Introduction

The relevance of religious denomination to mental health was first exemplified by Durkheim (1897) who distinguished Protestantism from Catholicism and Judaism in terms of their ability to offer integrative values. Since then, many researchers have examined the universals and the specifics of different denominations to explain differences in health and well-being across religious groups. However, to date only a handful of studies examined how these denominational differences are dependent of occupying a religious minority position, as well as how the meaning of occupying a minority position to mental health may be moderated by the broader religious context.

Available studies find that occupying a religious minority position within a particular context can lead to more depressive feelings in minority members (May and Smilde 2016; Van de Velde et al. 2017). According to the minority stress model, minorities may have less social support, a more vulnerable socioeconomic position, and/or face more prejudices and discrimination (Missinne and Bracke 2012; Pascoe and Smart Richman 2009). At the same time, people’s interpersonal relations are preferably homogeneous regarding many sociodemographic, behavioral, and intrapersonal characteristics (McPherson et al. 2001). This homophily preference in people’s networks yields large consequences for their mental health and social life experiences (e.g., level of happiness and sense of belonging: Okulicz-Kozaryn 2019). Consequently, being part of a specific religious community may provide (preferred) social networks that could even be strengthened, just because of this minority position, which could enhance one’s mental well-being (May and Smilde 2016; Verkuyten and Slooter 2007).

Furthermore, it may be of large interest to distinguish between religious denominations as many of them prescribe a code of conduct for both private and public life and thus study its importance within the particular religious context. For instance, many Muslims living in Europe find themselves in a minority situation. Because they occupy a minority position in many European societies, their religious denomination constantly feeds and is entangled with debates concerning and the extent to which Muslims should be loyal and follow religious prescriptions in their daily life habits and how such religious prescriptions should be expressed in social interactions (Haque 2004; Siddiqui 1997). The congruency and relationship between the norms and values of minority/majority religious denominations and how these relate to mental health are depend on the historical relationship between religious denominations and the religious organization of societies (Foner and Alba 2008), the religious heterogeneity within a particular context (Wiepking et al. 2014), a country’s social norm of religiosity (Stavrova et al. 2013), and the alignment of the moral messages (Ellison 1991). Making a clear distinction between different religious denominations is of utmost importance, given the increased importance of religious affiliation over ethnicity (Barth 1998; Gans 1994), and the increasing secularization trends in Westerns societies over the last decades (Berger 1967; Norris and Inglehart 2011), along with the stigmatization of particular religious denominations within a particular context (Siddiqui 1997; Titarenko 2008). In addition, certain religious denominations may be more tolerant toward other religious groups. A study by Pickel (2009), for example, found that in Catholic countries, there is a far more open-minded attitude toward the presence of religion in schools than in Protestant countries, which may reflect the overall tolerance of religious praxis in public spheres.

Available comparative research on this topic is, however, limited in scope. A number of studies find that the positive association between religious denominations and well-being is strengthened in regions where religion is normative (Elliott and Hayward 2009; Hayward and Elliott 2014; Okulicz-Kozaryn 2010). The study by Okulicz-Kozaryn (2011) shows that people are less happy in religiously diverse societies, but it did not examine how this relates to the concurrency with the individual’s religious denomination. Van Tubergen and colleagues (2005) compared suicidal behavior in Catholics, Protestants, and non-affiliated persons in the Netherlands. It finds that the congruency between the individual and majority denomination does not explain differences in suicidal behavior, but rather the presence of any strong religious community that prohibits suicide. Huijts and Kraaykamp (2011) compared subjective health differences in Protestants and Catholics in a wide range of European countries. It finds that the health advantage of Protestants as compared to Catholics is greater as the percentage of Protestants in a country is higher. Finally, May and Smilde (2016) examined levels of well-being of religious minorities in a number of majority Catholic nations. It finds that the personal benefits of participation in a minority religion are less pronounced in countries where the size of the majority religion is larger. However, because this study solely examined Catholic majority countries, it remains unclear whether these results can be translated to other religious contexts.

With this study, we build upon and extend this current knowledge in several ways. First, we compare levels of mental health across four of the most prevalent religious denominations within the European context (Catholic, Protestant, Eastern-Orthodox, and Muslim) and compare these with the growing number of people who are not affiliated to a Church. Given that comparative research on the Muslim community and the Eastern-Orthodox community in Europe is underdeveloped, our study fills this hiatus in the literature. Second, we examine whether these denominational differences in mental health are dependent on whether the person belongs to a religious minority group. Third, we examine how levels of mental health in religious minorities are moderated by the majority denomination in different European regions. Finally, given the recent influx of migrants within the European continent, along with the increase in anti-immigration attitudes (Meuleman et al. 2009), we examine whether the association between mental health and occupying a religious minority position can be attributed to either perceived experiences of religious discrimination and/or by a migration status. Finally, our study examines levels of depression across religious denominations. This health outcome is particularly interesting, given its strong association with religiosity (Koenig 2012; Schieman et al. 2013). People may turn to religion in times of distress, while at the same time certain religious behaviors may actually be harmful to mental health. To date, only one comparative study examined the association between levels of depression and religiosity, but focused on religious praxis, rather than religious denominations (Van de Velde et al. 2017). To the best of our knowledge, our study will be the first to present variations in levels of depression across and within the major religious denominations in Europe.

Data and Methods

We used data from wave 6 and wave 7 of the European Social Survey. Data were gathered during 2012 and 2014 by face-to-face interviews among a sample of the resident national population aged 15 or older, irrespective of citizenship or nationality. The sample was restricted to respondents from age 18 to 75. The unweighted sample, after listwise deletion of missing cases on the variables in the analyses, consisted of 71,627 individuals. These individual data were nested within 268 regions, which were nested within 28 countries. The regional variable constructed by the ESS was used.

Measures

Dependent variable—An 8-item version of the Center for Epidemiologic Studies–Depression Scale (CES-D-8) scale was used to measure the frequency and severity of depressive feelings. Respondents were asked to indicate how often in the week previous to the survey they felt or behaved in a certain way (felt depressed, felt that everything was an effort, slept badly, felt lonely, felt sad, could not get going, enjoyed life, or felt happy—last two items are reverse coded). Response categories forming a 4-point Likert scale ranged from none or almost none of the time (0) to all or almost all of the time (3). Scale scores for the CES-D-8 were assessed using non-weighted summed rating and ranged from zero to 24, with higher scores indicating a higher frequency and severity of depressive complaints. The reliability and the validity of the inventory were confirmed across countries (Van de Velde et al. 2010).

Individual-level variables—Respondents were asked whether they considered themselves as belonging to any particular religion or denomination, and if yes, they could choose between eight options. Based on those two questions, we constructed a categorical variable personal denomination including six categories: Protestant (reference category), Roman Catholic, Eastern-Orthodox, Muslim, other, and no affiliation. The other category covers a small sample of respondents belonging to the Jewish denomination, to Eastern religions, other Christian denominations, or to other non-Christian denominations.

The level of religiosity of respondents was measured by a self-constructed scale, based on the results of the principal component analysis, combining the items: ‘Regardless of whether you belong to a particular religion, how religious would you say you are?’ (correlation: 0.872), ‘Apart from when you are at religious services, how often, if at all, do you pray’ (correlation: 0.889) and ‘Apart from special occasions such as weddings and funerals, about how often do you attend religious services nowadays?’ (correlation: 0.860). Cronbach’s alpha of the scale is 0.804. A higher score on the scale refers to a higher level of religiosity.

Religious minority was captured by a dichotomous variable to indicate whether respondents were a member of a religious minority group in the region where they lived. Respondents were categorized as a member of a religious minority (score 1) if they shared their affiliation with less than 20 percent of the inhabitants of the region where they lived. Individuals who shared their religious affiliation with at least 20 percent of the inhabitants of the region where they lived were coded as not being a religious minority (score 0). This variable was constructed by combining, on the one hand, the individual-level information on religious affiliation and, on the other hand, by aggregating this information to the regional level.

A dichotomous variable distinguished respondents who reported perceived religious discrimination (score 1) versus respondents who did not report perceived discrimination (score 0), based on the questions ‘Would you describe yourself as being a member of a group that is discriminated against in this country?’ and ‘On what grounds is your group discriminated against?’ with ‘Religion’ as answer category (several grounds could be reported).

Migrant background was based on the respondent’s country of birth and that of his/her parents. We distinguished between natives (the country of birth of the parents and the respondents were similar to the country of residence), first-generation migrants (the country of birth was different from the country of residence), and second-generation migrants (the country of birth was similar to the country of residence, but the country of birth of one of the parents was different).

We controlled for gender, with men as reference category. To account for the nonlinear association between depression and age, we included both age and age squared in our analyses. Labor market position was included as a dichotomous variable that indicates whether respondents were in paid employment (1) or not (0: reference category). Educational level is a metric variable based on the total number of years of full-time education completed by respondents. Income position is a categorical variable representing the household income as a proportion of the national median equivalent income. The four categories represent people living in relative poverty (< 50% of the median equivalent income) (OECD 2019), in an income group below the national median income group (50–79% of the median equivalent income), in an income group around the national median (80–119% of the median equivalent income), and those with a relatively higher income (≥ 120% of the median equivalent income: reference category). The dichotomous variable ‘marital status’ indicates whether respondents were married or in a civil relationship (unmarried: reference category).

Contextual-level variables—At the regional level, we included information on the regional majority denomination, by aggregating data on personal denomination to the regional level. If one of the religious denominations had at least 50 percent adherents among the respondents, we identified that denomination as the regional majority denomination. If none of the denominations reached 50 percent within a certain region, then that region was considered to have no denominational majority. As a result, the variable regional majority denomination consists of six categories: Protestant (reference category), Roman Catholic, Eastern-Orthodox, Muslim, no majority, and non-affiliated. The latter category comprises regions where more than 50 percent of the respondents was not affiliated to any denomination. An overview of the majority denominations per region is provided in "Appendix".

At the country level, we controlled for GDP per capita, former Soviet membership, and political stability and absence of violence in order to take into account the uneven distribution of denominations across these country-level indicators. The average GPD per capita was included with a time lag of one year before the collection of the survey data (2011 to 2014). As the distribution of the denominational majorities is unevenly distributed between former USSR countries and non-former USSR countries, and there is also a difference in depression scores between both groups of countries (Van de Velde et al. 2010), we included a dummy that differentiates between countries that were former USSR countries or not. Finally, we controlled for the degree of political stability and absence of violence and terrorism through an index of political stability from the year 2012 (Kaufmann et al. 2013). This index measures perceptions of the likelihood that the government will be destabilized or overthrown by unconstitutional or violent means, including politically motivated violence and terrorism. An additional sensitivity analysis adjusted the results for the national level of religious fractionalization and for the national level of health-care expenditure. However, given that these controls did not change the results, we excluded these measures from the final analysis.

Statistical Analysis

A three-level multilevel model was applied with respondents (Ni = 71,627) nested within regions (Nj = 268), which were again nested within countries (Nk = 28). A total of seven models were estimated, of which the first two are presented in Table 2 and the other models in Table 3. The first model consists of personal denomination, the religiosity scale, and the individual (age, age square, gender, period, income, employment status education, and marital status) and country-level control variables (GDP, health-care expenditures, former USSR country-dummy). In model two, we estimated whether the relation between personal denomination and depressive feeling varied according individual’s level of religiosity, by adding an interaction effect between personal denomination and the religiosity scale. This interaction effect is additionally presented in Fig. 1. Model 3 is similar to model 1, but we included the religious minority measure instead of the religiosity scale. This enabled us to estimate the relation between individual denomination and depressive feelings, irrespective of being a member of a religious minority group. In Model 4, the regional majority denominations were added to the model, while the personal denomination was excluded from the model. In the next model (Model 5), we examined whether the relation between being a member of a religious minority and depressive feelings differed according the regional majority denomination by including a cross-level interaction effect. This interaction effect is also presented in Fig. 2. Thereafter, we added information on whether respondents reported perceived discrimination based their religion as well as whether respondent had a migrant background.

Fig. 1
figure 1

Estimated mean scores of depressive feelings by personal denomination and levels of religiosity

Full size image
Fig. 2
figure 2

Estimated mean scores of depressive feelings in religious minorities by religious majority regions

Full size image

Results

Descriptive Results

The distribution of religious denominations in our sample (results not presented in table) shows that the largest group of respondents was not affiliated to a religious denomination (43%), followed by the Roman Catholics (30%) and Protestants (12%). Almost eight percent of the respondents were Eastern-Orthodox, while 4 percent was Muslim and two percent was affiliated to another smaller denomination. Almost 40% of the respondents lived in a region where the majority did not belong to a religious denomination and 30% in a region with a Roman Catholic majority. Only three percent of the respondents lived in a region with a Muslim majority and almost four percent in a region with a Protestant majority (not presented in the table).

Table 1 presents the mean depression scores per personal denomination. Depression scores were highest in the Muslim sample (4.8), followed by the Eastern-Orthodox (4.2), while the lowest scores can be found among the Protestants (3.0) followed by the non-affiliated individuals (3.4). Almost half of the Muslim sample occupied a religious minority position (47%), while this is only for seven percent among the Roman Catholics and less than five percent in the sample of non-affiliated respondents. For Protestants and Eastern-Orthodox, this was around 1 in 5 individuals. All the respondents belonging to another smaller denomination were logically also members of a religious minority group. Those belonging to another denomination also had the highest mean score on the religiosity scale (0.9), followed by the Muslim (0.7), the Roman Catholic (0.6), and the Eastern-Orthodox (0.5). One in ten Muslims reported to have experienced discrimination based on their religion. Among those with another religion, this was eight percent, while for Roman Catholic, Eastern-Orthodox, and non-affiliated respondents, this is less than one percent.

Table 1 Descriptive results per personal denomination category
Full size table

Regarding the socioeconomic and demographic characteristic, the average age of the Muslim individuals was remarkable lower, and they were more likely to be married than the individuals of the other denominations were. On average, they were also less likely to be employed, received lower levels of education, and were more likely to be represented in the lower and moderate-income categories. Among the Muslim sample and those with another religion, almost 30% was a first-generation migrant. About 14% of the Eastern-Orthodox and Muslim sample was a second-generation migrant.

Results of Multilevel Analyses

Results in Model 1 of Table 2 show that the Protestants had a significant lower score on the depression scale compared to all the other religious denominations as well as compared to the sample that was not affiliated to a religious denomination. In addition, a higher score on the religiosity scale was related to more depressive feelings.

Table 2 Multilevel analyses on relationship between depressive feelings and religious denomination
Full size table

In the second model, the results show a significant moderation effect between the religiosity scale and a number of the individual denominations. This interaction effect is also depicted in Fig. 1. The differences in depression scores between the Protestant and Muslim samples (interaction term: − 0.165[0.060]), and between the Protestant sample and the ‘other’ sample (− 0.181[0.071]) reduced with higher levels of religiosity. The significant difference in depression scale between the protestant sample and people who are not affiliated to a specific denomination, however, increased with higher levels of religiosity (0.106 [0.037]).

Taking the religious minority status into account instead of the religiosity scale (Model 3, Table 3) did not moderate the differences in depression scores between the religious denominations in a significant manner. We find that members of a religious minority reported significantly more depressive feelings (0.192[0.034]) than respondents who were not a member of a religious minority. The regional majority denomination (Model 4, Table 3) was not related to individuals’ depressive feelings, but a number of the interaction effects between the regional majority denominations and being a member of a religious minority (Model 5, Table 3) were significant. This means that the relation between being a member of a religious minority and depressive feelings depends on the majority denomination of the region where the individual resided. Being a member of a religious minority group was significantly less depressive in a Roman Catholic majority region, a Muslim majority region, and in a region without a majority than in a region with a Protestant majority. As Fig. 2 shows, the depressive feelings in individuals who were not religious minority members were quiet similar across these regional majority denominations. At the same time, substantial differences in levels of depression could be established in religious minorities across these regional majority denominations. Most notably here is the estimated mean level of depression in religious minorities in Muslim majority regions; here we find that religious minorities reported lower levels of depression than inhabitants in that same region who were not a religious minority.

Table 3 Multilevel analyses on relationship between religious minority position, religious context, and perceived religious discrimination
Full size table

Model 6 (Table 3) additionally examines whether having a migration background and/or having perceived religious discrimination mediated the associations between depressive feelings and being a minority member in a specific religious context. Being a member of a religious minority in a Protestant majority context was no longer related to more depressive feelings than being a minority in a Roman Catholic majority region, nor in a region without a religious majority, after taking these control variables into account. The significant association between being a member of a religious minority and depressive feelings only held in the Muslim majority context (− 0.899 [0.267]). In addition, first-generation (0.297[0.034]) and second-generation migrants (0.250[0.036]) reported significantly more depressive feelings compared to individuals without a migrant background. Respondents who reported perceived discrimination based on their religion also reported significantly more depressive feelings (0.410[0.089]).

Discussion and Conclusion

Our study contributes to this literature by examining variations in levels of depression across and within the major religious denominations in Europe. We find that the Protestant population reports the lowest levels of depression, while the second lowest levels of depression are found in the Catholic population. In addition, we find that the non-affiliated population does not report the lowest levels of depression, which contrasts the vast majority of research on this topic (Dervic et al. 2004; Ellison 1991). The available research was, however, mainly restricted to the USA, that is, a more religious setting. Our finding shows that mental health benefits of being affiliated to a particular church are less straightforward in a less religious context, such as Europe. Highest levels of depression are found in the Muslim population, followed by the population affiliated to other smaller religions and the Eastern-Orthodox population. For the Muslim population, the higher levels of depression could be explained by the closely intertwined nature of mental health issues and religion, and the rigidity of the Islam when facing mental health issues. Taking care of one’s mental health and overcoming everyday life difficulties are seen as an individual responsibility, which is difficult to overcome without specialized care or support from the immediate environment—especially as it is highly stigmatized (Baasher 2001; Haque 2004; Husain 1998; Lester 2006). For the Eastern-Orthodox population, higher levels of depression are closely embedded in the lower levels of church attendance, interrelatedness with nationalism and the lesser extent in which they can personally shape their church (Wildmann 2012), providing less opportunities to use religion as a way to cope with their individual mental health problems.

Second, mental health differences between the religious denominations cannot be explained by differences in levels of religiosity between these groups, as significant differences hold when we adjust for levels of religiosity. In general, denominational differences in depression decrease among the more religious, with mental health differences between different denominations converging toward each other. Our results show that more religious people in general report more depressive feelings. In the current research, we measured religiosity through a composite index of three measures: frequency of service attendance, frequency of prayer, and religious salience. Previous research consistently shows that people who attend services more frequently also report fewer mental health problems (Schieman et al. 2013; Van de Velde et al. 2017). However, the relationship between prayer and mental health is less straightforward, with some studies showing that people who pray more frequently suffer more from mental health problems, while other studies provide evidence for the opposite direction. This association depends on the methods of prayer, as well as the perceived nature of divine relations or images, which shows substantial variation across denominations (Froese and Bader 2010). Future research should further entangle these specific interactions. Our finding may be further explained by a selection effect: People may be more inclined to turn to religious praxis in times of distress (Bradshaw et al. 2008). At the same time, it is much less likely to be selected into a particular denomination, especially within Christian heritage societies such as Europe, where religion is much more historically embedded in the cultural history of particular regions.

In contrast to the other denominations, within Muslim sample, more religious people report fewer depressive feelings. This variation due to religiosity within the Muslim sample may relate to the fact that higher levels of religiosity immediately have a larger impact on everyday religious practices. In the case of Muslims, these result in daily prayers five times a day, the active participation in social gatherings (e.g., festivities and the Ramadan), and in no substance use (Husain 1998). Furthermore, this could also relate to the higher levels of religiosity among Muslim migrant groups living as a minority in other European countries (Roy 2006). As high levels of religiosity serve as a protective factor against other stressors, such as discrimination, or serve as a feature on which interpersonal relationships are based on (e.g., McPherson et al. 2001; Okulicz-Kozaryn 2010, 2019), this finding could be due to a selection effect. Selection bias could occur as more religious or religiously visible Muslims could then experience higher levels of discrimination (Kunst et al. 2012), for which the benefits of being religious could also become more essential and make them more in need to search actively for coping mechanisms (Foner and Alba 2008). In this case, being more religious could result in a spiral due to negative encounters in the immigrant country and strengthen the feeling to hold on even more to their religious beliefs and practices.

Third, we find that the association between depressive feelings and being a member of a religious minority is moderated by the religious context: Being a member of a religious minority group is significantly less depressive in a region with a Roman Catholic majority, a Muslim majority, and in a region without a majority than in a region with a Protestant majority. The differences in levels of depression in religious minorities across these varying religious contexts can partly be attributed to migrant background or perceived religious discrimination. With the exception of religious minorities within a Muslim majority region, most of the differences between minority religious groups across different religious contexts do not hold when adjusted for these two factors. These findings are in line with a substantial body of research on mental health disparities in immigrants and ethnic minorities (Gkiouleka et al. 2018; Levecque and Van Rossem 2015; Missinne and Bracke 2012). In a similar line, these studies relate the health disadvantage in the migrant population to the distressing effects of the migration process itself, as well as to the precarious socioeconomic conditions and the experience of ethnic and religious discrimination in the country of residence. Our results additionally disprove a possible healthy migrant effect, which proposes a selective migration of healthy individuals. We, however, find that migrants report more depressive feelings than people without a migrant background do, and this health disadvantage is even more pronounced in the first than in second generation.

The significant difference in levels of depression between religious minorities in Muslim majority regions and Protestant majority regions, however, still holds after controlling for these factors. Within Muslim majority regions, we find that religious minorities in fact report lower levels of depression than individuals who are not a member of a religious minority. Perhaps, minorities within this setting benefit from engaging in homogeneous interpersonal networks within their religious communities, which may counteract the negative impact of other stressors related to occupying a minority position (McPherson et al. 2001; Okulicz-Kozaryn 2010, 2019). These results should, however, be interpreted with caution given the relatively small sample size of religious minorities in Muslim majority regions. In addition, with the exception of the Spanish Ceuta, these Muslim majority regions are all located in Kosovo and Albania. Both regions have been confronted with severe high levels of civil unrest and are among the poorest regions in Europe. While these conflicts were rooted in nationalism rather than religious intolerance (Musaraj 2013), the particularly higher levels of depression among religious minorities in these regions may nonetheless reflect the regions recent conflict-ridden and deprived circumstances, rather than a particular distressing position of religious minorities within a Muslim majority context.

Some additional limitations of our study need to be addressed. First, due to data restrictions, we were not able to distinguish between active and inactive members of a religious denomination. This might be of particular importance within the European context, where many individuals are formally members of a church, without actively participating in it. However, prevalence rates of active membership (thereby excluding the inactive members) based on the World value Survey are similar to the prevalence rates of our sample (May and Smilde 2016). Second, these data restrictions also limited us to only examine differences across the major religions in Europe, but thereby neglecting differences within these religious groups. However, previous research has shown that health behavior can vary considerably between different Protestant denominations (Van Tubergen et al. 2005). We may also expect similar differences within the other major denominations, especially if when a centralized leadership is not present. In the same line, the current study uses the homogenous cut off point of 50% to define the religious majority within a region. However, previous research showed that the mental well-being of religious minorities depends on the size of the religious majority. May and Smilde (2016), for example, find differences between religious minorities living in countries where the Catholic majority is 90% or more of the population and countries where the Catholic majority is between 51 and 89%. Unfortunately, we were unable to assess similar patterns due to small sample size at these higher cutoff points for religious majorities. An additional disadvantage of the data is that the surveys were conducted using the official national languages. This means that migrants who do not fully master one of the official languages of the country where they currently live are less likely to be interviewed. This probably resulted in a sample bias toward the more integrated groups of migrants (Van der Bracht et al. 2013). Finally, as with all other cross-sectional surveys, it is difficult to distinguish the cause from the effect. Religiosity and religious membership may help a person to cope with stress and therefore reduce the likelihood of developing depression. However, depressed people may also be more likely to turn to religion as a way of finding comfort in times of need or, in contrast, experience a loss in faith. We can, however, assume that changes in membership of a denomination are less likely than changes in religious behavior. At the same time, it cannot be ruled out that people join a specific denomination as a result of their depression, or renounce their membership due to a depression. When interpreting our results, it is therefore important to note possible reverse causation.

Nonetheless, our study contributes to the literature in several ways. First, our study showed that the established mental health patterns in the Protestant and Catholic population cannot be simply transcribed to the Eastern-Orthodox and Muslim population, which to date were underrepresented in empirical research. The same holds for the non-affiliated population, particularly within the European context. Second, we show that mental health in religious minorities is highly dependent on the religious context. Future research that ignores these contextual findings is limited in terms of generalization.