SOCIAL STATUS AND RELIGIOSITY IN CHRISTIAN EUROPE

The relationship between social status and religiosity is of major theoretical and empirical interest in the sociology of religion (and in studies of social stratification). One of the main tasks of the sociology of religion is to explore how socio-cultural factors influence religion and religiosity (without necessarily having the ambition to explain religiosity entirely away by social factors), and vice versa.

Review of the relevant literature reveals that many classical (sociologically minded) authors have written extensively about the matter. First, there are the French eighteenth-century Enlightenment materialists (e.g., Condorcet, d'Holbach, and Helvetius), who may now be almost forgotten but their basic premise was that knowledge, particularly scientific knowledge, will do away with all forms of ignorance, a category in which they included – if not religion itself – then the existing dominant forms of religion. For instance, d'Holbach wrote that ‘the ignorant’, not being able ‘to render themselves a satisfactory account’, are forced ‘to think the supernatural’ (1821: 88). More explicitly, he wrote that it is ‘ignorance’ that made ‘mankind form[ed] its first notions of Divinity’ (1821: 90). In this case, D'Holbach was not writing of French Catholicism only, a prime target of French Enlightenment criticism, but of religion in general, comprehended as an imperfect form of knowledge, as an imperfect comprehension of reality. This standpoint has often been ridiculed as simplistic and naive, particularly after Marx's introduction of the concept of alienation into social science and Weber's assertion of theodicy as central to the comprehension of religion. Nevertheless, the Enlightenment way of thinking led to a persistent assertion that, with respect to social status, education should be negatively linked to religiosity: the more educated people were, the less religious they would be. Such an argument rests on the belief that education diffuses knowledge, which, along with a systemic empiricist and methodically skeptical view of life, does not allow irrational creations of ‘Divinity’ and notions of the ‘Supernatural’.

Secondly, there are Marx's assumptions. Marx expressed varied views on this issue, probably the most famous being his statement about the ‘opiate nature’ of religion (Marx 1970 [1844]). This hypothesis links religiosity particularly to the poor and oppressed, since they need something to ease the pain of everyday life. Marx thus fathered the deprivation–compensation thesis, which asserts that ‘socioeconomic disparities cultivate the appeal of religion’ (Schieman and Pudrovska 2003: 4). This thesis was then further elaborated by Weber (1968 [1922]), Troeltsch (1960 [1912]), Stark and Bainbridge (1985), and others. In its essence, the idea hypothesizes that life is psychologically unbearable without religious consolation and explanation, particularly for the economically deprived and uneducated; since life is not comprehensible to the ignorant, lack of education fosters the production of religious representations of reality. The points meet in the idea that injustice and suffering in human life need both explanation and consolation.

At the other side of the spectrum are studies which imply that social status might be positively associated with religiosity (i.e., high social status leads to high religiosity) – a contention directly opposed to what was put forward by French Enlightenment materialists, classic theorists such as Weber and Marx, and authors whose empirical research found a negative link between social status and religious involvement (Pollner 1989; Coreno 2002; Schieman and Pudrovska 2003; Schieman et al.2003; Norris and Inglehart 2004). One of the more notable authors who implied such an association was Rodney Stark, who explained: ‘… wealth fails to satisfy many of those born into privilege and therefore they may turn to various religious … alternatives’ (Stark 2003: 7). In other words, Stark saw the upper class as economically well-off individuals who seek extra-economic rewards and who find these in the spiritual realm. Consequently, this is a rational choice for them under the given circumstances of economic opulence and privilege, where economic ends have lost their rewarding nature (conversely, all ‘ordinary’ subjects must toil for these worldly rewards, since their economic situation does not allow them to set such otherworldly goals). Nevertheless, it must be noted that Stark also wrote that the religiosity of the rich may partly involve ‘guilt about having wealth’ (2003: 16), that ‘the demonstrated (compensatory in nature) class effects on religious beliefs are very small within churches and quite unreliable in general populations, while those linking privilege and religious participation are strong and dependable’ (2004: 472). These points increased the complexity of what could otherwise be perceived as a rather simple theoretical framework. Consequently, some of Stark's contentions were more in line with the views that linked economic and social status, not (only) with the frequency and magnitude of religiosity, but also with its type and style – it may be more or less mystical, rational, ritualistic, emotional, strict, institution-linked or not, etc. (see, for example, Sherkat and Ellison 1999; Schieman and Pudrovska 2003; Schieman et al.2003). In other words, Stark's theoretical concept, together with other more contemporary views, went beyond attributing high religiosity to a particular social class in a simplistic way, acknowledging the complexity, not only of religious beliefs and practices, but also of class structures. Consequently, the idea that prevailed was that neither the upper nor the lower class entirely embraces or rejects religious beliefs and practices – i.e., that religion can and does take different forms and performs different functions across class strata (Schieman et al.2003). It thus seems that a range of diverse views about the linkage between religiosity and social status exists, although Smith and Faris (2005) recently noted that this relationship was still surprisingly understudied, especially in the matter of hypothesis testing by empirical research.

One line of empirical research was done within the frame of ‘religious stratification’, i.e., within an American frame, where religious groups cater for particular ethnic and status groups (Niebuhr 1929; Coreno 2002; Park and Reimer 2002; Smith and Faris 2005). For instance, Coreno (2002) found a difference between the social location of mainline and fundamentalist Christian religions in the United States, the latter being associated with lower status both culturally and from class perspective. In acknowledging that fundamentalist religion was ‘countering the penetration of secular values’ (Coreno 2002: 337) and that mainline religion was a diluted religion, which was ‘accommodated at the scientific-rationalist interpretations of the physical and social world’ (2002: 346), Coreno's findings gave important clues about where to look for non-religion.

Another strand of empirical research can be traced to the work of Schieman and Pudrovska (2003) who indicated that income and education, perceived as the ‘core components of socioeconomic status’ (Schieman et al. 2003: 202), played a significant role in one's sense of divine control, a concept known to correlate with measures of religiosity. Similar findings were obtained by Schieman et al. (2003), who established that a low sense of mastery correlated with religiosity, which in turn was negatively associated with education and income. Similar findings were reported by other studies also indicating that individuals in lower SES groups tended to report greater religious involvement, such as attendance, prayer, and various other forms of divine interaction (i.e., feeling close to God) (Pollner 1989; Ellison 1991; Sherkat and Ellison 1999; Norris and Inglehart 2004). Accordingly, Norris and Inglehart concluded that ‘religiousness is systematically related at individual-level to the distribution of income groups in postindustrial societies: the poor are almost twice as religious as the rich’ (2004: 108), mirroring the conten tion put forward by Sherkat and Ellison (1999), who stated that ‘higher levels of education have a negative impact on measures of traditional religious beliefs’ (Sherkat and Ellison 1999: 368).

It thus seems that a majority of recent studies give support to the hypotheses that education and income have a positive influence on individuals in the sense that they ‘attribute the events and conditions of their lives to their own actions’ and not to ‘forces external to themselves’ (Ross and Sastry 1999: 369) and that religiosity ‘helps individuals with fewer socioeconomic resources to accept the distressing reality of their powerlessness’ (Schieman et al.2003: 216). Consequently, one is tempted to ask why the available empirical evidence does not resonate within theoretical debates, which often warn against the conclusions mentioned above. We argue that more research, using more culturally and socio-economically diverse samples, is needed in order to cut through the maze of relationships between social status and religiosity.

Acknowledging the diversity of views within the debate about religiosity and social status, the main aim of the current study was to empirically test, using different national samples (known to differ dramatically in terms of religious salience and level of economic development) the following hypotheses:

Two dimensions of social status were tested independently in terms of their connection to religiosity and in terms of their robustness in predicting religiosity.

Among dimensions and measures of religiosity, we have limited ourselves to the notion of religiousness as it appears within Christianity, i.e. to the notion of a personal, omnipotent and ethical God. Clayton and Gladden, when analyzing the dimensions of religiousness, wrote that the ideological dimension (belief in the supernatural) was central to the study of religiousness in comparison to attendance, prayer and other components (Clayton and Gladden 1974: 141). A similar argument was put forward by Stark, who noted that ‘… belief in God is religiousness per se …’ (2001: 624), thus dismissing prayer and service, since the latter might possibly contain ‘many non-religious reasons’ (2001: 624). In other words, it was argued that ‘belief measures’ could be of greater validity than measures of religious practice and attendance, not only because religious attendance varies among religious groups and because some people attend religious services for social reasons, but also because the importance of holidays and prayer may differ significantly across cultures.¹

Among the dimensions and measures of social stratification (and social status as a unit of observation) which may also be studied via various dimensions and measures (occupation, education, political power, economic status, including current income, accumulated assets, etc.), the current study was limited to what can be seen as key components of social status – income and education. The most common aspect of one's social position is economic status (or class as conceptualised by Marx), with income being the most self-evident measure of economic status. Grusky (2004: 21) writes: ‘If one is truly intent on assessing the market situation of workers (Weber), there is much to recommend direct measurement of their income and wealth’. Similar importance could be ascribed to the level of formal education completed, which is connected to how and where one enters the labor market, with one beliefs and values, with one's ‘cultural capital’ (Bourdieu 1986). At the end of Grusky's Social Stratification (2004) Meyer concludes: ‘Every story told in these chapters amounts, in essence, to either emphasizing or explaining the rising stratificational importance of education and its credentials’ (2004: 88). In other words, education and income are ‘core elements of one's social and material conditions’ (Schieman et al.2003: 204), where the former is indicative of economic status and the latter of cultural status, which is often considered to be the more important one, especially if one sees contemporary society as a network society (Castells 1996) and as a knowledge society (Bell 1973).

The current study used data from the World Values Survey (1999–2001 waves) (Inglehart et al. 2004) in order to test the proposed hypotheses. More specifically, the data used covered 39 countries with n=48.263 respondents, where three main religious traditions prevailed: Eastern Orthodox, Lutheran and Roman Catholic. Countries selected ranked from being first on the Human Development Index rank (Norway, with a per capita GDP of 37,670 US dollars in 2003) to 115th (Moldova, with a per capita GDP of 1,510 US dollars in the same year) (Human Development Report 2005).

Religiosity was assessed by a single item: ‘How important is God in your life?’ (1= ‘not at all’, 10= ‘very important’).

Income was assessed by a single item on a ten-point scale which measured the relative position of the respondent within his/her country: ‘1’ indicated the lowest income, ‘10’ indicated the highest income. This measure was used for correlation and regression analysis. For the mean level comparisons, respondents were divided into three income groups of equal number within each country. The first was labeled LI (low income), the second MI (middle income), and the third HI (high income) respectively.

Education was also assessed by a single item, where the format of the question on completed education (which was standardised to fit varied educational systems) was 1 = ‘inadequately completed elementary education’, 8= ‘higher education degree’. This measure was used for correlation and regression analysis. Again, for the mean level comparisons, respondents were divided into three groups: the first comprised all those who indicated 1, 2 or 3 on the eight-point education scale (inadequately completed or completed elementary school), the second, all those who indicated 4, 5 or 6, and the third, all those who indicated 7 or 8 (some university without degree, university with degree or higher). These three groups were then named LE (lower education), ME (middle education) and HE (higher education), respectively.

Other measures included standard demographic variables used as controls in regression analyses and included gender, age, and size of town. Responses given were in the following format: gender 1= ‘male’, 2= ‘female’; age 1= ‘15–24’, 20= ‘25–34’, 3= ‘35–44’, 4= ‘45–54’, 5= ‘55–64’, 6= ‘65–98’; size of town 1= ‘2000 and less’, 2= ‘2000–5000’, 3= ‘5000–10,000’, 4= ‘10,000–20,000’, 5= ‘20,000–50,000’, 6= ‘50,000–100,000’, 7= ‘100,000–500,000’, 8= ‘500,000 and above’. Although possibly relevant, ethnicity was not included, as this question was not asked in the majority of the country samples included in the analysis.

First, zero-order correlations between the measure of religiosity and both status variables were computed (using full scales, i.e., ten-point for the income and eight-point for the education variable). Second, mean level comparisons were conducted (one-way ANOVA's with post-hoc tests) for different income (LI, MI, HI) and educational (LE, ME, HE) groups on the measure of religiosity. At this point, the possibility of a curvilinear relationship between status variables and the measure of religiosity was examined. Finally, to test the robustness of each status variable included, a series of hierarchical regression analyses were completed (again, full scales were used), where demographic variables such as age, gender and size of town were included as controls. At this final step, differences by confessional background were also investigated.

First, results from the zero-order correlation analyses indicated that income and the importance of God were significantly correlated in 28 out of 37 countries (P<0.05). All statistically significant relationships were negative, indicating that in 28 cases H1 could not be rejected (Table 1).

Second, results from the mean analysis indicated that significant mean differences (P<0.05) between income groups existed in 26 out of 37 samples. All of the 26 indicated that those in the high income group were on average less religious than those in the low income group; i.e., not a single country could be found where the richest segment of the population was significantly more religious. That said, there were cases where no significant mean difference could be found between the HI and MI groups (13 cases out of 26) or between the MI and LI groups (four cases out of 26). There were also cases where means for the MI groups were not significantly different from either the LI or the HI group, although a statistically significant mean difference could be observed between the HI and the LI groups (five cases out of 26). In the remaining four samples (out of 26), a statistically significant mean difference could be observed between all three income groups. Consequently, it could be argued that results from the correlation analyses and mean comparisons indicated a predominant situation where higher income meant, on average, lower religiosity. This pattern is also discernable from Figure 1, which also indicates that the relationship between the two observed variables is not curvilinear.

As indicated in the introduction, the next step was to examine the relationship between the selected measure of religiosity and another key component of social status, i.e., educational attainment (Table 2).

Results from zero-order correlation analysis indicated that, in 32 countries out of 39, the correlation between education and religiousness was significant (P<0.05) and negative. There was no national sample where the relationship was positive; i.e., in all of the remaining seven samples the relationship was non-significant (and still negative).

Similarly, results from the mean analyses indicated that a statistically significant (P<0.05) mean difference could be observed in 33 out of the 39 European countries with a Christian tradition. Again, all 33 significant differences indicated higher levels of religiosity among the less educated. There were four cases out of 33 where significant mean difference could be observed between all three groups, and 28 cases (out of 33) where significant difference could be observed between LE and HE (while in all of those no significant difference could be observed between the HE and ME groups, there were also two cases where no significant difference could be observed between the LE and the ME groups). In the one remaining case (Denmark), a significant mean difference could be observed between LE and ME, but not between LE and HE; i.e., the mean for the ME group was lower than the mean for the HE group (although the difference was not great enough to reach statistical significance). Results thus indicated an overwhelming predominance of a situation that could be predicted from the Enlightenment point of view – the higher the education attained, the lower the importance of God in a respondent's life. In addition, based on those findings and on findings that were obtained for the compound sample (see Figure 2), it could also be argued that the relationship was not curvilinear.

Consequently, it could be argued that using a different indicator of social status did not alter the overall picture – higher social status (speaking generally) was, on average, associated with lower levels of religiosity. The same could be argued when entire groups of countries were analyzed, based on their historical and actual predominance of certain confessions (Table 3).

Significant (P<0.001) and negative correlations were found between the importance of God, and both status variables within all confessional groups. This mirrored the results that were obtained from per country analysis; in general, those who were better off financially and those who were more educated on average reported a lower importance for God in their lives (regardless of their confession). However, it should be noted that the significant mean difference between all three income and education groups could be observed only in the Catholic group, and that, although significant, these differences were generally small in magnitude. One can only speculate on the factors explaining the lack of such differences in the other two confessional groups. The most evident explanation might be that all Orthodox countries are post-Communist transitional ones whereas in the other two confessional groups there are only isolated instances of such a situation, one bringing about general reorganisation and transitional disorganization. On the other hand, Lutheranism has undergone a possibly greater transformation itself, being more and specifically involved in catering for contemporary welfare problems, and thus better serving all strata of society (Fix and Fix 2005).

At this point, results indicated that neither H1 nor H2 could be easily rejected, i.e., that income and level of education could be seen as important correlates of religiousness. However, bivariate analyses fail to take into consideration other potential factors that may confound the relationship between social status and religiosity. For instance, variables such as age, gender, and residence type might reduce the effect of education and income on religiosity. In addition, one could also hypothesize that the effect of income would disappear if income was controlled for education (and vice versa). Thus, the employment of multivariate analyses is also needed when examining the relationship between social status and religiosity. Consequently, a series of hierarchical regression analyses using national, confessional and compound samples were conducted. For the sake of space, we present results for the compound and confessional samples only (Table 4).

Findings from regression analyses provided evidence that the effect of income on religiousness remained significant even after it was controlled by education (Model 1), regardless of whether compound or confessional samples were used. Similarly, income remained a significant predictor of religiosity even after age, gender (sex) and size of town were introduced into the analysis (Model 2). The same could be said for education: its effect remained significant when controlled for income, age, gender and size of town. Again, this was true regardless of the sample used (compound/confessional). Nevertheless, it should be noted that the effect of both status variables on religiosity was relatively weak and that, when combined, it explained less than two percent of the total variance. In addition, the effect of status variables was smaller than the effect of gender, which proved to be the strongest among the predictors included, and which alone explained around four percent of the variance.

The weak effect of the status variables has to do with the fact that results from analyses using national samples indicated that the effect of status variables on religiousness was so marginal as to be non-significant. Specifically, looking at the second model, in 11 out of 35 countries (in addition to Norway and Portugal, Armenia and Georgia were also excluded from this set of analyses since there was no data for the variable ‘size of town’), both status variables proved to be non-significant predictors. In six countries, both predictors proved to be significant. In 18 countries, one of the two status variables proved to be a significant predictor of the outcome. More specifically, income proved significant in seven and education in 11 countries. Still, there was not a single country sample where the opposite could be found (i.e., there was not a single country sample where higher income or level of education attained pointed to higher religiosity).

Finally, age and size of town also demonstrated a significant effect in compound and confessional samples, except in the Lutheran sample, where the size of town proved to be non-significant, and in the Eastern Orthodox country group, where age demonstrated a negative effect, indicating that younger cohorts were more religious.

The current study assessed the relationship between two social status variables (income and education) and religiosity (importance of God in one's life). Results from zero-order correlations indicated that, in the European Christian environment, religiosity tended to be negatively associated with both key components of social status. More specifically, whenever a statistically significant relationship was identified, this relationship was negative, i.e., there was not a single case where the relationship ran in the opposite direction (higher education/income, higher religiosity). The same pattern was replicated by mean comparisons and by multivariate analyses, although the latter indicated that the effect of both status variables on religiosity was relatively weak, especially after demographic controls (gender, age, size of town) were introduced in to the models. Still, a significant effect of income on religiosity was found in 13 countries, and a significant effect of education on religiosity was found in 18 countries (out of 35). In addition, no significant effect was found in the opposite direction where higher social status led to higher religiosity. Results also indicated that the nature of the relationship between status variables and religiosity, at least when assessed with the measures used in the current study, was not curvilinear.

In sum, results, despite the weakness of the association between social status and religiosity, and the low robustness of the former, gave limited support to the hypothesis that in environments where the social desirability of religion exerted little pressure (due to secularization), religiosity was relatively more concentrated among the lower strata. Results thus gave some credence to the classical positions that were outlined in the introductory part. More specifically, although it would be too much to insist that the negative association between income and the importance of God indicates an ‘escape from the deprivations of lower-class life’ (Lefever 1977: 225), findings indicated that higher levels of religiosity among lower classes in a European context could be seen as a combination of clinging to traditionalism, of lower levels of education among certain classes and their exclusion from the scientific-rational mode of thinking. In other words, the findings lent some credence to the contention that education (which in the context of modern society is inseparably associated with income) was conducive to what may be called a more enlightened worldview.² Consequently, our findings fit well with those of Coreno (2002), who interpreted the association between low status and pronounced forms of religiosity via differences in ‘class culture’. They also fit well with findings of Norris and Inglehart (2004), although the current study concentrated on one component of religiosity, which may, on the other hand, be more valid than measures that deal with ritual aspects (Stark 2001: 624).

One may thus conclude that social status variables are, at this level of analysis, weak predictors of religiosity. That said, the current study has limitations which should be mentioned at this point. First, we already mentioned that we did not deal with the issue of strength, dimensionality, and centrality of religiosity. We might have ended with notably different results if had used different measure(s), which in turn means that the current study was limited by its use of a single religiosity item. This is attributable to the fact that attempts to form and use more complex measures for assessing particular study constructs resulted in many missing cases, which then further limited our ability to generalize findings across different samples. Our suggestion for further research would thus be to do a similar study of church attendance, where some previous research has found curvilinear relationship that was not evident in the current study. Next, the current study also did not deal with the issue of the denomination effect inside national samples, which, judging by the results (see Table 4) might also play a role. We also did not analyze between-country differences, as this would go beyond the scope of the current paper. Nevertheless, it is obvious that individual historical and cultural environments also mold the relationship between social status and religiosity. Last but not least, our study was also exclusively cross-sectional and thus does not permit longitudinal inferences.

Sergej Flere is Professor and head of Department in Sociology, University of Maribor, Slovenia. He has wide experience in the empirical study of religion and ethnicity. His recent works include: Flere, S. and Klanjšek, R. (2008) ‘Serbian Orthodox Religiousness: An Empirical and Comparative Portrait’, Review of Religious Research 50(1): 35–48; Flere, S. and Lavrič, M. (2007) ‘Is intrinsic religious orientation a culturally specific American Protestant concept?: The fusion of intrinsic and extrinsic religious oriention among non-Protestants’, European Journal of Social Psychology 38: 521–30.

Rudi Klanjšek is an Assistant professor at the Department of Sociology, University of Maribor, Slovenia. His research interest includes religion, deviance and value change. He recently co-authored: Vazsonyi, A. T. and Klanjšek, R. (2008) ‘A test of Self-Control Theory across Different Socioeconomic Strata’, Justice Quarterly 25(1): 101–31; Flere, S., Klanjšek, R., Musil, B. &amp; Lavrič, M. (2008) ‘Testing the Allportian Religious Orientation Schema on Slovenian Catholics by Multidimensional Scaling’, Personality and Individual Differences 44(2): 512–20.