The influence of religious attachment on intended political engagement among lower-secondary students

Abstract

Religious attachment has been identified as an important correlate of civic participation, civic engagement, and civil participation among adults. This study investigates two aspects of relationships between religiosity and intended political engagement among lower secondary school students in 2009 and 2016. One aspect is the extent to which religious attachment is associated with an endorsement of the influence of religion in society. This can be viewed as the converse of secularity which asks for the separation of social and political institutions from religion. A second aspect investigated is the extent to which religious attachment is associated with expected adult electoral participation and expected adult active political participation after controlling for the effects of other characteristics. While the results from this study show no strong or consistent relationships between religious background and expected political participation among lower-secondary students, findings suggest that young people’s endorsement of religious influence in society depends strongly on their religious background and in turn shows associations with expected active political participation.

Background

This article addresses questions about religion and expected political participation among lower-secondary students. It presents results from analyses of rich comparative international data generated by two cycles of International Civic and Citizenship Education Study (ICCS 2009 and 2016) (Schulz et al., 2010, 2018). These data provided the basis for our investigation of the associations of religious background, and endorsement of the influence of religion in society, on expected electoral and expected active political participation as future adults. Our analyses also include associations of students’ endorsement of influence of religion and expected political participation with factors reflecting their social and personal background and attributes such as civic knowledge.

Recent surveys have shown that across most countries, and especially in North America and Europe, people view religion as having a diminished role in society. However, within this overall pattern, there were variations among countries and more religious people tended to favour a more important role for religion in society (Pew Research Center, 2019a).

There is an extensive literature concerning the extent to which religiosity is associated with civic engagement among adults (Ben-Nun Bloom et al., 2021; Pancer, 2015; Putnam & Campbell, 2010; Verba et al., 1995). An international survey, conducted across 26 countries, concluded that adults who are active members of religious organisations tended to be more engaged in civic life than other adults (Pew Research Center, 2019b). This conclusion is consistent with the conclusions of studies conducted in only one national context (Guo et al., 2013; Lundasen, 2021; Pancer, 2015; Putnam & Campbell, 2010; Verba et al., 1995). Even after controlling for many other variables, religious tradition and attendance of religious services tend to be related to indicators of civic participation (Storm, 2015). Many of these studies take a broad view of civic participation which includes voluntary contributions to, and participation in, community activities (e.g. through volunteering or involvement in community service) as well as political engagement. These differences have consequences for interpretations of relationships between religiosity and civic engagement.

While many of the studies looking at the nexus between religion and civic engagement are concerned with adults, there are also some studies that have focused on secondary school students. Perks and Haan (2011) found that involvement in a religious organisation among Canadian youth predicted measures of adult community participation. Ekström and Kvalem (2013) reported a relationship between belonging to a religion and political engagement among high school students in Sweden. Pancer (2015) presented evidence that schools and neighbourhoods may contribute to both civic engagement and religious formation among adolescents. Vermeer (2010) viewed religious education at schools as a contributor to socialising young people in ways that had civic value. Francis et al. (2015) regarded church attendance and education about religion at school as factors that nurture tolerance in a religiously diverse society, arguing that engagement with religion could also be viewed as an important part of a broader civic engagement.

Several interpretations of the association between religiosity and civic participation have been suggested. One interpretation is that religious participation encourages adherents to consider features of society (a world view) that they see as aligned with their beliefs and to seek to promote those features through political participation (Campbell, 2004; Jones-Correa & Leal, 2001; Pancer, 2015; Putnam & Campbell, 2010). Pavolini et al. (2017) review the ways in which religious background can influence attitudes to social policy at various levels. Specifically, they argue that, for individuals, being religious (measured by attending services regularly), appears to be a crucial influence on preferences and behaviours towards social welfare provision. Similarly, van Oorschot (2006), based on analyses of data from the European Values Survey, found that people who attended church more frequently were less conditional in their solidarity towards groups in need of welfare. Religious background, even religious attendance, is a broad category that can include a range of religions exercised within various national contexts. That range of religions encompasses varied emphases on the extent to which there is an endorsement of the roles of religion in society and the ways in which those roles are enacted (Castles, 1994).

A second interpretation of the association between religiosity and civic participation is that civic skills and knowledge, including communication and organisational capacities, are acquired early in life, and are developed through institutional settings such as religious organisations as well as schools (Verba et al., 1995). However, this interpretation is contested as some studies reported negative associations between religious participation and levels of political knowledge or political efficacy (Scheufele et al., 2003).

A third interpretation is that religious organisations provide networks focused on political recruitment and motivation (Pancer, 2015; Putnam & Campbell, 2010). This appears to apply to civic engagement through activities such as volunteering, which may take place in religious charities, rather than political engagement. Research among US adolescents (Porter, 2013) indicated that moral identity may be positively associated with voluntary service and expressive political involvement but negatively related to traditional forms of political involvement. There are also suggestions that the strength of the association between religious participation and civic engagement may depend on contextual factors such as the level of religious attachment in society and the extent to which a political system is secular with clear separation between religion from social and political institutions (Zuckerman & Shook, 2017). Further, religious attachment may possibly play different roles in secular societies compared to contexts where religion and politics tend to be more intertwined.

In this study we focus on expected political participation among lower-secondary students. We build on some of the findings reported in the research literature in pursuit of our underlying research questions that ask about the relationship between religiosity and expected political engagement among young people in lower secondary education. Our analyses include consideration of the extent to which respondents endorse the implementation in society of policies related to religious beliefs and principles. This was one of the interpretations suggested in the literature of how religious attachment might be related to political engagement. Our review of the literature also identified an argument that civic knowledge and skills might be developed through religious participation as well as school education, and that these might be related to political participation. Our investigation includes measures of knowledge of civic principles and practices as well as experiences with civic participation at school.

We distinguish between expected electoral participation, which involves voting and associated formal activities, and expected active political participation, which involves participating in political parties or organisations, being a candidate, or supporting candidates in elections. Based on previous research we anticipated different influences on these two forms of participation.

There is evidence in the research literature of several other factors related to electoral and active political participation. Therefore, we investigated the concomitant influence of citizenship beliefs such as citizenship self-efficacy, trust in civic institutions, and civic engagement at school (Schulz, 2023) and background factors such as gender, socioeconomic background, and school location (Schulz et al., 2018) as well as parental interest in political issues (Lauglo, 2016).

In recognition of the importance of context, especially the varying levels of attachment to religion across national contexts, our analyses are conducted separately in each country but also include reviews of average results across all participants in this study.

Research questions

From our evaluation of the research literature, we identified several research questions that provided foci for our study.

• 1. To what extent is religious background associated with an endorsement of the influence of religion in society, after controlling for the associations of this endorsement with social and personal background, school-related factors, and general citizenship beliefs.

• 2. To what extent is religious background associated with expected electoral and expected active political participation as future adults, after controlling for associations of these forms of expected political participation with social and personal background, school-related factors, and general citizenship beliefs.

• 3. To what extent does endorsement of the influence of religion in society predict expected electoral and expected active political participation as future adults, after controlling for associations of these forms of expected political participation with social and personal background, school-related factors, and general citizenship beliefs.

Data and methods

The first two cycles of the International Civic and Citizenship Education Study (ICCS 2009 and 2016) provided data that allow comparative analyses of civic education outcomes (civic knowledge, attitudes, and behaviour-related measures) with contextual data at the level of education system, schools, and students (Schulz et al., 2010, 2018). In both cycles the student questionnaire included an international option on religious affiliation and engagement (religious attachment), as well as on students’ endorsement of religious influence in society (rejection of secularity) that was administered in most participating countries.

Data

The research for this study was based on data from 13 countries participating in ICCS 2016 that included all questions pertaining to the international option regarding religion. The target population for ICCS 2016 included students in their 8th year of formal schooling (Weber, 2018). It further makes use of data from students of the same target population that participated in ICCS 2009 for eight of those countries.¹ Parallel analyses were conducted of the 2016 and 2009 data. We only included only data from those countries that had met the IEA sample participation requirements (Weber et al., 2018) and included records from 52,403 students enrolled at 2003 schools in ICCS 2016, and 31,594 students from 1236 schools in ICCS 2009. Data from the Dominican Republic and Estonia were also available but excluded because fewer than 70 percent of their students had valid data for the multivariate analyses after we had applied a list-wise exclusion of missing data. Results from countries where data for multivariate analyses from one or both surveys were available for less than 80, but more than 70, percent of their students were annotated in the reporting tables.

Variables and measures

The study makes use of two types of indices: simple indices and scale indices. Simple indices were based on direct information provided in response to a question (e.g. a student’s stated affiliation with a religion) or a transformation of information provided (e.g. whether they attended a religious service on a monthly or more frequent basis). Scale indices were derived through the scaling of several related items. Typically, exploratory and confirmatory factor analyses were used to examine cross-national measurement equivalence (Schulz et al., 2018). Item Response Theory (IRT) scaling was applied to derive scales related to expected engagement and other variables expected to be related to expected engagement (Schulz & Friedman, 2018).

For this study, we selected simple and scale indices that were identified as relevant to our underlying research question. The following sections describe the measures that were included in this study encompassing both outcome variables and factors expected to explain the variation in those outcome variables.

Political participation

Measures of expected electoral participation and expected active political participation were based on students’ ratings of their expectations to get involved in different ways that adults can take an active part in political life.² Three items were used to construct the scale expected electoral participation (“vote in local elections”, “vote in national elections”, and “get information about candidates before voting in an election”) and five items were used to construct the scale expected active political participation (“help a candidate or party during an election campaign”, “join a political party”, “join a trade union”, “stand as a candidate in a local election”, and “join an organization for a political or social cause”). Higher scores for these two scales indicated greater expectancy of participation in the respective activities. The average scale reliabilities (Cronbach’s alpha) across all ICCS 2016 countries were 0.83 and 0.85. The scales were equated across ICCS 2009 and ICCS 2016 and transformed to an original reporting metric with a mean of 50 and standard deviation of 10 for all equally weighted countries that participated in ICCS 2009. For the multivariate path analyses where both scales were used as dependent variables, we used nationally standardised scores with average values of 0 and standard deviations of 1 within each country.

Influence of religion in society

The scale measuring endorsement of the influence of religion in society was a central mediating variable between religious attachment and the two political participation measures. It was based on the level of agreement³ that students indicated in relation to a set of six statements concerned with the notion that religion should play an important role in shaping society (“religion is more important to me than what is happening in national politics”, “religion helps me to decide what is right and what is wrong”, “religious leaders should have more power in society”, “religion should influence people’s behaviour towards others”, “rules of life based on religion are more important than civil laws”, and “religious people are better citizens” (the last item was only included in ICCS 2016). Higher scores on this scale indicated stronger agreement with the notion that religion should play an important role in shaping society. The average scale reliability (Cronbach’s alpha) across ICCS 2016 countries was 0.87. The scale was equated across ICCS 2009 and ICCS 2016 and set to an original international reporting metric (used in Table 1) with a mean of 50 and standard deviation of 10 for all equally weighted countries that participated in ICCS 2009. For the multivariate path analyses, in which religious influence was both an independent and dependent variable, we transformed scores to a nationally standardised metric with average values of 0 and standard deviations of 1 within each country.

Student background

Student background was included through gender, school location, socioeconomic background, and parental interest in political and social issues. Gender (female) was coded dichotomously (female = 1 and male = 0) based on student responses and school location was coded dichotomously (urban = 1 and non-urban = 0) based on information provided for the school. Socioeconomic background was measured using a composite indictor from student-reported parental occupation and education (the highest levels of parental occupation and education were used wherever data were provided for two parents), and the number of books at home (Schulz & Friedman, 2018, pp. 151–152). Scale scores for socioeconomic background were nationally standardized to have averages of 0 and standard deviations of 1 in each country. Higher scores on this indicator corresponded to higher status occupations, higher levels of education and a greater number of books in the home. Parents’ interest in political and social issues was based on student reports of parental interest (using the highest level of interest where data were reported for two parents) and included as a dichotomous variable (with a value of 1 for students who reported that at least one of their parents was quite or very interested and a value of 0 indicating no or little parental interest).

School-related factor

Civic knowledge was based on a test concerned with understanding four domains of civics and citizenship: civic society and systems, civic principles, civic participation, and civic identities (Fraillon, 2018). In ICCS 2016 the measurement of civic knowledge was based on 87 test items, 42 items of those items had already been included in ICCS 2009 and were used to equate the test scores from ICCS 2016 with those from the first ICCS survey in 2009.⁴ In the analyses for this study we used all five plausible values transformed into a standardised metric set to have national averages of 0 and national standard deviations of 1.

We also included civic participation at school in our analyses. While this variable was not central to our research questions, it had been found to be related to expected political participation (Schulz et al., 2018), and it was included as a control variable in our analyses. This variable was measured based on six items reflecting past or current participation in civic activities at school (“active participation in an organised debate”, “voting for a class representative or school parliament”, “taking part in decision making about how the school is run”, “taking part in discussions at a school assembly”, “becoming a candidate for class representative or school parliament”, and “participating in an activity to make the school more environmentally friendly”).⁵ Scale scores were derived using IRT scaling (nationally standardized scores with averages of 0 and standard deviations of 1) with higher scales scores indicating higher levels of civic participation at school. It had only marginally satisfactory reliability across ICCS 2016 countries (average Cronbach’s alpha = 0.67).

Citizenship beliefs

Two variables that reflected citizenship beliefs were also included as control variables because of previous findings about their associations with expected political participation. Citizenship self-efficacy reflected how well students believed they could undertake seven different civic-related activities (“discuss a newspaper article about a conflict between countries”, “argue your point of view about a controversial political or social issue”, “stand as a candidate in a school election”, “organize a group of students in order to achieve changes at school”, “follow a television debate about a controversial issue”, “write a letter or email to a newspaper giving your view on a current issue”, and “speak in front of your class about a social or political issue”).⁶ IRT scaling was used to develop a scale with a high reliability (Cronbach’s alpha averaged 0.84 across ICCS 2016 countries). Trust in civic institutions referenced six institutions: national government, local government, national parliament, police, courts of justice, and political parties.⁷ IRT scaling was used to derive a scale reflecting students’ trust in civic institutions. This IRT scale had high reliability on average across ICCS 2016 countries (Cronbach’s alpha = 0.85), and, for this study, standardised scores were used for both measures set to have averages of 0 and standard deviations of 1 within each country.

Religious background

Three indicators of religious background were included in the analyses. The first was religious affiliation based on student responses to the question “what is your religion” that were coded dichotomously (1 = named religion stated and 0 = no religion indicated). The second was religious attendance based on question about how often students attended a religious service outside their home with other people, scored as having done this at least once each month (coded 1) or less frequently (coded as 0). The third indicator was derived from student responses indicating that they had been involved in activities of a religious group, where responses indicating any previous involvement were coded as 1 while others were coded as 0.

Analytic methods

ICCS 2009 and 2016 employed two-stage cluster sampling procedures (Weber, 2018; Zuehlke, 2011). In the first stage, schools were sampled in each country from a sampling frame with a probability proportional to size. In the second stage, intact classrooms were randomly sampled within schools. Although this approach supports efficient data collection, the resulting samples are not simple random samples. Therefore, the jackknife repeated replication technique (JRR) was used for all analyses to obtain appropriate sampling errors for population means, percentages, regression coefficients, and any other population estimates (Schulz, 2018).

Two types of analysis were used in this investigation. The first analyses were comparisons of national score averages reflecting endorsement of the influence of religion in society and percentages of indicator variables of religiosity across countries, and also the first two cycles (ICCS 2016 compared to ICCS 2009) for countries that had participated in both surveys. The second analyses employed path modelling (see, for example, Kaplan, 2009) conducted with the MPLUS (version 7) software package, which allows using the jackknife (JK2) procedure for estimating sampling variance for data from complex samples as well as the measurement variance based on the five plausible values reflecting students’ civic knowledge (Muthén & Muthén, 2012). These analyses were conducted for each cycle in each country using regression coefficients to represent the strength of each relationship.

The analytic model (Fig. 1) postulates that the three criterion variables (endorsement of religious influence in society, expected electoral and active political participation) may be potentially influenced by factors that belong to four groups:

• Social and personal background (gender, socio-economic status, school location, and parental interest in political and social issues);

• Religious background (religious affiliation, attendance of religious services, and participation in religious groups);

• School-related, or school-developed factors (civic knowledge is central to our research question, but we also included civic participation at school as it has been identified as a predictor of future engagement); and

• Citizenship beliefs (citizenship self-efficacy and trust in civic institutions), which were included as control variables known to be associated with expected political participation.

Fig. 1

Path model of influences on expected electoral and active political participation among lower secondary school students

Not all these factors were expected to influence all three of the criterion variables, decisions about assuming paths from predictor to criterion variables were based on exploratory analyses or prior research findings (see, for example, Schulz et al., 2018). Beliefs (citizenship self-efficacy and trust in institutions) were identified as important predictors of expected political engagement, the same holds for civic participation at school as one of the school-related factors. Civic knowledge is assumed to be related to all three criterion variables (depicted with an individual arrow connecting it to the endorsement of religious influence).

In addition, endorsement of religious influence is modelled as a predictor of the two forms of political participation as well as being influenced by the other sets of variables. In this sense it is seen as a mediating variable.

To reduce the complexity of estimating this model across a range of different countries, the path model is based only on manifest indicators. Given the mostly high reliabilities for scales used in this study, in the case of variables that represent latent variables, we used the IRT scales without specifying the measurement model for each latent factor. Models were estimated using maximum likelihood (ML) estimation for each national sample separately, and average results with their corresponding standard errors were also computed to provide findings at the level of the combined study. As civic knowledge is represented by five plausible values, the multiple-imputation procedure provided by MPLUS was applied to incorporate the measurement error for this latent variable.

Table 10 in the annex shows model fit indices⁸ for the path model for each national samples as well as on average. The RMSEA and SRMR both suggested an acceptable fit (values below 0.08) on average and for all national samples. The CFI also indicated good fit (value > 0.95) on average and in almost all countries except Peru.

Results

National and between-cycle comparisons of religious background

Table 1 records for 2009 and 2016 the national mean scores in 2009 and 2016 students’ endorsement of religious influence in society and average percentages indicating a religious affiliation and monthly attendance of religious services. Those results show considerable variation among countries for each cycle of ICCS. In ICCS 2016 the mean percentages declaring a religious affiliation ranged from 41 percent in the Netherlands to 95 percent in Croatia. In ICCS 2009 the range was from 64 percent in Belgium (Flemish) to 92 percent in Malta. For countries that had participated in both cycles of ICCS there was an overall drop of 1.5 percentage points in the proportion of students stating a religious affiliation.

Table 1 National results for students’ endorsement of religious influence in society, religious affiliation, and attendance of religious services in 2009 and 2016

Percentages of students indicating at least monthly attendance at a religious service, in ICCS 2016, ranged from 13 percent in Belgium (Flemish) to 78 percent in Croatia. In ICCS 2009 monthly attendance at a religious service ranged from 12 percent in Belgium to 70 percent in Malta. In ICCS 2016 the between-country correlation coefficient for religious affiliation with monthly religious service attendance was 0.72. Across countries, there were slightly lower percentages of students reporting at least monthly attendance of religious service (− 1.1 percentage points), however, changes were not consistent across countries.

Table 1 also shows variation across national score averages for students’ endorsement of religious influence in society. In ICCS 2016 the lowest average was 44 in the Netherlands and Norway and the highest was 56 in Peru followed by Colombia and Croatia recording mean scores of 55. This represents a range of more than one standard deviation (1.2 standard deviations between the Netherlands and Norway, and Peru). In ICCS 2009 there was a similar range from 45 in Belgium (Flemish) to 55 in Malta. In ICCS 2016, the between-country correlation coefficient for endorsement of religious influence with religious affiliation was 0.83 and with monthly religious service attendance it was 0.81. The results also illustrate that, between 2009 and 2016, there was a (statistically not significant) drop of just under one scale point (equivalent to about one tenth of an international standard deviation) in the international scale score for endorsement of religious influence in society based on the nine common countries.

Relationships with endorsement of religious influence in society

Our analytical model (Fig. 1) represents a range of relationships between social and personal background, religious background, and civic knowledge (as well as the further control variables reflecting civic engagement at school and citizenship beliefs), and students’ endorsement of religious influence in society and expected political participation. Table 2 records the coefficients representing relationships between endorsement of religious influence in society, three background factors (gender, socioeconomic status, and school location) and the school factor, civic knowledge within each country as well on average across national samples.

Table 2 Civic knowledge and background factors as predictors of students’ endorsement of religious influence in society

It is evident that there is a moderate negative relationship between civic knowledge and endorsement of religious influence in society, which was observed in all countries and in both cycles of ICCS. In 2016 the average value of the coefficient for common countries was − 0.29 and in 2009 the average value was − 0.23. The difference between these values suggests that the negative relationship was slightly stronger in 2016 than in 2009. A simple interpretation of this association is that students who have a greater understanding of civic principles are less inclined to endorse religious influence in society. Moreover, the strength of the relationship did not appear to differ between countries with high and low levels of religiosity (measured as religious affiliation or participation in religious services shown in Table 1). The data in Table 2 also show only small and inconsistent associations of the background factors gender, socioeconomic status, and school location with endorsement of religious influence in society.

Table 3 records coefficients representing relationships between religious background and endorsement of religious influence in society. Those coefficients show strong and consistent associations of endorsement of religious influence in society with religious affiliation (coefficients averaging more than 0.7 across countries in both cycles) and moderate but also consistent associations with religious service attendance (with averages of about 0.4) and participation in a religious group (with averages of about 0.3).

Table 3 Religious background factors as predictors explaining students’ endorsement of religious influence in society

These patterns suggest that religious affiliation was strongly associated with endorsement of religious influence in society. In comparison, indicators of religious practice (service attendance and participation in a religious group) were only moderately associated with endorsement of religious influence in society. This difference suggests that religious affiliation (stating that one adheres to a religion) carries with it an endorsement of religious influence in society possibly through a belief that religious principles should apply in society. In other words, religious affiliation might be associated with advocacy of features of society seen as compatible with their religion (e.g. “religious leaders should have more power in society”, “religion should influence people’s behaviour towards others”), whereas religious practice might reflect a view that religion should focus on personal behaviour (e.g. “religion helps me to decide what is right and what is wrong”, “rules of life based on religion are more important than civil laws”).

In our view, these differences raise issues that deserve further research that should consider this as well as other study results. For example, using survey data from the Netherlands, Van Tubergen (2007) found that social integration led to a decrease in religiosity among immigrants. Aksoy and Wiertz (2024), based on analyses of panel study data in Britain, provided evidence about differences among religions in the extent of social involvement of religious adherents. It may be that there are also differences within religions in terms of outward-looking (extrinsic) and inward-looking (intrinsic) beliefs (see Tiliopoulos et al., 2007).

Relationships of background and school factors with expected electoral participation

Table 4 records coefficients representing relationships of expected electoral participation with religious, personal, and social background. While there were only small associations in a few countries of expected electoral participation with religious affiliation or reported participation in religious groups in the community, gender, and socioeconomic status, the results reveal moderate but consistent association with parental interest in political and social issues is evident in all countries and across both cycles. This latter finding is consistent with previous research (Lauglo, 2016; Schulz et al., 2010, 2018).

Table 4 Religious affiliation and religious group participation and other background factors as predictors of students’ expected electoral participation

Table 5 records coefficients representing relationships of expected electoral participation with and endorsement of religious influence, civic knowledge, citizenship self-efficacy, trust in civic institutions, and civic participation at school. Even though we found statistically significant positive associations of expected electoral participation with endorsement of religious influence in society in almost half of the countries, these were only relatively small and inconsistent (ICCS 2016 average of 0.04). The results show, however, consistent moderately strong relationships (on average 0.31 across ICCS 2016 countries) with civic knowledge indicating that understanding civics and citizenship is associated with electoral participation. Slightly smaller but still consistent associations with the control variables citizenship self-efficacy (0.19) and trust in civic institutions (0.20) are also evident, while relationships with civic participation at school were very only relatively small and inconsistent. The results from ICCS 2009 were overall very similar to those from ICCS 2016 with ICCS 2009 average coefficients for the relationship between expected electoral participation and endorsement of religious influence in society recorded as 0.03 and with civic knowledge as 0.26.

Table 5 Civic knowledge and endorsement of religious influence as predictors of students’ expected electoral participation, controlling for school civic participation and citizenship beliefs

Relationships with expected active political participation

Table 6 records coefficients representing relationships of expected active political participation with religious and other background factors. In 2016, there were moderate negative associations of expected active political participation with religious affiliation in Croatia and Malta (both countries with high levels of religiosity) and small negative associations with religious affiliation in a few other countries. There are only small but positive associations in a few countries with religious participation, and we found mostly smaller negative associations of expected active political participation with gender in eight countries in 2016 and seven countries in 2009. This can be interpreted as indicating that female students recorded lower expected active political participation than male students, but the association was not consistent across countries. While socioeconomic status had no consistent relationships with expected active participation, we found consistently positive associations for parental interest. When comparing results from ICCS 2016 with those from ICCS 2009, they were generally similar. However, there were more instances of a negative association with religious affiliation, and somewhat larger and more frequent positive associations with students’ participation in religious groups.

Table 6 Religious and other background factors as predictors of students’ expected active political participation

Table 7 records coefficients representing relationships of expected active political participation with endorsement of religious influence in society and civic knowledge as well as the control variables citizenship self-efficacy, trust in civic institutions, and civic participation at school. We found mostly positive but relatively small associations (on average across ICCS 2016 countries 0.11) of expected active political participation with endorsement of religious influence in society in all countries. Civic knowledge was negatively related to expected active political participation albeit to a relatively small extent (ICCS 2016 average of − 0.12). Among the control variables, citizenship self-efficacy and trust in institutions had consistent positive associations (on average, respectively 0.29 and 0.14), while participation in civic activities at school had weak positive associations related to expected active political participation (0.07) in most countries. Results from the previous cycle in 2009 were similar to those from ICCS 2016.

Table 7 Endorsement of religious influence in society, civic knowledge, and other factors as predictors of students’ expected active political participation

Associations between participation variables and variance explanation

Our model in Fig. 1 explained, on average across ICCS 2016 countries, 30 percent of the variance in endorsement of religious influence in society, 26 percent of the variance in expected electoral participation and just 14 percent of the variance in expected active political participation (Table 8). The results also show that there were only moderate correlations between expected electoral participation and expected active political participation (on average across countries: r = 0.22). The proportions of explained variance and correlations from ICCS 2009 were of similar size compared to those from ICCS 2016.

Table 8 Percentages of explained variance by model and correlations between expected electoral and active political participation

Summarising results across the two ICCS cycles

Table 9 provides a comparative summary of the average path coefficients in ICCS 2016 and ICCS 2009 using only the results from the same nine countries with data from both cycles. The results show that across the two surveys there was a consistent positive relationship between religious background factors and endorsement of religious influence in society. Further, students with higher levels of civic knowledge tended to be less likely to endorse religious influence in society. Endorsement of religious influence in society tended to be more associated with expected active political participation than with expectations to engage in elections.

Table 9 Comparison of average path coefficients in ICCS 2016 and ICCS 2009 (common countries)

Conclusion

At the outset we looked at the extent of religious affiliation, religious practice, and endorsement of religious influence in society as well as at changes between the two first cycles of ICCS in 2009 and 2016. We observed differences across participating countries on these three characteristics. In some national contexts majorities of students saw themselves as part of a religion and reported attending at least once a month religious services while in other countries fewer than half of their young people identified with a religion. Results from comparisons across the first two cycles suggest a slight decrease in religious affiliation and attendance of religious services on average across those countries that participated in both cycles.

Our research questions focused on the influence of religious background on the endorsement of the influence of religion in society and expected political participation (in the form of both electoral participation and active political participation) as well as the relationship between endorsement of religion in society and political participation.

In relation to our first research question, endorsement of religious influence in society was most strongly related to religious affiliation but only moderately related to religious service attendance and participation in a religious group. From our analysis of ICCS 2009 and 2016 data we concluded that endorsement of religious influence on society was associated with religious background (especially to religious affiliation) and appeared to be higher in countries with higher proportions of students expressing a religious affiliation. Social and personal background factors were not associated with endorsement of religious influence on society. However, knowledge and understanding of civic principles and practices was negatively related to endorsement of religious influence on society. This relationship was evident in all countries and warrants further investigation. It may be that higher levels of civic knowledge provide young people with a better understanding of secular principles and dissuade them from viewing society solely in terms of their religious beliefs and principles.

In relation to our second research question, we distinguished between associations with expected electoral participation and expected active political participation, which were only weakly correlated with each other. We did not observe any consistent associations between our measures reflecting religiosity (religious affiliation and participation) and expected electoral participation. However, religious affiliation appeared to be related to expected active political participation to a small to moderate extent in almost half of the countries that participated in ICCS 2016. In some countries, there were also weak but significant associations between religious group participation and expected active political participation.

Our third research question concerned the role of endorsement of religious influence in society on expected political participation. We found that endorsement of religious influence in society was related to expected active political participation to a small but consistent extent. In ICCS 2016, the effects of endorsement of religion in society on expected active political participation were relatively modest but were statistically significant in all countries. In ICCS 2009, the coefficients were of similar magnitude and statistically significant in all but two countries.

Overall, our conclusion is that religious background is not strongly or consistently directly related to expected electoral or active political participation. However, we also found that endorsement of the influence of religion in society was related to religious background and, to a small extent, also to expected active political participation. This may suggest a transmitted influence of religious background on endorsement of the influence of religion in society through to expected active political participation. However, we found that endorsement of religious influence in society was not consistently related to expected electoral participation, which represents formal but limited political participation.

As already shown in earlier research of ICCS data (see Schulz et al., 2018), knowledge and understanding of civic principles and practices was negatively related to expected active political participation. This suggests that in terms of secular influences, as well as religious influences, it is beliefs rather than civic knowledge that portend active political participation (Veugelers, 2021). One may conclude that active political participation appears to be driven by ideals rather than knowing how civic and political systems work. This interpretation is consistent with the lack of an association of religious background with expected electoral participation which is constrained to voting as a formal aspects of civic participation.

It is important to acknowledge that with data from cross-sectional surveys such as ICCS it is not possible to make inferences to causality. While the models used in the analyses presented in this article, assume criterion and predictor variables, results from such models only reflect net associations between variables that may interpreted as indicative but not as proof of underlying causal relationships. It would be ideal to pursue the nexus between religiosity and political engagement in a longitudinal perspective on how political engagement and religiosity as well as the relationship between these factors develop throughout adolescence.

Annex

See Table 10.

Table 10 Model fit indices