Introduction. It seems that in the last of couple decades the secularisation thesis has been exposed to the harsh criticism to such extent that it can even by labelled as an “unfashionable theory” (Bruce, 2011). Most famously, an early proponent of the theory, Peter Berger (1999), criticized it for it supposed ideological content prone to mixing secularisation (science) with secularism (a value-loaded political stance), and for its alleged empirical falseness. He even went that far to declare the secularisation to be a Western phenomenon and, more specifically, reserved it to the secular Western business and academic elites. Another strand of research criticized secularisation by putting forward an alternative theoretical framework know as supply-side approach or market-model of religion (Stark, Bainbridge, 1987; Iannaccone, 1992; Iannaccone, 1998; Stark, Iannaccone, 1994; Stark, Finke, 2000). In a milder manner, some authors called for a more nuanced approach to secularisation, thus effectively “sectorizing” secularization (Warner, 2010). For instance, Dobellaere (1981) divided secularisation into three arguably independent components: laïcisation (a decrease in social importance of religion), individual secularization (a decline in commitment to religion and religious organizations) and organizational secularization (an adaptation of religious organizations to secularised societies). In a similar manner, Casanova (1994) posits that secularisation can be divided into three independent empirical hypotheses: secularization as differentiation (independence of societal spheres), secularization as the decline of religion (decline of individual religiosity), and secularization as the privatization of religion (religion as solely a private matter). However, on this issue we concur with Steve Bruce (2011) in that we believe that the secularisation theory is unjustifiably accused as a pure ideological and political thesis with scarce empirical validity. As Bruce notes, the main cause of confusion is the alleged prognosis of the secularisation theory that religion is going to inevitably decline, and even that this is a good thing. However, secularisation theory does not necessarily include such far-fetched prognosis, it merely posits that modernisation is intertwined with secularisation, i.e. that social conditions that we label as „modernity“ make less likely the further social importance of religion and religiosity. Given the certain amount of vagueness that is attached to the notion of “modernity”, we can even define secularisation theory as a cluster of different ideas that theoretically connect modernity and decreased social importance of religion. It can be also argued that in the postmodern (or late-modern) fragmented and mediated social conditions religion can indeed develop differently in different religious fields, i.e. that secularisation can proceed in only some religious fields, thus effectively transforming religion in new directions and not destroying it (Pavić, 2016a, Pavić 2016b, Pavić, Kurbanović, Levak, 2017).

To sum up, such ideas and the propositions that follow from secularization can be independently tested. This is precisely the intention of this paper. Namely, we aim to explore whether religiosity is empirically related to the two processes that can be seen as the cornerstones of modernity: societalisation and economic development. First, we lay out a theoretical overview which draws on several authors that proposed such theoretical links, namely Bryan Wilson, as well as Pippa Norris and Ronald Inglehart. Furthermore, we also test the idea that the impact of these processes, when it comes to different levels of secularisation, can be very uneven. Namely, societalisation and economic wealth can cause religion to decline in one aspect (i.e., church attendance), and possibly be vital in other (e.g., religious beliefs). In order to avoid the difficult question of the possible uneven secularisation throughout the World and the European „exceptionalism“ (Davie, 2002), we test the theoretical propositions on the data collected within the European Values Study project, i.e. using the data collected in European countries.

Methodology and methods.Theoretical overview and research questions. In his book Religion and Secular Society (first published in 1966)[1], Bryan Wilson argues that social conditions of industrial societies are poised to lead to the decline of religion, i.e. to secularisation. For Wilson, the pursuit of individual goals by appealing to supernatural intervention is less plausible in a world that offers more realistic, goal-orientated, rational political possibilities and solutions. Individuals become more “this-worldy” orientated and less emotionally intensive and subject to emotional gratification previously provided by religious organizations. On the macro level, all social processes are becoming rationalized, with ready-made roles in rationalized social organizations to be filled in with the individuals who are fulfilling their goals by participating in collective actions. For instance, leisure and diversions provided by mass media offered a new outlet for spending free time, thus competing with a church ministers, previously the sources of life lessons (Wilson, 2016: 47-48). In the new industrial economy, the control over the entire persons of the worker is no longer needed, and thus religious and moral socialization ceases to be an imperative. The control over the work process in industrial societies is now relegated to the new principles of “scientific management”, which motivate worker through financial reward and direct control (Wilson, 2016: 53-54). This process leads to the “de-moralization”, i.e. everyday moral concerns become less and less important as a behavioral guide, since the social scripts are provided by large impersonal organizations. Previous personal social relations are being replaced with impersonal interactions guided by formal social roles which usually need no further reflection as to their morality. In consequence, the moral philosophy in industrial societies is becoming more concerned with morality of large-scale organizations (corporations, nation-states, etc.), and less concerned with the morality of individuals (Wilson, 2016: 45). Wilson named this process “societalisation” and defined as a decline of close-knit communities which comprised units of everyday life, work, faith and leisure, and were religion was the source of ideas about life that were transferred unreflectively through the collective life. The decline of such, mostly rural, communities deprived religion of its natural social setting which managed to provide for the effective religious socialisation for centuries.

Economic development is one of the constituent elements of the concept of modernity and, as such, it can be conceived as one of the propellers of secularisation. The most clear and theoretically simple statement about this link can be found in a work of Pippa Norris and Ronald Inglehart (2007). They posit a simple hypothesis according which religiosity is negatively related to the level of physical, personal and social existential security. Consequently, “the importance of religion persists most strongly in vulnerable populations, especially those living in poorer nations who face risks to their survival” (Norris and Inglehart, 2007: 18). In industrial and postindustrial societies, due to the rising levels of economic wealth and existential security, religion should be in decline. Comparing their theoretical proposition with the data coming from World Values Survey, they note that, for instance, 44% of individuals in agrarian societies attend church at least once a week, while this is the case for 25% of individuals in industrial societies, and for 20% in post-industrial societies. However, they also note that the differences are very small, maybe even non-existent, when it comes to religious beliefs and other indicators of the more personalised spiritualities. In other words, a quest for a meaningful life is probably something that is not going to vanish with the rising levels of material wealth, and religion may be very well suited for the fulfillment of such human needs (Nikodem, Jurlina, 2018: 295).

The data cited above can be interpreted as a firm sign of secularisation, since it may be assumed that those indicators of religiosity which are less costly and obligatory will decline in a more gradual and slow manner. However, they can also be seen as an indicator of the transformation of religion in contemporary postindustrial and postmodern societies. Religion may be changed and accustomed to fit to the changing social cirumstances, some social contexts may favor one, and other some other types of religion. For instance, Luckmann (2003) proposed four social forms of religion which are intertwined with the social context in which they existed. In archaic societies religion and social order where one and the same, while in the first civilisations there was a division between religious and secular state power, but as a rule the state power was legimitised by means of religious arguments. In the Middle Ages until the modern western societies sacred and secular were further divided, with stronger functional specialisation in comparison to earlier societies. And finally, in contemporary societies the so-called invisible religion (Luckmann, 1967) is a dominant social form of religion which is characterised by ecleticism, individualisation and privatisation of religion. Such type of religion implies that secularisation is indeed a complex and multidimensional phenomenon, and that it is perfectly possible that societies can be secular in one sense, but very religious in some other sense. Similar ideas are proposed by Grace Davie (1990, 1994), wherein she attributes the decline of the communal aspects of religion to the general decline of communalism, i.e. to the rise of the individualist mentalities and lifestyles (Putnam, 2000). Even though there are some indications that the decline of belonging and believing is proceeding at the same rate (Voas and Crockett, 2005), the possibility that individual and collective aspects of religion are not equally susceptible to secularization still remains a viable hypothesis to be tested.

Bearing in mind the formulated theoretical framework, we have outlined the following research hypotheses to be tested in the empirical part of the paper:

H1. Societalisation (decline of family and rural communities on one side, and the rise of rational voluntary organisations and political activity on the other) is inversely correlated with religiosity.

H2. Economic wealth (personal wealth and GDP per capita) is inversely correlated with religiosity.

H3. Societalization is more strongly correlated with indicators of church-oriented religiosity (religious services' attendance) in comparison with indicators of non-church religiosity (prayer outside religious services, personal importance of god).

H4. Economic wealth (personal wealth and GDP per capita) is more strongly correlated with indicators of church-oriented religiosity (religious services' attendance) in comparison with indicators of non-church religiosity (prayer outside religious services, personal importance of god).

Data and methods. In this study we used a dataset from European Values Study, i.e. the integrated dataset of the 2017 EVS wave (EVS, 2020). This dataset was used for the multilevel analysis, i.e. for testing the research hypotheses. Various measurements from the dataset, described below, were used in the subsequent analyses.

As first-level predictor variables we chose a couple of indicators of societalisation, personal wealth and other demographic control variables. As first-level indicators of societalisation importance of family in life, importance of democracy, size of the place of residence and membership in voluntary associations were measured. All these variables are supposed to measure the impact of societalisation on the individuals. Importance of family in life is measured on a four-level scale (from very important to not important at all), but we recoded into a dummy variable (1 – very important, 0 – other answers). The reason was the fact that, overall, almost 88% of all respondents stated that their family is very important to them. The size of the place of residence was measured on a scale ranging from 1 (less than 5,000) do 5 (500,00 and more). Membership in voluntary associations is measured as a dummy variable (1 – belongs to at least one of the listed organisations, 0 – belongs to none of the listed organisations), wherein the respondents were asked to indicate whether they belong to various types of associations – trade unions, political parties, ecological groups, sports and recreation groups, charitable/humanitarian organizations, etc. Overall, about 41.5% of all respondents declared that they do not belong to any of the ten listed organisations. Importance of democracy was measured on a scale ranging from 1 (not at all important) to 10 (absolutely important). Income per household member was measured as household monthly net income (x1000), corrected for purchase power parity in euros and divided by the number of household members. As the first-level control variables, gender (0 – female, 1 – male) and age were used.

As the second-level predictors, country GDP per capita, as well aggregated country levels of importance of family in life, importance of democracy and membership in voluntary associations were used. Country GDP per capita was measured as PPP, current international dollars, and the data source was World Bank Database. GDP per capita is taken for the year 2019, and not for the year when the data for the individual countries were collected. The main reason for such decision was COVID-19 health crisis, which led to the GDP decline in 2020, which may have introduced some measurement error to the data.

All continous predictors were grand-mean centered.

As criterion variables, three indicators of religiosity were used – frequency of prayer outside religious services (from 1 – never to 7 – every day), importance of god in one's life (from 1 – not at all important to 10 – very important), and attendance of religious services (from 1 – never, practically never to 7 – more than once a week).

Before the analyses and hypotheses testing, in the Table 1 the descriptive statistics of the study variables is listed.

Research Results and Discussion. As analytical strategy, we conducted a series of multilevel analyses with three measures of religiosity as three criterion variables. Each multilevel analysis consisted of three stages/models. The first stage was an empty model (no predictors), the second stage included random intercept model with first-level predictors, and the third stage random intercept with both first-level and second-level predictors. As already noted, all continuous predictors were grand-mean centered.

As can be seen from Table 2, there is a significant group-level variance, i.e. country difference with regard to attendance of religious services. To be precise, about 15.08% of the total variance can be attributed to the country level (the so-called intraclass correlation coefficient – ICC).

From the second model (individual level predictors), we can note that all predictors are statistically significant, with the exception of the importance of democracy. Women have 0.36 points higher score on the religious services attendance than men, while every year of age increases the attendance score by 0.01 points. An increase of income of one thousand dollars per family member leads to a decrease in the attendance score by 0.14 points. One-point increase in the size of the place of residence decreases the attendance score by 0.04 points. The respondents to whom their families are very important have higher score on the attendance scale by 0.21 points, when compared to those who did not indicate that their families are very important to them. And finally, the persons who are members of at least one voluntary association have higher attendance score by 0.51 points in comparison to the persons who are not members of such associations.

When the second-level predictors are added (the third model), we can note that none of the second-level predictors are statistically significant. However, we can also not that even in the final model there is a significant and substantial intercept variance, i.e. that the countries from the sample are different even when the selected variables are accounted for.

When it comes to the multilevel regression with prayer outside religious services as the criterion variable (Table 3), from the empty model we can note that ICC amounts to 19,23% (i.e. 0.1923), which means that there also exists a significant clustering effect on the country level.

We can see that the same predictors are statistically significant and that the direction of the predictors is the same as in the regression with the attendance level as the criterion variable. The only difference is the third model, wherein one-thousand dollars increases country-level GDP per capita decreases the score on the frequency of prayer scale by 0.04 points. Other group-level predictors, i.e. aggregated levels of family importance, importance of democracy and membership in horizontal associations, are not statistically significant.

The empty model from the regression on the importance of god as the criterion variable (Model 1 in Table 4) we can also conclude that there are significant country-level differences, given that 25.29% of the total variance can be attributed to the country level. Other results are very similar as in the previous multilevel regression. Namely, the same predictors are statistically significant with the coefficients pointing to the same direction.

The parameters of the model fit show that in all regression analyses the subsequent models represent the better fit to the data. Namely, we can see that fit indices are lower for the subsequent nested models, and this is also confirmed by testing the differences between -2 restricted log likelihood with chi-square tests with degrees of freedom that amounts to the differences in the total number of parameters between two models that are being tested.

Conclusion. Overall, the results or our analyses confirmed our first and second hypothesis, but were not in accordance with our third and fourth hypothesis. Namely, there is a significant negative association of almost all measures of societalisation and the measures of individual and societal wealth with the indicators of religiosity. Such correlation was not stronger, in fact it was a little weaker, for the attendance of religious services, in comparison to the prayer outside religious services and the personal importance of god. In consequence, it cannot be confirmed that the impact of societalisation is uneven, and that it provides a space for more personal forms of religion and spirituality.

As for the specific predictors, the importance of the size of the place of residence affirms the validity of the societalization thesis. It seems that small communities, with less social and cultural pluralism that is inherently present in larger settlements, can maintain the sacred canopy (Berger, 1967) that is important as a plausibility structure which is vital for religious life. The same can be concluded from the finding that the importance of family is positively related to religiosity. Family is also an institution that provides religious plausibility structures connected to socialisation, and the decline of importance of family is a sign of societalisation, and consequently of secularisation. With regard to the positive association between the associational membership and religiosity, we might interpret it as a success of religion in building social capital (Smith, 2003; King, Furrow, 2008), i.e. we can assume that the membership in voluntary associations is an inadequate measure of societalisation. Contemporary societies are not dependent on social capital as their driving force, but on the impersonal rules. In Portes and Vickstrom's (2011: 473) words, “large corporations and impersonal markets do not run on social capital; they operate instead on the basis of universalistic rules and their embodiment in specific roles”. Even though voluntary associations indeed express active citizenship and the willingness to actively engage in the try to change one's own social conditions, precisely religion can be a driving force behind a rich associational life. Namely, we should not equate secularization with quietism and privatization of religion (Casanova, 1994). Therefore, our findings might be more a vindication of a thesis that secularization is not equal to privatization of religion, than a recantation of the societalization thesis. As visible from the results' tables, we did not establish any connection between the perception of importance of democracy and religiosity. A possible explanation might be a curvilinear association between the development and the faith in democracy. In other words, citizens of wealthy societies are becoming more critical of democracy since they have high expectations about its effects. Thus, the perception of the importance of democracy might not be a perfect indicator of secularisation.

As for the individual and societal wealth, our results confirm Norris and Inglehart's (2007) hypothesis about the role of scarcity and existential security in the decline of religion. It is particularly interesting that in the case of the analyses with frequency of prayer and importance of god as criterion variables both individual wealth and societal wealth are negatively and independently correlated with religiosity. That means that the individual wealth creates a sense of security, but that also happens with societal wealth, which also creates the overall sense of security that permeates the wealthier societies.

When it comes to the other control variables, i.e. gender and age, our results are in general agreement with the previous knowledge. Namely, according to numerous research studies, women are more religious than men, at least in the Christian countries (Trzebiatowska, Bruce, 2012). Age differences are not so clear-cut, although older persons are more often found to be more religous than the younger ones (e.g., Bengtson et al., 2015; Argue, Johnson and White, 1999), mostly due to the age and period effects. This happen to be the case for most European countries, with countries in East Europe showing no difference between the age cohorts (Pew Research Center, 2018).

As for the limitations of our research, it has to be borne in mind that our analysis of the impact of societalisation and economic wealth is conducted on a cross-sectional data and thus, strictly speaking, cannot form a basis for conclusion about the religiosity trends.

As noted at the beginning, our analysis covers only Europe. A contextualized secularisation theory (Pickel, 2009) needs to take into account possible socio-cultural differences and path dependencies that cannot be explained by a universal theory. In short, socio-cultural and socio-political (Martin, 1978; Martin, 2005) specificites of Europe can influence the position of religion and thus create a spurious connection between modernisation and religiosity. In addition, our results are confined to the “traditional” religion, i.e. they tell us nothing about possible “spiritualities of life” (Heelas, 2008) or “the spiritual revolution” (Heelas et al., 2005) that might replace organized religion in a conventional sense, even though there are some findings that demonstrate that they cannot fully compensate for the decline of traditional religion in Europe (Pollack, 2008).