Religiosity Decline in Europe: Age, Generation, and the Mediating Role of Shifting Human Values

Religiosity is usually defined as an organized system of beliefs, practices, rituals, and symbols related to a higher power (Paterson & Francis, 2017, p. 2; see also: Han & Richardson, 2010; Kidwai et al., 2014; Lavretsky, 2010). It serves different psychological needs. For example, it soothes existential anxiety in the face of the inevitability of death (Pyszczynski et al., 2010) and helps provide cognitive closure and reduces perceived uncertainty (Duriez, 2003; Saroglou, 2002). Religiosity is governed by the underlying processes of identity and motivation. Along with social identity theory (Tajfel & Turner, 1979), religiosity is a significant element of human identity, which regulates a person’s self-concept (“I am a religious person”) and their group belonging (with those who practice the same religion; Van Cappellen et al., 2016). It can be expressed and strengthened through individual and social religious practices. The practices that are carried out alone (e.g., praying) express a person’s self-concept, and they are more likely to be a sign of an intrinsic religious motivation (as opposed to extrinsic; Allport & Ross, 1967), whereas the practices that are carried out with others (e.g., attending religious services) strengthen social bonds and are also representative of social processes.

Although generational affiliation is the key focus of this study, the aspect of age in explaining religiosity levels cannot be ignored. Numerous studies have demonstrated a significant relationship between age and religiosity, with older people being systematically more religious than people from younger groups (Argue et al., 1999; Shulgin et al., 2019; Wilmoth et al., 2015). Most researchers explain the age and religiosity relationship by referring to developmental mechanisms connected to aging and claim that as individuals get older, they become more involved in religious practices and are more likely to identify as religious (Bahr, 1970; Barker et al., 1992; Davie & Vincent, 1998).

Theoretical explanations for the developmental mechanism behind the age-religiosity relationship can be derived from Terror Management Theory (for overviews, see: Kesebir & Pyszczynski, 2012; Pyszczynski et al., 2010) and Socioemotional Selectivity Theory (Carstensen et al., 1999; Löckenhoff & Carstensen, 2004). Terror Management Theory is based on the idea that humans, unlike other animals, are sophisticated enough in their mental abilities to be aware of the fragility of life and the inevitability of ultimate death, and proposes that the awareness of mortality has the potential to generate paralyzing anxiety and that the management of this potential anxiety is essential for effective functioning (Pyszczynski et al., 2010). According to the theory, people develop an anxiety buffering system that, as long as it is functional, protects against existential anxiety, and provides psychological equanimity. Religiosity may also have this function, and therefore may become increasingly important as death gets closer (Jonas & Fischer, 2006) because it serves as emotional compensation and increases well-being toward the end of life (McCoy et al., 2000). Socioemotional selectivity theory proposes that as their time horizons become shorter, people tend to change their priorities and focus more on positive events and cues than on negative ones (Löckenhoff & Carstensen, 2004). Religion may provide these positive stimuli, and its social aspects may help to increase a sense of community, leading to decreased anxiety, increased well-being, longevity, and somatic health (Abdel-Khalek et al., 2019; Cox & Hammonds, 1989; Hall, 2006; Saleem & Saleem, 2020).

Because people become more religious with age and, at the same time, societies are aging (Vancea & Solé-Casals, 2016), we should be able to observe an increase in religiosity over the past decades. However, what we have recently witnessed is actually a clear decline in religiosity (Shulgin et al., 2019; Voas & Chaves, 2016). A possible explanation for this decline is that the observed relationship between age and religiosity is a statistical artifact associated with cohort replacement (Argue et al., 1999): older people represent earlier generations and it is their generational affiliation that explains their religiosity levels and not mere age. If this is true, then we should be looking at the characteristics of each subsequent generation that could explain religiosity decline, and not only at developmental mechanisms connected to age. One such pattern underlying religiosity decline may be the intergenerational shift in human values.

In the psychological sense, values are beliefs or cognitive representations of motivations and goals (Schwatz, 1992). They constitute the core of human identity and serve as cross-situational standards or criteria for evaluating behaviors and decisions (Rokeach, 1973). According to Schwartz’s model (Schwatz, 1992), 10 basic values—self-determination, stimulation, hedonism, achievement, power, security, conformity, tradition, benevolence, and universalism—exist in a circular structure that determines the relationships between them (positively related or conflicting). This perspective also allows for the grouping of basic values into higher-order dimensions: openness to change (self-determination, stimulation, and hedonism), self-enhancement (achievement and power), conservation (security and tradition), and self-transcendence (benevolence and universalism) (Schwartz et al., 2001; Sortheix & Schwartz, 2017). Openness to change and self-enhancement represent personal focus, expressing individual needs and attitudes, whereas self-transcendence and conservation represent social focus, regulating how a person relates to others. Furthermore, openness to change and self-transcendence represent self-expansive and growth values, whereas self-enhancement and conservation represent self-protective, anxiety-based values (Saroglou et al., 2004; Schwartz et al., 2001).

The relationship between values and religiosity is twofold. First, specific religious traditions relate to individual value hierarchies (Schwartz & Huismans, 1995) because these two areas are strongly intertwined as they fall within a broader framework of personal and social identity. Being religious means upholding specific values (“I am Christian = I hold Christian values”) and specific religious traditions dictate which values people are supposed to uphold and how they should function within society (e.g., Weber, 2013). Indeed, empirical evidence confirms that socialization with respect to religion can be viewed as a process of social transmission of values (Saroglou et al., 2004).

Second, and more pertinent to this paper, the level of religiosity can be reinforced by some values and weakened by others, regardless of a specific religious tradition, because psychologically some values play a similar role to religiosity. Specifically, values falling into the dimension of conservation, such as tradition and security, are rooted in anxiety reduction and uncertainty avoidance (Schwatz, 1992; Sortheix & Schwartz, 2017), so they serve psychological functions similar to religiosity (Duriez, 2003; Greenberg et al., 1986; Saroglou, 2002). Other values, such as conformity and benevolence, have social functions similar to those of religious identity, consistent with social identity theory (Tajfel & Turner, 1979).

Moreover, religiosity is primarily identified with faith in delayed reward (eternal life in exchange for temporal duties). Therefore, the values associated with gratification "here and now" (e.g., hedonism) have a negative association with religious beliefs (Schwartz & Huismans, 1995). Values that are most commonly expressed through behaviors inconsistent with the requirements dictated by most religious traditions—engaging in sensory or bodily pleasures (hedonism), seeking out novelty (stimulation) or freedom and independence in choosing one’s life goals (self-direction) (Bardi & Schwartz, 2003)—are difficult to align with religious doctrines. It is therefore natural that higher levels of openness to change are related to lower religiosity. Indeed, empirical evidence confirms that religiosity is positively associated with tradition and conformity, and to a lesser extent, with security and benevolence. It is also negatively associated with hedonism, stimulation, and self-direction, and, to a lesser extent, with achievement, power, and universalism (Chan et al., 2018; Hitlin & Piliavin, 2004; Kuşdil & Akoğlu, 2014; Maio, 2010; Rohan, 2000; Saroglou et al., 2004). These findings seem robust across different contexts (religious denominations, countries, etc.). Therefore, in this study, we also expect religiosity to be related mainly to high endorsement of conservation values and low endorsement of openness to change values. We also predict that the relationship between religiosity and self-transcendence and self-enhancement values will be weaker or even absent.

The term “generation” signifies more than just being born at a specific time within a specific culture; it also denotes shared life experiences, attitudes, goals, and ways of interpreting reality (Mannheim, 1970). For example, the term “Generation X” (Coupland, 1991) is used to describe people born in the 1960s or 1970s for whom career and material wealth are of key importance. Their aspirations in the spiritual domain are low. Tapscott (2009) suggested that people’s generational affiliations can be determined by the media that was dominant when they were at the peak of their cognitive and social functioning. After the Silent Generation, or Matures, (born between 1928 and 1945), there are three post-World War II generations: Baby Boomers (born between 1946 and 1964), for whom radio and television were the key media; Generation X (born between 1965 and 1976), who experienced a shift in the dominant media; Generation Y, or Millennials, (born between 1977 and 1997), whose mindset was formed mostly by the Internet, and Generation Z (born after 1998).

Distinctions between generations based on year of birth are quite arbitrary, but there is no doubt that to some extent, they define groups of people who share common values and are aware of their uniqueness with respect to other groups. For example, the Matures, who experienced the Great Depression and World War II, are considered to be hardworking and devoted, and willing to sacrifice for their loved ones, conform to tradition, and accept authority (Strauss & Howe, 1992). Their values are therefore conservation rather than openness to change and self-transcendence rather than self-enhancement (Lyons et al., 2007). Baby Boomers, who were born and raised in the era of economic prosperity in the post-war period, had to compete for attention in their childhood and for jobs as they entered the job market, but they also grew up with a sense of entitlement, expecting prosperity and satisfaction in their lives. They are said to be self-indulgent, hedonist, pleasure-seeking, and achievement oriented; however, they share a concern for society and are perceived as optimistic idealists (Lyons et al., 2007; Strauss & Howe, 1992). Compared with Matures, they are more likely to endorse openness to change values, be less attached to conservation values, and be similar to the previous generation in terms of self-transcendence and self-enhancement (Lyons et al., 2007).

Generation Xers were raised in the age of economic uncertainty. They experienced increased parental divorce rates and grew up in single-parent families much more often than members of previous generations (Howe & Strauss, 2009). They are often described as highly cynical, severely independent, and entrepreneurial (Lyons et al., 2007; Wey Smola & Sutton, 2002), preferring to rely on their own efforts rather than expecting help from others, favoring change over stability and tradition, and self-enhancement over self-transcendence (Howe & Strauss, 2009; Lyons et al., 2007). Like Generation Xers, Millennials are characterized as independent, innovative, adaptable, and change-oriented; however, they are considered more optimistic and self-absorbed than Generation Xers (Tapscott, 2009).

Using the Schwartz Value Survey, Lyons et al. (2007) demonstrated that the two younger generations, Generation Xers and Millennials, value self-transcendence and conservation less than the two older generations, Matures and Baby Boomers. The same pattern had been observed in earlier research that examined age rather than generation; age was positively correlated with conservation (tradition, conformity, and security) and self-transcendence (benevolence and universalism) and negatively correlated with openness to change (self-direction, stimulation, and hedonism values) and self-enhancement (power and achievement values) (Schwartz et al., 2001). Hence, it seems difficult to unravel the effects of age and generation in cross-sectional research. However, some longitudinal studies and meta-analyses have suggested that people and societies are shifting toward more individualistic values and away from social values. Over the past few years, various studies have demonstrated increasing levels of individualism (Marcus et al., 2017; Twenge, 2014), self-enhancement, extrinsic values (Kasser, 2016; Twenge & Campbell, 2014; Twenge et al., 2012) and agentic orientation (Twenge, 1997), but decreasing levels of intrinsic values (Twenge et al., 2012), empathy (Konrath et al., 2011) and secure attachment styles (Konrath et al., 2014). Most of these studies draw their data from Western or highly educated urban samples; hence, value changes observed in these studies may be explained by the fact that as societies become more prosperous, they also become less traditional (Bakan, 1966). This global shift toward individualistic, extrinsic, and self-enhancement values (Kasser, 2016; Marcus et al., 2017) has also been connected to the development of new media, such as the Internet, underlining the fact that more recent generations are born and raised in different societal contexts than the earlier ones, and are not just younger (Tapscott, 2009).

As stated above, values and religiosity are connected (Roccas & Elster, 2013; Saroglou et al., 2004), and, therefore, a change in values that occurs between generations should also be reflected in changes in religiosity. For example, Gay et al. (2015) found that Generation X is a group that seeks spirituality, but not necessarily religiosity, whereas Generation Y is less religious than Baby Boomers but similar to Generation X in terms of engaging in religious practices within a specific tradition. It seems that generational affiliation may be connected to religiosity in the sense that the decline in religiosity is possibly related to generational patterns rather than age per se. This “cohort” theory (Firebaugh & Harley, 1991; Hout & Greeley, 1987) has received some support with analyses of big data sets conducted by the Pew Research Centre (Taylor & Keeter, 2010). Data gathered since the 1970s showed that differences in religiosity were significant between generations, but that religious attitudes within each cohort were relatively stable (Fig. 1). Additionally, to confirm these findings with another longitudinal demographic set, we explored data collected in the 32 subsequent waves of the General Social Survey (spanning the period from 1972 to 2018) in an independent analysis (see Supplemental Material for details). These results provide further support for the superiority of the cohort effect over the aging effect.

Given these findings, we expect more recent European generations to be less religious than earlier ones. We propose that this effect goes beyond the simple age differences between people from different generations and stems from generational shifts in human values.

This research tests the hypothesis that a generational shift in values explains the observed decline in religiosity beyond the effect of age. First, we expected that more recent generations would be less religious than earlier generations, and that this relationship would be independent of the developmental effect of age on religiosity. Second, with each subsequent generation, we expected a decline in social focus values, especially in conservation and, to a lesser extent, in self-transcendence, and a parallel increase in personal focus values, including openness to change and self-enhancement. Third, considering the results of the aforementioned meta-analysis (Saroglou et al., 2004), we expected that diminished conservation and increased openness to change values would predict a deterioration in the level of religiosity, and that this effect could be attributed to generational changes (above any effects of age found).

To verify our predictions, we first analyzed data from the seventh round of the European Social Survey (ESS7), conducted in 19 countries in 2014, and included participants representing three post-World War II generations: Baby Boomers, Generation X, and Generation Y. We tested whether four dimensions of basic human values—conservation, self-transcendence, openness to change, and self-enhancement—mediate the relationships between generational affiliation and religiosity. We found that each subsequent generation was less religious, and this effect could mainly be attributed to a shift from conservation values to openness to change values. However, because these data were collected at one point in time, it was impossible to disentangle the effect of age and the effect of generation in the analyses. The year of the participant’s birth determined age and their generational affiliation, and there was a full match between those two variables within one survey; that is, participants of the same age belonged to the same generation. Therefore, in the second step, we employed data from all rounds of the European Social Survey, from round 1 conducted in the period of 2002–2003 to round 9 conducted in the period of 2018–2019. This approach allowed us to partially separate age and generation. Year of the participant’s birth determined age and their generational affiliation, but because the time of the study changed, there was no longer a full match between those two variables across surveys; there were participants of the same age in subsequent surveys who belonged to different generations. We demonstrated that older people were less religious than younger people, and this effect was partially explained by the shift from openness to change values to conservation values. However, using two different operationalizations of generation, we found that generational decline in religiosity could also be attributed to generational increases in self-expansive, growth-based values constituting openness to change that was independent of the effect of age.

Mediation analysis was conducted using Mplus 8.1 (Muthén et al., 2017). We used the robust full information maximum likelihood estimation method to estimate the assumed model of relationships with generational affiliation as an independent variable, four higher-order value dimensions as parallel operating mediators, and religiosity as the dependent variable. We also allowed for a correlation between the mediators. More specifically, we tested (1) the total effect of generation on the level of religiosity, (2) indirect effects of generation on the level of religiosity via each value dimension, controlling for the other dimensions, and (3) the total effect of generation on the level of religiosity controlling for four mediators. All continuous variables were Z-scored before the analysis to obtain the standardized coefficients. Because generation was a discrete ordinal variable with just three values, we used sequential coding (Hayes & Preacher, 2014). With sequential codes, the relative direct and indirect effects can be interpreted as the effect of membership in one group relative to the group one step sequentially lower in the ordered system. In our case, we used one dummy variable representing the comparison between Generation X and Baby Boomers, and a second dummy variable representing the comparison between Generation Y and Generation X. We examined the goodness of fit using multiple indices: the root mean square error of approximation (RMSEA) and 90% confidence interval (90% CI), standardized root mean square residual (SRMR), comparative fit index (CFI), and Tucker-Lewis index (TLI). We used multiple fit indices to assess different types of model fit (e.g., model parsimony and absolute fit), and for a more reliable and conservative evaluation when used together (Brown, 2006). The model was evaluated using the criteria proposed by Hu and Bentler (1999): the lower bound for a good fit was 0.95 for CFI and TLI, 0.06 for RMSEA, and 0.08 for SRMR. Indirect effects were tested by examining bootstrap confidence intervals for indirect effects using a bootstrap procedure with 10,000 samples.

The model fitted the data perfectly because it was saturated (RMSEA = 0, 90% CI [0, 0], SRMR < 0.001, TLI = 1, CFI = 1). All the hypothesized path coefficients were significant, except for the effect of Generation Y versus Generation X on self-transcendence. After excluding this insignificant path from the model, it fitted the data very well in light of all the examined indices (RMSEA = 0.005, 90% CI [0, 0.017], SRMR = 0.001, TLI = 0.99, CFI = 1). The results (standardized path coefficients) are presented in Table 2.

We observed significant effects of generations on all four value dimensions. For instance, Generation Xers, when compared with Baby Boomers, were less likely to endorse both social focus values (i.e., conservation and self-transcendence) and more likely to endorse both personal focus values (i.e., openness to change and self-enhancement). The effects for Generation Y in comparison with Generation X revealed a similar pattern, except that the path to self-transcendence was not significant. The total effects of generation on religiosity were significant, with Generation X being less religious than Baby Boomers, and Generation Y being less religious than Generation X. While controlling for mediators, the direct effects of generation remained significant, albeit weaker. The effect of conservation values on religiosity was significant and positive, similar to the effect of self-transcendence values, with the former being significantly stronger than the latter, Z = 17.97, p < 0.001. The effect of openness to change values on religiosity was significant and negative, whereas the respective effect of self-enhancement values was weak and positive. The predictors and mediators accounted for R² = 9.1% of the variance in religiosity.

Further investigation of the 95% bootstrapped confidence intervals (95% bootCIs) for the relative indirect effects from generation to religiosity (see Table 2) confirmed significant indirect effects of Generation X (vs. Baby Boomers) on religiosity via conservation values, 95% bootCI [− 0.04, − 0.03] and via openness to change values, 95% bootCI [− 0.03, − 0.02], and indirect effects of Generation Y (vs. Generation X) on religiosity via conservation values, 95% bootCI [− 0.03, − 0.02] and via openness to change, 95% bootCI [− 0.05, − 0.03]. The indirect effects via self-enhancement and self-transcendence were much weaker (95% bootCI [0.002, 0.01] for the effect of Generation X [vs. Baby Boomers] on religiosity via self-enhancement, 95% bootCI [0.003, 0.01] for the effect of Generation Y [vs. Generation X] on religiosity via self-enhancement, and 95% bootCI [− 0.003, − 0.001] for the effect of Generation X [vs. Baby Boomers] on religiosity via self-transcendence), and their significance was most likely a result of the large sample size.

To summarize, using data from EES round 7, we demonstrated that Generation Y is less religious than Generation X, and Generation X is less religious than Baby Boomers. These differences in religiosity between generations are at least partially explained by generational differences in values, that is, lower endorsement of conservation values and higher endorsement of openness to change values in later generations compared with the previous ones. The results for self-enhancement and self-transcendence were inconclusive, with relatively weaker effects. This study provided preliminary support for our assumption that generational changes in values lead to a decrease in religiosity.

However, because we employed data that was collected at one point in time and coded generation as a function of year of birth, we could not separate the effect of generation from the effect of age; participants from later generations were not only born later, but they were also younger than participants from earlier generations. Furthermore, because of the relatively low representation of participants born before 1945 and after 1998 in the ESS7 sample, we had to limit our analysis to three generations only. Therefore, we decided to rerun our analysis using combined data from all rounds of the European Social Survey.

As in Study 1, mediation analysis was conducted with Mplus 8.0 (Muthén et al., 2017) using the robust full information maximum likelihood estimation method. We tested a model of relationships with generational affiliation and age as independent variables, four dimensions of values as parallel operating mediators, and religiosity as the dependent variable. We also allowed for a correlation between the mediators. More specifically, we tested (1) the total effect of generation on religiosity, (2) indirect effects of generation on religiosity via each value dimension, controlling for the other dimensions, and (3) the total effect of generation on religiosity controlling for four mediators. All variables were Z-scored before the analysis to obtain the standardized coefficients.

The model fitted the data perfectly (RMSEA = 0, 90% CI [0, 0], SRMR < 0.001, TLI = 1, CFI = 1). All the hypothesized path coefficients were significant, except for the effect of generation on self-enhancement. After excluding this insignificant path from the model, it fitted the data very well in light of all the examined indices (RMSEA = 0.003, 90% CI [0, 0.007], SRMR < 0.001, TLI = 1, CFI = 1). The results (standardized path coefficients) are presented in Table 4.

We observed significant effects of age on all four value dimensions, such as older participants who, when compared with younger participants, were more likely to endorse conservation and self-transcendence values, and less likely to endorse openness to change and self-enhancement values. Controlling for age, we found significant effects of generation on the three value dimensions. In line with our predictions, younger (later) generations were more likely to endorse openness to change values. However, contrary to our hypotheses, they were also more likely to endorse conservation and self-transcendence values. The total effect of age on religiosity was significant and positive. Controlling for age, the total effect of generation on religiosity was significant and negative. While controlling for mediators, the direct effect of age remained significant and positive, whereas the direct effect of generation remained significant and negative.

As in Study 1, the effect of conservation values on religiosity was significant and positive, similar to the effect of self-transcendence values, with the former being significantly stronger than the latter, Z = 63.65, p < 0.001. Again, as in Study 1, the effect of openness to change values on religiosity was significant and negative, and the effect of self-enhancement values on religiosity was weak and positive. The predictors and mediators accounted for R² = 12% of the variance in religiosity.

Further investigation of the 95% bootCIs for the relative indirect effects of age on religiosity revealed significant indirect effects via conservation values, 95% bootCI [0.058, 0.064], and via openness to change values, 95% bootCI [0.028, 0.031]. The indirect effect via self-transcendence was positive but much weaker than the other effects, 95% bootCI [0.001, 0.002], and the effect of age on religiosity via self-enhancement was weak and negative, 95% bootCI [− 0.007, − 0.005]. The significance of the two latter indirect effects may be due to the large sample size and should be considered with caution. The overall indirect effect of age on religiosity via human values was significant and positive, β = 0.09, SE = 0.002, Z = 49.29, p < 0.001, 95% CI [0.08, 0.09].

Finally, we investigated the 95% bootCIs for the relative indirect effects from generation to religiosity. In line with our expectations, the indirect effect of openness to change was significant and negative. The indirect effects via conservation and self-transcendence were positive but weaker than the other indirect effects. The overall indirect effect of generation on religiosity via human values was significant and negative, β = − 0.004, SE = 0.002, Z = − 2.81, p < 0.001, 95% CI [− 0.01, − 0.001].

To summarize, using the data from multiple rounds of the ESS, we were able to independently analyze the effect of age and generation on human values and religiosity in the same model. We demonstrated that, after controlling for generation, age was positively related to religiosity. This means that within the same generation, older people were more religious than younger ones. However, this effect was accompanied by a negative effect of generation after controlling for age, indicating that two people of the same age but coming from different generations (thus, surveyed in different waves of ESS [e.g., a 19-year-old in round 1 vs a 19-year-old in round 2) would also differ in terms of religiosity, with members of more recent generations being less religious than members of older ones. Furthermore, the effects of age and generation were mediated by basic human values. Older people were more prone to endorse conservation values and less prone to endorse openness to change values, which led to higher religiosity. However, we also observed a unique indirect effect of generation via openness to change on religiosity, present for two different operationalizations of generation (see Supplemental Materials for additional analyses). People from more recent generations were more open to change than those from less recent ones, which in turn led to lower levels of religiosity, and this effect was present over and above the effect of age on religiosity both directly and via basic human values. Interestingly, when controlling for age, we observed a positive—and not negative as expected—effect of membership in a younger generation on self-transcendence. However, this effect translated into increased religiosity, with an extremely small effect size. Finally, we also systematically did not observe the effect of generation on religiosity via decreased conservation values. In fact, we found some increase in conservation in more recent generations when compared with the older ones, but this increase translated into only slight proliferation of religiosity. Hence, the increase in conservation values leading to higher religiosity should be attributed to age differences rather than to generational affiliation, while the alterations in openness to change affecting religiosity could be attributed both to age (with older people being less open to change and therefore more religious than younger people) and generational differences (with more recent generations being more open to change and therefore less religious than older generations).

One major limitation of our study is its cross-sectional design, which did not allow us to draw causal inferences about the relationships between our variables of interest. The mediation model was built on the assumption that value structure, which is closely related to basic personality traits and people’s conscious goals (Fischer, 2018), affects levels of religiosity; however, this relationship is bidirectional (Chan et al., 2018; Roccas & Elster, 2013). On the one hand, people may feel motivated to belong to groups with shared social goals (Gandal et al., 2005), and an important aim of religions is to shape its members’ values according to its ideology (McCullough & Willoughby, 2009). Unfortunately, the ESS did not include the same participants in each round, so it does not provide longitudinal data at the individual level that would at least partially allow us to answer the question of causality. Future research should collect longitudinal data to better understand the causality of the investigated effects.

Another important limitation of our study is that the list of countries participating in the ESS is rather limited, focusing on countries strongly embedded in the European tradition. The advantage of this dataset may be the strong cross-country variation in religiosity (from very religious Poland to strongly secularized Czechia), while the undoubted disadvantage is its geographical and cultural homogeneity. Therefore, our findings can only be generalized to other countries with similar cultures or European origin (e.g., North America), but one should be very cautious when using these data in highly collectivistic countries. Furthermore, we did not investigate the country-specific relationships between age, generation, values, and religiosity. However, the strength of the relationship between residents’ values and individual religiosity may differ depending on the general level of religiosity in a specific country. On the one hand, in highly religious societies, socially oriented people may be naturally motivated to identify as religious, because they then reach a person-environment congruence in values known to be associated with better well-being (Sortheix & Schwartz, 2017). In addition, this effect can be explained in light of the aforementioned terror management theory (Greenberg et al., 1986), which describes ways of coping with the perspective of inevitable death. In highly religious societies, where believing in God and the afterlife is the most common way of reacting to the salience of mortality, conservation values (tradition, conformity, and security) are core elements of social identity. Therefore, people high in those values try to fulfill them through a socially activated worldview: religion. On the other hand, in highly secularized countries, relationships between values and religious affiliation may be much weaker, presumably because of other (non-religious) foundations underlying people’s beliefs and goals. Confirming self-esteem (related to achievement and power) is the most common method of coping with death anxiety among non-religious people (Kashima et al., 2004), and the openness to change values such as hedonism, stimulation, self-direction are more likely to dominate in highly individualistic cultures (Cukur et al., 2004). In secularized countries, conservation values may poorly reflect individual religiosity, mainly because of different patterns of social identity bonds. In their meta-analysis on the relationship between human values and religiosity, Saroglou et al. (2004) demonstrated that the more developed a country, the less positive the correlation of religiosity with conservation, and the less negative the correlation of religiosity with self-direction and achievement. Because country religiosity declines as material security and quality of life improve (Barber, 2013), this may also translate into diverse relations between values and religiosity in different countries. Hence, future research should investigate the relationship between age, generation, values, and religiosity, not only at the individual level, but also taking country specificity into account.