35 in the fifth waves. The sample size for country-wave units ranged from 838 (New Zealand, Wave 5) to 2851 (South Africa, Wave 5). The analytic sample represents all habitable continents and contains highly diverse countries in economic development, political institutions, and religious traditions.

Table 1. Summary Statistics for all Study Variables.

Variables	Mean/Percent	Standard Deviation	Min	Max
Dependent variable
Support for clerics’ influence in politics	0.035	1.000	−1.273	2.802
Key Independent variables
Individual religiosity	0.035	0.990	−2.110	1.185
Logged GDP per capita	8.525	1.514	5.371	11.371
Individual level controls
Age	40.409	16.154	15	97
Women	0.504	0.500	0	1
College education	0.145	0.352	0	1
Income	4.676	2.292	1	10
Religious affiliation
Protestant	0.184	0.388	0	1
Catholic	0.241	0.428	0	1
Orthodox	0.115	0.319	0	1
Jew	0.004	0.060	0	1
Muslim	0.213	0.410	0	1
Buddhist	0.046	0.209	0	1
Hindu	0.025	0.156	0	1
Other religions	0.050	0.217	0	1
Religious none (reference)	0.122	0.327	0	1
Country level controls
Polity index	5.958	5.289	−7	10
Aggregate religiosity	0.032	0.624	−1.192	1.034
Religious fractionalization	0.438	0.230	0	0.757
Official religion	0.159	0.366	0	1
Gini index	0.469	0.768	0.238	0.683
Wave 5 (Wave 4 as reference)	0.863	0.343	0	1

presents descriptive statistics for all study variables, with detailed information for each country-wave case summarized in

Table A1. Country Level Variables for Each Country or Region.

Country	Mean Support for Clerical Influence in Politics	Log GDP Capita	Polity Index	Aggregate Religiosity	Religious Fractionalization	Having Official Religion	Gini Index	Wave
Albania	−0.249488	7.515152	5	−0.4316816	0.6034439	0	0.476	4
Algeria	0.6881459	8.135804	−3	0.3860891	0	1	0.392	4
Australia	−0.1170668	10.77743	10	−0.9485121	0.7053581	0	0.48	5
Bangladesh	0.0126096	6.205823	6	0.5979469	0.1509351	1	0.385	4
Brazil	0.2452431	9.142708	8	0.2500831	0.5659658	0	0.587	5
Bulgaria	−0.4126491	8.555276	9	−0.6606858	0.4097842	0	0.365	5
Canada	−0.1483017	10.66329	10	−0.4212575	0.712308	0	0.467	5
Chile	0.0982304	9.280486	9	−0.2984607	0.5530643	0	0.519	5
Taiwan region	−0.3725775	9.713053	10	−0.8442396	0.6496228	0	0.321	5
Cyprus	−0.2993729	10.29441	10	−0.173938	0.5450889	0	0.477	5
Ethiopia	−0.1212688	5.371356	1	0.7216944	0.5202475	0	0.357	5
Finland	−0.0738791	10.69836	10	−0.7614214	0.2882941	1	0.471	5
Georgia	0.1042932	7.697381	7	0.2643952	0.1225969	0	0.481	5
Germany	−0.0305474	10.55933	10	−0.5595562	0.660979	0	0.503	5
Ghana	0.1402342	6.94943	8	0.8515619	0.6065158	0	0.455	5
Guatemala	0.1875306	7.877909	8	0.7856147	0.5772846	0	0.5	5
Hungary	−0.2287762	9.439443	10	−0.7394633	0.6280375	0	0.513	5
India	−0.0232628	6.886822	9	0.2071664	0.3791364	0	0.477	5
Indonesia	0.0453334	7.791687	8	0.738721	0.1410488	0	0.422	5
Iran	0.2837504	8.634744	−6	0.2995407	0.0282612	1	0.442	5
Italy	−0.08131	10.52479	10	−0.0347184	0.2193928	0	0.484	5
Japan	−0.2799127	10.68445	10	−1.037838	0.5123495	0	0.423	5
South Korea	0.1864124	9.823904	8	−0.4762917	0.7568887	0	0.324	5
Kyrgyzstan	0.0442265	6.421447	−3	−0.3507998	0.4165829	0	0.456	4
Malaysia	0.2281527	8.951677	3	0.8050887	0.6152129	1	0.46	5
Mali	−0.1025772	6.473231	7	0.7741945	0.0992789	0	0.436	5
Mexico	0.447214	9.125952	8	0.2304932	0.4319671	0	0.478	5
Moldova	0.1229348	7.556458	8	−0.2698564	0.1471106	0	0.565	5
Morocco	0.5463964	7.747246	−6	1.034188	0.0116319	1	0.446	5
New Zealand	−0.1924069	10.39306	10	−1.023207	0.6812041	0	0.464	5
Norway	−0.6790965	11.37089	10	−1.065837	0.5003922	1	0.455	5
Peru	0.3365832	8.189522	9	0.1235698	0.4505062	0	0.548	5
Philippi	−0.1788049	7.389787	8	0.6317183	0.4589064	0	0.476	4
Poland	−0.4140203	9.170604	10	0.3127929	0.1027375	0	0.519	5
Romania	−0.3540958	8.770152	9	0.1683259	0.2386505	0	0.423	5
Rwanda	0.1510123	6.074065	−3	0.7227616	0.6145089	0	0.533	5
Viet Nam	0.00792	6.86218	−7	−0.9883463	0.6142151	0	0.41	5
Slovenia	−0.137392	9.960676	10	−0.7220406	0.4802308	0	0.396	5
South Africa	0.1847639	8.775257	9	0.3792233	0.5876378	0	0.683	5
Zimbabwe	0.4284406	7.329317	−6	0.5938547	0.548572	0	0.525	4
Spain	−0.2096974	10.3237	10	−0.9483094	0.3132815	0	0.453	5
Sweden	−0.3225363	10.80094	10	−1.191893	0.4675993	0	0.468	5
Thailand	0.1050583	8.340572	9	0.4598396	0.0585343	0	0.471	5
Trinidad	0.2179921	9.499336	10	0.4009708	0.6997756	0	0.441	5
Turkey	−0.0785363	9.106031	7	0.1369063	0.0206156	0	0.451	5
Uganda	0.1789983	6.259138	−4	0.6706904	0.6423087	0	0.455	4
Ukraine	0.0538353	7.885588	6	−0.5758852	0.5165529	0	0.238	5
Tanzania	−0.1354793	6.214129	−1	0.7967934	0.7164364	0	0.4	4
United States	0.4584492	10.764	10	−0.1355668	0.7461435	0	0.488	5
Burkina	−0.1241264	6.303274	0	0.7355806	0.6025797	0	0.482	5
Uruguay	0.0707868	9.041828	10	−0.9799876	0.577112	0	0.532	5
Venezuela	0.2546787	9.451435	8	0.104834	0.4882758	0	0.445	4
Serbia	0.1905078	8.430487	6	−0.3047742	0.217977	0	0.529	5
Zambia	0.4190331	6.982711	5	0.6077806	0.6510764	1	0.587	5

.

3.1. Dependent Variable

Two questions drawn from the WVS tap the support for clerical influence in politics. Respondents were asked, “How strongly do you agree or disagree with each of the following statements? (1) Religious leaders should not influence how people vote in elections. (2) Religious leaders should not influence government decisions.” Response options are “strongly agree”, “agree”, “neither agree not disagree”, “disagree”, “strongly disagree”. For the two questions, 15.6% and 18.6% of the respondents stated “disagree” or “strongly disagree,” respectively. Previous studies have shown that the two items are conceptually coherent and combined them in empirical analysis (Buckley 2015; Carlson and Listhaug 2006; Karakoç and Başkan 2012). Accordingly, I found the items were highly correlated (their principal-component factor loadings are 0.8726). Keeping in line with these studies, I combined the two questions to generate an index using the principal-component factor method. I recoded the variables so that higher scores mean more support for clerical influence in politics.

3.2. Key Independent Variables

Individual religiosity. I generated an index for individual religiosity by combining three religious variables: religious service attendance (“Apart from weddings and funerals, about how often do you attend religious services these days?” Responses were reverse coded from [1] “never” to [7] “more than one a week”), religious importance in life (“For each of the following, indicate how important it is in your life: Religion” Reverse coded responses range from [1] “not at all important” to [4] “very important”), and the importance of God, Buddha, Allah, or other deities (a 10-point scale ranging from [1] “not at all important” to [10] “very important”). The three religious variables are highly correlated (their principal-component factor loadings are 0.7800, 0.8810, and 0.8605, respectively). Such an integrated index based on multiple religious variables should be more reliable than a single indicator in terms of capturing religiosity which manifests in different ways across religious traditions. Thus, using this index to measure religiosity fits with the cross-national scope of this study.

Economic development. I measured the economic development with the natural log of gross domestic product per capita (in 2010 U.S. dollar, GDP hereafter). The data was drawn from the World Bank. The economic measure lagged one year, so respondents’ attitudes toward religious politics were linked to the previous year’s economic level.

3.3. Control Variables

For individual-level controls, I included standard demographic variables, which are age (in year), sex (0 = men, 1 = woman), education (from [1] “no education” to [7] “university education and above”), and self-report income (from [1] “lowest level” to [10] “highest level”). Previous studies have noted that these demographic variables are related to people’s attitudes toward religion’s political role (e.g., Buckley 2015; Karakoç and Başkan 2012). Moreover, I controlled for religious affiliation (Protestant, Catholic, Orthodox Christian, Jew, Muslim, Buddhist, Hindu, the follower of other religions, and religious “none” as reference), because theological views of religion-politics relationship are different between religious traditions (Philpott 2007).

Country-wave level control variables include (1) democratic level, measured by the polity index. The Polity IV dataset provides information on the authority characteristics of most states in the world. The polity index accounts for the competitiveness of executive recruitment, the openness of executive recruitment, constraints on the chief executive, and the competitiveness of political participation. The index ranges from −10 (strongly autocratic) to 10 (strongly democratic) and lagged one year in this study. 2) Aggregate religiosity is the country-wave mean of the individual religiosity variable. Aggregate religiosity was found to be another contextual factor of public opinion on religious politics (Buckley 2015). (3) Religious fractionalization index is calculated based on the proportion of the religious traditions specified above following the Herfindahl formula. Some research has portrayed religious fractionalization as a source of political struggles (Brubaker 2015), suggesting that people in diverse societies may support the role of religion in politics. (4) Official religion (0 = no official religion, 1 = having official religion) is a dichotomous variable based on Religion and State Round 3 data. Religious leaders from official religions tend to be less responsive to laypeople, changing people’s views on religion’s role in politics (Stark and Finke 2000). (5) Gini index, provided by the Standardized World Income Inequality Database (SWIID; Solt 2019). Previous research has noted that people in countries with higher degrees of economic inequality are more likely to support religious politics (Karakoç and Başkan 2012). (6) A wave indicator, with the fourth wave (1999–2002) set as the reference.

3.4. Model

I employed multilevel (hierarchical) linear regression modeling techniques in this study. There are two levels in the data, as individual respondents were nested within different countries. The analysis consists of two parts. First, I examined the effects of individual religiosity on the attitudes toward clerical influence in politics. At the next stage, I examined how economic development conditioned the effects of individual religiosity. The multilevel equations, taking the cross-level interaction between economic measure and personal religiosity as an example, are shown below:

$$Y_{ij}={\beta }_{0j}+{\beta }_{1j}{X}_{1ij}+{\beta }_2^{\prime }{x}_{ij}$$

$${\beta }_{0j}={\gamma }_{00}+{\gamma }_1^{\prime }{w}_j+u_{0j}$$

$${\beta }_{1j}={\gamma }_{10}+{\gamma }_1^{\prime }{w}_j+u_{1j}$$

In the level-1 equation, the attitude variable is a function of personal religiosity (X_1ij) and a vector of additional individual-level covariates (x_ij). This equation specifies a random intercept and a random slope for personal religiosity that vary across country-wave samples. The level-2 equations show the random intercept and random slope as functions of a vector of country-wave-level characteristics (w_j). The level-2 equation for the random slope permits the effect of personal religiosity to vary according to the economic contexts.

4. Result

4.1. Main Analysis

In an unconditional model without any covariates (not shown), the intra-class correlation (ICC) of the support for clerical influence in politics is 0.073. It means that about 7.3% of the outcome variable’s variance is at the country level. It is thus appropriate to employ multilevel modeling techniques due to the substantial variation at the country level.

Table 2. Multilevel Linear Regression Models on Support of Clerics’ Influence in Politics, World Values Surveys 1999 and 2005 Waves (Individual N = 59,951, Country N = 54).

	Model 1	Model 2
Key independent variables
Individual religiosity	0.149 *** (0.021)	−0.253 * (0.108)
Logged GDP per capita	0.073 * (0.031)	0.071 * (0.031)
Individual religiosity * GDP		0.047 *** (0.012)
Individual level controls
Age	−0.001 ** (0.000)	−0.001 ** (0.000)
Women	0.008 (0.008)	0.008 (0.008)
College education	−0.058 *** (0.012)	−0.058 *** (0.012)
Income	−0.003 (0.002)	−0.003 (0.002)
Religious affiliation
Protestant	0.010 (0.018)	0.009 (0.018)
Catholic	0.007 (0.017)	0.006 (0.017)
Orthodox	0.029 (0.028)	0.027 (0.028)
Jew	−0.106 (0.068)	−0.107 (0.068)
Muslim	0.089 *** (0.024)	0.089 *** (0.024)
Buddhist	0.081 ** (0.031)	0.080 * (0.031)
Hindu	−0.007 (0.040)	−0.009 (0.040)
Other religions	0.007 (0.024)	0.004 (0.024)
Religious none (reference)
Country level controls
Polity index	−0.016 * (0.008)	−0.016 * (0.008)
Aggregate religiosity	0.053 (0.062)	0.053 (0.061)
Religious fractionalization	0.157 (0.143)	0.163 (0.142)
Official religion	0.067 (0.095)	0.069 (0.094)
Gini index	0.851 (0.435)	0.836 (0.432)
Wave 5 (Wave 4 as reference)	−0.037 (0.089)	−0.035 (0.088)
Intercept	−0.881 ** (0.290)	−0.857 ** (0.288)
Variance components
Residual	0.902 (0.005)	0.901 (0.005)
Intercept variance	0.044 (0.009)	0.043 (0.009)
Slope variance	0.022 (0.005)	0.017 (0.004)

Note: Unstandardized coefficients are presented with standard errors in parentheses. One-tailed tests. Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001.

presents the logistic regression findings. Model 1 is a random-slope model with the effects of individual religiosity being set to vary across country-wave units. Individual religiosity is positively associated with the support of clerical influence in politics (β = 0.149, SD = 0.021), adjusting for all individual and country controls. The finding suggests that more religious individuals are more likely to support clerics to influence politics. However, there are substantial cross-national variance in personal religiosity slope (slope variance = 0.022). If accounting for one standard deviation, individual religiosity’s coefficients vary between 0.001 (=0.149 − $\surd 0.022$) and 0.297 (=0.149 + $\surd 0.022$). This finding implies that although individual religiosity strongly increases support for clerics in some countries, its effect is weak in others.

We now move on to the cross-level interaction models. Model 2 estimated the cross-level interaction between logged GDP per capita and individual religiosity. The interaction coefficient is significant and positive (β = 0.047, SD = 0.012), meaning that the effects of individual religiosity are larger in more economically developed countries. Figure 1 plots the interplay with all control variables set at their respective means. It shows that as the GDP level increases, high-religiosity (one standard deviation above the mean) individuals’ support rises substantially. There is no significant difference between high-religiosity individuals and low-religiosity individuals (one standard deviation below the mean) when the GDP is at the lowest levels. This figure supports Hypothesis 2, which expects larger effects of individual religiosity in more economically developed countries.

4.2. Sensitivity Analyses

I conducted sensitivity analyses for the key findings. I re-estimated the models by using separate variables for individual religiosity. As presented in Model 1 to 3 (

Table A2. Robustness Check for Separate Individual Religiosity Variables.

Dependent Variable	Model 1	Model 2	Model 3
Key independent variables
Importance of God/Allah/Buddha…	−0.111 *** (0.025)
Religious service attendance		−0.099 * (0.040)
Importance of religion			−0.342 *** (0.084)
Logged GDP per capita	−0.085 * (0.038)	0.009 (0.046)	−0.089 (0.057)
Importance of God/Allah/Buddha …* GDP	0.016 *** (0.003)
Religious service attendance * GDP		0.019 *** (0.005)
Importance of religion * GDP			0.051 *** (0.009)
Individual level controls
Age	−0.000 (0.000)	−0.000 (0.000)	−0.001 * (0.000)
Women	0.019 * (0.008)	0.020 * (0.008)	0.016 * (0.008)
Education	−0.063 *** (0.012)	−0.065 *** (0.012)	−0.063 *** (0.012)
Income	−0.003 (0.002)	−0.006 *** (0.002)	−0.003 (0.002)
Religious affiliation
Protestant	0.112 *** (0.016)	0.062 *** (0.017)	0.100 *** (0.016)
Catholic	0.098 *** (0.016)	0.055 *** (0.017)	0.089 *** (0.016)
Orthodox	0.107 *** (0.027)	0.077 ** (0.028)	0.107 *** (0.027)
Jew	−0.005 (0.067)	−0.085 (0.066)	−0.029 (0.067)
Muslim	0.181 *** (0.023)	0.157 *** (0.024)	0.179 *** (0.023)
Buddhist	0.160 *** (0.030)	0.112 *** (0.030)	0.125 *** (0.031)
Hindu	0.082 * (0.039)	0.051 (0.039)	0.082 * (0.039)
Other religions	0.102 *** (0.023)	0.072 ** (0.024)	0.083 *** (0.023)
Religious none (reference)
Country level controls
Polity index	−0.012 (0.010)	−0.011 (0.012)	−0.012 (0.015)
Aggregate religiosity	0.151 * (0.072)	0.210 * (0.090)	0.221 * (0.110)
Religious fractionalization	0.123 (0.172)	0.110 (0.209)	0.210 (0.260)
Official religion	0.005 (0.112)	0.060 (0.139)	0.005 (0.174)
Gini index	−0.022 (0.505)	0.851 (0.633)	0.975 (0.772)
Wave 5 (Wave 4 as reference)	0.147 (0.110)	−0.015 (0.131)	0.147 (0.164)
Intercept	0.365 (0.342)	−0.680 (0.420)	−0.128 (0.523)
Variance components
Residual	0.913 (0.005)	0.908 (0.005)	0.911 (0.005)
Intercept variance	0.049 (0.013)	0.091 (0.019)	0.129 (0.030)
Slope variance	0.001 (0.000)	0.002 (0.000)	0.010 (0.002)
Individual N	61859	61043	61871

Note: Unstandardized coefficients are presented with standard errors in parentheses. One-tailed tests. Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001.

), all religious variables—religious service attendance, religious importance in life, and the importance of deity—significantly interacted with logged GDP per capita. The results with separate variables are remarkably similar to those with the integrated index of individual religiosity, suggesting that the findings are robust across different coding methods.

The limited sample size at the country-wave level, around 50 cases, makes the analysis vulnerable to the influence of potential outliers. To check the possibility that some country cases are too influential, I did a drop-one-at-a-time analysis in which I repeated the models with each country case being dropped at a time. If excluding any particular country case substantially alters the findings, it would mean that this influential outlier drives the findings. If not, it would provide evidence of the robustness of the results. Figure A1 exhibits the distributions of interaction coefficients and standard errors from the drop-one analysis. The x-axis indicates the point estimates for all country cases in the drop-one analysis, and the y-axis is the frequency of the estimates. The bold vertical lines refer to the full sample estimates of Model 2 reported in Table 2. The left graph (Figure A1) summarizes the drop-one analysis for the interaction coefficient of GDP and individual religiosity. The drop-one coefficients range between 0.04 to 0.06, which do not substantially depart from the full-sample estimate of 0.047 (Model 2 in Table 2). The right graph shows the distribution of the interaction standard errors. The drop-one standard errors range between 0.011 to 0.013, close to 0.012 in the full-sample model (Model 2 in Table 2). Generally, I did not find that dropping one country case would substantially change the interaction result between GDP per capita and individual religiosity. The above findings show that potential outlier cases do not drive the findings.

To examine the effects of religious affiliations, I estimate an additional model in which the Protestant group is set as the reference (Model 1,

Table A3. Robustness Check for Individual Religious Affiliation.

Dependent Variable	Model 1	Model 2	Model 3
Key independent variables
Individual religious affiliation
Protestant (reference)
Catholic	−0.001 (0.014)	−0.039 (0.036)	−0.005 (0.020)
Orthodox	−0.004 (0.026)	0.006 (0.063)	0.044 (0.053)
Jew	−0.122 (0.067)	0.046 (0.140)	−0.149 (0.112)
Muslim	0.071 *** (0.022)	0.081 (0.064)	0.044 (0.065)
Buddhist	0.062 * (0.031)	−0.019 (0.063)	0.023 (0.050)
Hindu	−0.009 (0.039)	−0.078 (0.069)	−0.036 (0.055)
Other religions	0.025 (0.023)	−0.026 (0.058)	−0.018 (0.042)
Religious none	−0.037 * (0.017)	−0.062 (0.049)	−0.021 (0.035)
Country Catholic size		−0.014 (0.128)
Country Muslim size			−0.263 (0.150)
Catholic * Country Catholic size		0.109 (0.085)
Orthodox * Country Catholic size		−0.154 (0.259)
Jew * Country Catholic size		−0.518 (0.416)
Muslim * Country Catholic size		−0.050 (0.239)
Buddhist * Country Catholic size		0.284 (0.309)
Hindu * Country Catholic size		0.314 (0.360)
Other religions * Country Catholic size		0.014 (0.162)
Religious none Country Catholic size		0.143 (0.119)
Catholic * Country Muslim size			0.094 (0.129)
Orthodox * Country Muslim size			−0.357 (0.220)
Jew * Country Muslim size			0.422 (0.336)
Muslim * Country Muslim size			0.146 (0.152)
Buddhist * Country Muslim size			0.043 (0.210)
Hindu * Country Muslim size			0.087 (0.180)
Other religions * Country Muslim size			0.012 (0.188)
Religious none * Country Muslim size			−0.020 (0.182)
Individual level controls
Individual religiosity	0.183 *** (0.006)	0.182 *** (0.006)	0.182 *** (0.006)
Age	−0.001 * (0.000)	−0.000 (0.000)	−0.000 (0.000)
Women	0.007 (0.008)	0.008 (0.008)	0.008 (0.008)
Education	−0.057 *** (0.012)	−0.055 *** (0.012)	−0.055 *** (0.012)
Income	−0.004 (0.002)	−0.003 (0.002)	−0.003 (0.002)
Country level controls
Log GDP per Capita	0.056 (0.030)	0.054 (0.031)	0.062 * (0.029)
Polity index	−0.020 ** (0.008)	−0.020 * (0.008)	−0.024 ** (0.008)
Aggregate religiosity	0.018 (0.060)	0.022 (0.060)	0.055 (0.059)
Religious fractionalization	0.122 (0.137)	0.276 (0.144)	0.168 (0.148)
Official religion	0.042 (0.091)	0.074 (0.094)	0.086 (0.091)
Gini index	0.665 (0.420)	0.465 (0.423)	0.360 (0.408)
Wave 5 (Wave 4 as reference)	0.021 (0.085)	0.054 (0.087)	0.043 (0.082)
Intercept	−0.685 * (0.279)	−0.684 * (0.284)	−0.590 * (0.283)
Variance components
Residual	0.913 (0.005)	0.907 (0.005)	0.907 (0.005)
Intercept variance	0.041 (0.008)	0.036 (0.008)	0.032 (0.007)
Catholic variance		0.003 (0.002)	0.003 (0.002)
Orthodox variance		0.023 (0.017)	0.014 (0.012)
Jew variance		0.052 (0.047)	0.055 (0.052)
Muslim variance		0.043 (0.016)	0.042 (0.016)
Buddhist variance		0.007 (0.012)	0.005 (0.010)
Hindu variance		0.000 (0.000)	0.000 (0.000)
Other religions variance		0.023 (0.015)	0.022 (0.015)
Religious none variance		0.025 (0.009)	0.028 (0.009)
Individual N	59,951	59,951	59,951

Note: Unstandardized coefficients are presented with standard errors in parentheses. One-tailed tests. Standard errors in parentheses. * p < 0.05, ** p < 0.01, *** p < 0.001.

). The attitudes of Catholics, Orthodoxy Christians, Jews, and Hindus are not significantly different from Protestants. However, Muslims and Buddhists are more likely to support clerical influence in politics relative to Protestants. This finding about Muslims aligns with the literature noting the Islamic tradition that emphasizes religion’s role in politics (e.g., Esposito 1998).

The effects of individual-level religious affiliations may depend on country-level religious group size. For example, Catholics may have different attitudes in Catholic-dominated countries and in countries where Catholics are a minority group. To explore this issue, I use the Catholic and Muslim groups as examples. I estimate a model to examine the interaction between individual-level religious affiliations and country-level Catholic group size (Model 2, Table A3). The interaction coefficient between individual Catholic affiliation and Catholic group size is not significant. Therefore, Catholics’ attitudes toward clerical influence in politics do not vary across countries with different Catholic group sizes. In addition, I explore if Muslims’ attitudes change across countries (Model 3, Table A3). According to the model result, individual Islamic affiliation does not significantly interact with country-level Muslim group size. Thus, country-level Muslim group size does not moderate Muslims’ attitudes toward clerical influence in politics.

5. Conclusions and Discussion

In this study, I have examined the cross-national variation in the relationship between individual religiosity and the support of clerical influence in politics. By combining the fourth and fifth waves of the WVS data, including more than 59,000 respondents from 54 countries or regions, I have found that how individual religiosity affects people’s attitudes toward clerical influence in politics is contingent on national economic contexts. Specifically speaking, high-religiosity individuals’ support for clerical influence in politics is higher in more economically developed societies. This finding suggests that economic development motivates religious people to support cleric influences in politics.

I propose explanations to understand national economic contexts’ moderating effects. Economic development is often associated with the growing influences of liberal and secular social norms and values. Highly religious individuals may choose to defend traditional values and norms through political means. For highly religious individuals, economic development may thus galvanize them to endorse the political role of clerics. However, religious people living in developing societies, usually dominated by traditional and conservative norms, may find fewer benefits from supporting clerics to influence politics.

The findings of this study have important implications for the literature. A considerable size of literature has emphasized the role of individual religiosity in shaping one’s opinion of religious politics (e.g., Esposito 1998; Hout and Fischer 2002; Inglehart and Baker 2000; Tessler 2015; Whitehead et al. 2018; Wilkins-Laflamme 2016). However, the present study shows that personal religiosity is not by itself a promoting factor for support of religious politics. The significance of personal religiosity in this regard is contingent upon the economic environment. According to the findings, only in highly developed countries do more religious people report stronger support for clerics’ political influence. This study’s findings align with a few past works highlighting the meaning of social contexts in understanding mass support for religious politics (Buckley 2015).

This study posits a challenge to the secularization paradigm that has long predicted that economic development undermines religion and religious leaders’ political significance. As a latest and refined version of the secularization paradigm, Norris and Inglehart’s theory (2011) argues that the improving human security during economic development has diminished religion’s importance as a major source of psychological assurance. This theory predicts the decline of popular support for religious politics in developed societies. However, the theory has failed to recognize a possible scenario: instead of withdrawing from the political arena, highly religious individuals may use political means to protect their conservative values. From a cross-national comparative perspective, this study’s findings echo case studies on religion-related movements in the developed world, such as the Christian Right in the United States (Woodberry and Smith 1998).

One direction for future research is to explore the diversity of socio-religious values in developing societies. As discussed earlier, developing societies are diverse regarding religious values and practices and of the religion-politics relationship. Such diversity may also influence religious people’s attitudes toward clerical influence in politics. Future research may further analyze the heterogeneity within developing countries. In addition, future research may explore the relationship between religion and populist movements from a cross-national perspective. Populist politicians may draw moral support from religion (Gorski 2017). It is unclear how national contexts may condition the relationship between populist movements and religion.

Overall, the present study expands on the literature on religion, politics, and societal contexts. The relationship between individual religiosity and the support for clerics’ political roles varies across countries. In societies with higher levels of economic development, religious people are more likely to support clerical influence in politics. This study contributes to the literature by examining how economic development influences attitudes about the privatization of religion. Although research shows that economic development is related to declining religious beliefs and practices (Norris and Inglehart 2011), economic development may provoke more extensive support for clerics’ political role among religious individuals. This finding suggests that economic development may have complicated consequences on the public role of religion in modern societies.