Religion, identity, and party preference: the role of Catholicism in the 2022 Italian national elections

Abstract

This paper contributes to the debate on the role of religiosity in party competition in democratic political systems by analyzing the role of religious practice and belonging in the 2022 Italian elections. By using the newest Italian National Election Studies dataset, we combine indicators of both religious affiliation and practice to show how the electorates of the main Italian parties are composed in terms of voters' religiosity, to highlight how these differences influence the probability to vote for each of those parties and to further investigate the relationship between religious affiliation/practice and voting behavior. Results show the emergence of an increasingly identity-based relationship between religiosity and vote within the Italian context. We find that religiosity – in the form of Catholic affiliation – maintains a direct effect on vote choice which is also partially mediated by attitudinal indicators, such as those toward immigration, homosexuality, and abortion. We conclude that party competition in Italy is mainly fueled by identity dynamics – and less on religious practice or beliefs – concerning the whole group of those who identify as Catholics opposed to the group of those who do not.

Introduction

The relationship among religiosity, politics, and party preference has always been a crucial issue for both political sociology and electoral studies. Even though many studies have claimed that the relation between religious identities and party choice has become increasingly weaker in the last few decades (Wellhofer and Ignazi, 2011), the role of religious beliefs and practice for voting behavior remains a prolific topic for at least two main reasons.

First and foremost, for a contextual issue. If secularization changed how people deal with religion and develop their own beliefs and practice, especially in the West, this process can advance at a different pace in different institutional and political contexts. In this regard, there is evidence that Italy is on the same path of diffuse religious decline as the other European countries, but it also went through a phase of religious stability at the turn of the 1980s and 1990s (Vezzoni and Biolcati-Rinaldi, 2015). The impact of this peculiar pattern on systems of belief and practices – and party preferences of course – is still underexplored.

Second, for a party competition reason. The relationship between religiosity and vote choice has returned to the fore concurrently with the electoral results obtained by radical right parties (RRPs) throughout Europe. Such a success has progressively called for a reconceptualization of the effect of religiosity on voting behavior, because, if the radical right family is tightly connected to conservative and traditionalist issues for sure, religious voters tend not to vote massively for RRPs, or at least most of regular churchgoers tend to prefer more moderate alternatives (Immerzeel et al., 2013). However, newer research shows an increase of religious references within RRPs' political programs, as they use these references under a nativist frame, shaping an “us vs. them” rhetoric against immigrants, especially against Muslims (Schwörer and Fernández-García, 2021).

In conclusion, the whole breeding ground of political competition seems to have changed, and the classic distinction between religious and secular voters, which purportedly gave birth to confessional parties and fueled party identification for many years, seems to have gradually given way to a newer conflict between traditionalists and liberals (Bornschier, 2010).

Given these considerations, our proposal contributes to the debate by analyzing the role of religion in the latest Italian elections under a new light. We take a step further from the classic account, which focuses mostly on the role of religious practice, and explore the role of “nominal Catholicism” more explicitly. In the Italian context, this phenomenon partially overlaps with that of “cultural Catholicism” and refers to people who identify as Catholics even though they do not practice their faith regularly nor coherently follow religious commands; still, they cling to religion as a repository of traditions and values (see, among others, Garelli, 2020). By referring to this category, we combine indicators of both religious affiliation and practice, which allows us to overcome a unidimensional concept of religiosity and also to investigate more identity and culture related aspects of religiosity. Such a role of nominal Catholicism has been recently analyzed in relation to political attitudes toward immigration, LGBT issues, abortion, and other ethical topics with interesting results (cf. Ladini et al., 2021).

Our research aims at assessing whether being part of this category holds meaningful direct effects on vote choice or on the development of attitudes that may place voters closer to specific parties. To do so, we rely on the newest Italian National Election Studies (ITANES) dataset and structure our research along three main steps. First, we show how the electorates of the main Italian parties are composed in terms of voters' religiosity, highlighting similarities and differences among parties and how these differences influence the probability to vote for each of those parties. Second, we further investigate the relationship between religious affiliation/practice and voting behavior. Finally, we test whether religiosity holds a direct effect on vote choice, or its effect may be mediated by attitudinal indicators. Results show the existence of a more identity-based relationship between religiosity and vote, yet in a way which is not related to the opposition between nominal and practicing Catholics, but it encompasses the whole category of religiously affiliated. In a nutshell the Italian context seems to be fueled by identitarian dynamics concerning the whole group of those who identify as Catholics opposed to the group of those who do not.

Religiosity and voting behavior

The slow decline of the relationship between religiosity and voting behavior

Religious identity has always been a major driver of political conflict, and the interplay between religion and voting behavior has shaped political science literature for decades. From the 1960s, the traditional religious cleavage has represented one of the main focuses of scholars' interests. On the one hand, religious beliefs triggered the rise of confessional parties whenever the latter could politicize religious conflicts representing religious against secular interests (Lipset and Rokkan, 1967). On the other, but also consequently, religion was regarded as one of the most influential factors of voting behavior, if not the main one in specific contexts where it showed even greater effects than social class divisions (Rose and Urwin, 1969). Still, even though religion maintains an impact on political behavior, the magnitude of the effect varies across contexts; and, when we narrow it down to Western Europe, the cruciality of religiosity to explain vote choice appears to have weakened consistently (Broughton and ten Napel, 2000).

One of the more celebrated theories to explain this variability and the widespread lowering intensity of religiosity is the secularization theory. Intended as a process by which increasing sectors of society cease to be dominated by religious institutions and symbols (Berger, 1973), secularization theory purports that the increasing social modernization and rational thinking decrease the need for religion, consequently shrinking the scope of religious authorities (Chaves, 1994). Strictly connected with this, Western Europe has been facing a significant and constant decline in Christian belief and practice since the end of World War II (Gorski, 2003; Voas, 2009; Molteni and Biolcati, 2023), which is only partially counterbalanced by what Casanova (1994) – among others – defines as the “privatization of religion,” meaning that religion itself can remain relevant in the private rather than in the public sphere.

Therefore, the spread of a secular society and the consequent privatization of religion stand as a common trait for at least Western European countries, with all due contextual differences affecting the evolution of the process. However, when scholars dealt with the religious cleavage, they tried to explain its decline (and the decline in influence of all cleavages, broadly speaking) using different perspectives. Bottom-up explanations underline the cruciality of factors like modernization, globalization, and individualization in weakening societal and party cues, thus marking the decline of cleavage influence in voting behavior. This process can be behavioral when it assumes the impairment of a formerly strong bond between a social group and a political party, or it can be structural when it assumes a collapse in size of a certain social group (Lachat, 2007). Still, both processes claim the cruciality of societal changes in enforcing newer and more flexible voting patterns based on policy preferences, spatial evaluation, and cognitive mobilization (cf. Franklin et al., 2009; Dalton, 2013). Other explanations adopt a top-down perspective by giving prominence to political changes. In this perspective, it is party strategy or structural systemic changes that trigger a decline or reassessment of cleavage politics (cf. Przeworski, 1986; Elff, 2007). Consequently, voters simply react to the changing structure of political choice and to the stimulation and mobilization strategies of political elites, as notable works on class-cleavage suggest (Evans and Tilley, 2017; Heath, 2018).

In sum, whenever we face a deep change in political behavior, we can rely on two main approaches depending on which side of the coin we find prominent. Therefore, the decline of religious voting can be explained as the consequence of the decrease of religious voters but also as the consequence of strategic or systemic changes in the supply side. Both approaches have succeeded in explaining vote changes in different contexts, however, when we narrow it down to the Italian scenario, scholars convincingly explained the decline of the religious cleavage by focusing on a sudden change of political supply. Although scholars did not always find systematic evidence for top-down explanations in comparative studies (cf. Goldberg, 2020), notable case studies showed that the sharp decline of religious voting in Italy was not solely due to the visible decrease in the number of religious people, but mostly the consequence of supply-side factors (Bellucci and Heath, 2012). These contributions show a dramatic change after 1993, when the main confessional party (Democrazia Cristiana) disappeared and Italian Catholics had to choose between two new coalitions after a complete restructure of the political supply in 1994. Afterward, in line with the Politics Matter theories, the interlocking between parties and religious organizations grew dim and the effect of religion on voting behavior almost disappeared, replaced by the effect of ideology (Segatti and Vezzoni, 2008). Given the relevance of the ideological dimension, the political dimension seems to have conquered a certain autonomy, and the effect of religiosity on voting has been relegated as a property of the past for quite a long time, not only in Italy but throughout Western Europe at least.

However, not a single contribution completely closed the door for a potential comeback of religion as a driver of voting behavior in the next future. As Segatti and Vezzoni (2008: 16) emphasize “given the lack of the social conditions (diminishing number of practicing Catholics), there is no room for a revival of the traditional religious cleavage […] If religiosity will increase its influence on vote, it will not be through an activation of membership to a traditional religious institution like a church, but by means of a mobilization on religious issues taking place in the political sphere.” This shift is precisely what occured when religiosity experienced a resurgence in voting behavior literature, largely driven by the electoral consolidation and strategic evolution of the radical right family.

Radical right success and religiosity comeback

In the last few years, RRPs increased their vote share in Europe, becoming one of the most studied phenomena in political science. RRPs' relationship with religion has been a rather investigated subject, for at least two main reasons: first, for the salience of immigration framing in RRPs' mobilization strategies; second, for RRPs' leaders often claim to be among the last defenders of Christian values and civilization (Art, 2011; Minkenberg, 2018; Morieson, 2021). However, the relationship between RRPs and religious voters is not trivial. In fact, religious voters (intended as frequent church attendants) tend not to vote for a radical party when a sufficiently strong moderate alternative is available, not to mention when the political supply offers a confessional party (Immerzeel et al., 2013). Furthermore, RRPs do not have a stronghold on religion: they do not own confessional and moral issues (not even a traditionalist appeal, which they share with conservatives) and, especially during the first and second waves of the radical right, RRPs were recognized as more secularized than their center-right counterpart (cf. Mudde, 2007). However, scholars noted a growing alignment between RRPs and religious stances in Western Europe, where RRPs have historically struggled in mobilizing churchgoers. We explicitly emphasize the focus on Western Europe, as in Eastern Europe, RRPs have consistently integrated religious and nationalist platforms since they first appeared on the political scene especially in Poland, Hungary, and Slovakia (cf. Minkenberg, 2018 for details and further references).

However, even though the degree of adherence of RRPs to religious stances remains contextual, what really represents the common trait among Western and Eastern RRPs is their nativist approach. When we consider nativism (the division of society between native in-groups and non-native out-groups) as the main ideological outlook of RRPs, the references to Christianity get a more cultural nuance rather than a confessional strength. In this perspective, the appeal to Christianity helps to foster the boundary between us and them on which RRPs base their mobilization and their pans of policies (from expulsions to welfare chauvinism).

In fact, recent contributions show that religious stances have been increasingly tackled by RRPs in their manifestos, especially after migrants' crises in Europe; yet, RRPs tend to use religion strategically and to frame it according to their nativist ideology, which is why the exclusion of Muslims and negative evaluations about Islam hold much more space than the praise of Christianity (Schworer and Romero-Vidal, 2020; Schworer and Fernandez-Garcia, 2021). Quite ironically, this anti-Islamic orientation prompts both positive references to Christianity and support for secular arguments, depending on the situation.

This framework seems to hold in Italy as well, where religion has become one of the fundamental tenants of national identity within RRPs' communitarian and nativist worldview, withChristian symbols serving as symbols of community identity (Ozzano and Giorgi, 2013). Scholars also commented on the separation of Christian identity from the recognition of religious authority, which makes Italian RRPs more adaptive to the contingent needs. Such flexibility enables them to invoke conservative Catholic arguments when addressing social modernizers, while also defending free speech and individual civil rights when addressing Muslims (Ozzano, 2019; Ganesh and Froio, 2020).

To sum up, RRPs bring religion back into the picture, but in a completely different way than European politics were accustomed to. This use of religious stances in opposition to the non-native outgroup marks a difference from how religiosity have traditionally been interpreted by confessional parties. It may even significantly impact the type of voters RRPs mobilize, for instance people who identify as Catholics but are less devoted and participatory. More properly, we may be witnessing a non-linear, but still crucial, pattern between religiosity and voting behavior in the last Italian elections.

Research questions and expectations

Considering the literature, Italy is a promising case study for several reasons, from its prominent religious cleavage to the history of its party competition, but also for contingent reasons, as we witness similarities between supply and demand during the last Italian elections. Parties who used a strong nativist appeal and emphasized Christianity as one of the keepers of national identity, exploiting religious symbols yet with no explicit ties to Christian doctrine, reaped huge electoral rewards in a country where most people still identify as Catholics, but practice and attend religious services much less than before.

To grasp this growing category of Catholics, we adopt an increasingly popular multidimensional approach to investigate religiosity, which not only considers church attendance as a marker but also includes identification (cf. Jagodzinski and Dobbeleare, 1995; Storm, 2017; Piumatti and Russo, 2019). A multidimensional approach disentangles religious identity and commitment; therefore, it can help to identify new ways in which religiosity affects voting behavior in contemporary societies that increasingly moved away from the traditional religious cleavage.

In a society where the number of non-practicing Catholics keeps increasing, we must investigate nominal Catholics in depth. By “nominal” Catholics we mean those who identify as Catholics mainly for environmental and contextual reasons rather than for specific religious beliefs, thereby developing a sense of belonging based on identity and an ethno-cultural matrix (Garelli, 2020). As Garelli shows – and our data confirm – the nominal Catholic is the most frequent religious profile to be found in contemporary Italy.

So, based on the theoretical framework detailed beforehand and given the connotations of supply and demand, two main research questions guided our work.

The first mainly deals with the past theories on religious voting and, specifically, the way in which religiosity has been operationalized in many studies throughout indicators of church attendance. Our contribution aims at making a step forward from the one-dimensionality of attendance to the multidimensionality of the combination between religious participation and identification. This step is useful for two main reasons: first, it helps to separate different layers of religiosity (identification, belief, practice, participation) that would be otherwise compressed into the single use of the attendance indicator; consequently, it allows to directly tackle the category of nominal Catholics which would otherwise get missed.

RQ1. Does religiosity still influence voting behavior? And if so, which dimension matters the most?

Based on theory, we expect this category of Catholics to hold a distinct and specific voting behavior. We expect that nominal Catholics voted massively for RRPs in the last Italian elections. More generally, we expect religiosity to play an important role in shaping voting behavior, but in a new way that is identity-based and culturally framed, leading to discrimination within the category of believers.

The second research question explores the role of religiosity on voting behavior more deeply. Religiosity might even have a direct effect on voting behavior, but it could also be mediated by attitudes and values. We can find a vast amount of literature investigating the relationship between religiosity and the emergence of specific attitudes among the public. Some examples include the association between religiosity and anti-immigration stances (Storm, 2018), but also aversion toward homosexuality (Adamczyk and Pitt, 2009) or the micro and macro roles of religion in helping the development of attitudes toward ethical issues, such as abortion (Adamczyk, 2022). This stream of literature highlights the many effects religion can have on culture and society, while also cautioning against the risks of focusing solely on the potential direct effect of religion on voting behavior. In a nutshell, the level of religiosity as such may have a risible role in shaping one's vote choice, nonetheless it could foster and develop attitudes holding a robust effect on voting. Therefore, given that religiosity can trigger specific attitudes and voters may target parties intercepting those attitudes, we cannot discount the possibility of a mediated effect of religiosity on voting behavior, which may even be greater than its potential direct influence.

RQ2. Does religiosity hold a direct effect on voting, or is it mediated by value or attitudinal patterns?

To answer these two research questions, the present study will be structured in two main sections: the first will tackle religiosity and voting behavior, while the second section will take also values and attitudes and their relationship with religiosity into the analysis.

Data and methods

We investigated the research questions outlined above by using the latest ITANES election study.¹ The study was fielded between September and October 2022 and covered the Italian national elections held on September 25th, 2022, with a pre-electoral and a post-electoral survey. Since the aim of our work is to investigate the role of religiosity on voting behavior, we relied mainly on data from the post-electoral survey, and we integrated the indicator about attitudes toward immigration from the panel section of the pre-electoral study.

The dependent variable is the vote choice indicator measured in the post-electoral survey, which we restricted to the six major parties in terms of vote shares (Fratelli d'Italia – FdI; Lega; Forza Italia – FI; Partito Democratico – PD; Azione/Italia Viva – A-IV; Movimento 5 Stelle – M5S) and later dichotomized between right-wing parties and others (more on this in the Results section).

The main independent variable is a typology that combines information on two indicators: religious affiliation (“Which religion do you belong to?”), which aims at capturing religious identity, and religious practice (“Can you tell me how often you go to church (excluding ceremonies like baptisms, funerals, weddings)?”), which aims at bringing together participation and religious commitment. By combining affiliation and practice we can discriminate within the area of non-practicing citizens between those who do not practice because they do not identify as Catholics and those who do not (or do it seldomly) even though they identify as Catholics. In a few words, we split the sample in three categories: the practicing Catholics, the nominal Catholics- and the non-religious. It is crucial to state that “non-religious” category includes atheists and agnostic, and more generally those who declared as non-believers.² Furthermore, to avoid inconsistent religious profiles, we also excluded those individuals who declared themselves both as church attendants and not belonging to any religious denomination. Table 1 describes the procedure we employed to build our religious typology.

Table 1. Religious typology

We conducted several robustness checks to test for different thresholds in the information about attendance to the service. The results are fully consistent with those presented in the article. These tests will be available upon request to the authors.

Afterward, we rely on a set of three dichotomous variables gauging negative attitudes toward abortion (“we should make abortion more difficult,” the negative attitudes correspond to the answers “Quite agree” and “Agree a lot”), homosexuality (“It is positive to make homosexual marriages legal,” the negative attitudes correspond to the answers “Do not agree at all” and “Do not agree completely”) and immigration (positive values correspond to values 1, 2, 3 of a scale ranging from “1 – we receive too many immigrants” and “7 – We could welcome more immigrants”).

We also employed a few standard control variables to check for possible spurious effects in the multivariate models: Gender, Education, Age class, Income class, and Area of residence (geopolitical). The distribution of the different variables involved in the study is reported in Table A1.

To investigate our research questions and test our expectations we organized our analytical strategy into three different steps. First, we used a multinomial logistic regression model with controls to investigate the effect of religiosity on the probability to vote for each party. The existence of a clear and quite sharp separation between center-right/right-wing parties and the rest convinced us to dichotomize the dependent variable and investigate the probability to vote for right-wing parties in the following parts. For this second step we relied on a set of logistic regression models to test the effect of affiliation, participation, and the interaction between them. Third, after establishing the existence of an effect of religious belonging on voting behavior, we kept on using logistic models to investigate the existence of a possible mediated effect of religiosity by using attitudinal variables. To give also a more precise reading of the relative strength of these two components, we also perform and report an analysis of mediation (Nguyen et al., 2021). The idea lying behind this is to test out the potential role of religiosity on the development of specific attitudes and values that, in turn, may have a direct relation with vote choice.

Results

Does religiosity matter for voting behavior?

We will start the presentation of the results with Figure 1 which, albeit being descriptive, is useful to arrange and understand the distribution of religious and non-religious voters throughout the electorate of the six major Italian parties at the 2022 elections. What it shows is the distribution of voters for the six major parties in terms of our religious typology (bottom).

Figure 1. Distribution of voters within six major parties by religious typology.

By looking at the voters' distribution we get multiple information. First, each party has its own share of regular church attendants, regardless of parties' position on the political spectrum, ranging between 25.8 and 39.4%. To be more precise, if it seems quite reasonable to claim that regular churchgoers may feel closer to conservative and traditionalist parties, our data do not show wide differences between – for example – the share of churchgoers of the PD electorate and the one within Lega voters (26.5 vs. 30.0%). Our data also question the literature on RRPs claiming that regular churchgoers tend to vote for moderate parties whenever they can. Still, even though FI shows the highest percentage of churchgoers among the six selected parties (39.4%), Lega and FdI could mobilize that segment of voters as well.

Figure 1 also displays the disproportion between nominal Catholics and non-religious, and most importantly how this disproportion seems to be framed by a political border. In fact, all right-wing parties have a much larger share of nominal Catholics than the left-wing; on the contrary, left-wing parties show a much higher share of non-religious than the right-wing, with the latter disproportion seemingly much more pronounced than the first. Nominal Catholics, albeit representing more than half of right-wing voters, can be found quite equally also among left-wing voters: more than 40% of left-wing voters are nominal Catholics, and the difference never exceeds 15%. Quite the opposite, non-religious left-wing voters are three times the right-wing ones.

Figure 2 adds an important piece to the puzzle by showing vote distribution within the typology. The vote distributions of practicing and nominal Catholics are almost identical, while not religious voters stand out as they reward center-left parties much more than religious voters.

Figure 2. Distribution of votes for six major parties by religious typology.

These figures suggest that the difference stands between those who identify as Catholics and those who do not. The main discriminant may not lay between practicing and nominal, as we expected, but rather between Catholics and non-religious.

The data shown in Figure 3 follow this direction. Based on a multinomial logistic model also including gender, education, age class, income, and area of residence as controls, Figure 3 shows the changes in the probability to vote for each of the six parties depending on the category of our religious typology. No matter the party, the probabilities remain virtually untouched when we get from practicing Catholics to nominal Catholics. On the contrary, being non-religious raises the probability to vote for a left-wing party and lowers quite sharply the probability to vote for a right-wing party, and most notably for FdI.

Figure 3. Predicted values for vote choice (six parties) based on religious typology (results from a multinomial logistic model with control variables).

We get two main points from this part of the analysis: first, we show a strong difference between left-wing parties and right-wing parties while there is no strong evidence of heterogeneity within the two wings in terms of the effect of the religious typology; on the contrary, all parties contained inside the left or right group show a very similar electoral composition in religious terms. Second, this strong difference is not based on the “quality” of Catholicism but on Catholic identity itself, as it is not a matter of practice but a matter of affiliation.

To further frame the greater role of affiliation (identity-based) than church attendance (practice-based) in shaping vote choice, we tested the two indicators in three different models. First, we used attendance on its own (M1), then we added affiliation to the model (M2), and eventually we made them interact (M3). Given the evidence of homogeneity of religious characteristics of the electorate within the right-wing parties and the left-wing counterpart, we chose to dichotomize the dependent variable and to conduct our analysis on the probability to cast a vote for the right-wing rather than for the left-wing.³ The results are shown in Table 2 by means of odds ratio.

Table 2. Effect of attendance and affiliation on right-wing voting (odds ratio)

When taken on its own, church attendance has a quite robust effect on the probability to vote for right-wing parties. However, the coefficients are quite the same both for regular church-attendants and those who practice seldomly. So, the effect remains quite strong for both categories when we keep “not-practicing” as the base item, meaning that the great difference stands between those who do and those who do not attend religious services. However, attendance effect almost disappears when we add religious affiliation to the picture. Both the addiction of affiliation (M2) and the interaction between attendance and affiliation (M3) strongly reduce (and make it not-statistically significant) the effect of the former on right-wing vote choice.

Just to give a sense of the magnitude of the effects, being Catholic increases the probability to vote for a right-wing party by 0.16 compared to being not religious, while attending church at least monthly or occasionally increases – among those who identify themselves as religious – it by 0.049 and 0.051 respectively compared to never attending (and such differences are not statistically significant).

This is a further confirmation of the stronger role of religious affiliation (i.e. identity) on vote choice in Italy, and a further element to reconsider the essence of religious voting, which, on the one hand seems to still exist at least in the Italian scenario, but on the other it appears to have changed quite drastically as it seems not to be a sole matter of practice and attendance to religious services.

Direct or mediated? The nature of the effect of religiosity on voting behavior

In this second part of the study, we want to further test the goodness of our results by investigating if, and to what extent, the relationship between religious belonging and voting preferences is mediated by a specific set of attitudes. The idea underlying this last part of the essay is that religion, as repeatedly shown in the literature, might fuel and foster certain values and attitudes toward specific issues,⁴ therefore the effect of religiosity on voting behavior may be mediated by such attitudes, in the sense that parties can propose themselves as “recipients” of that specific moral and social issues and mobilize voters accordingly.⁵

The data at our disposal gave us the opportunity to investigate three different attitudes whose correlation with religiosity has been investigated in the literature: attitudes toward abortion, homosexual marriage, and immigration. The first empirical evidence we propose is a set of logistic models in which we test for the effect of the Catholic denomination and the different attitudes, by controlling for the usual set of potential confounders. If the effect of the Catholic denomination on party choice diminishes after introducing the variable concerning specific attitudes, this suggests that the impact of the former is, at least partially, mediated by the latter.

Table 3 presents the results. As anticipated, M1 confirms the significant role of Catholic affiliation in influencing the likelihood of voting for right-wing parties. Upon introducing three variables measuring negative attitudes in separate models (M2, M3, M4, and M5), we observe that the impact of religious affiliation persists but diminishes. This empirical evidence demonstrates that its effect on party choice is partly direct and partly mediated by a set of attitudes.

Table 3. Effect of religious affiliation mediated by abortion, homosexual marriage, and immigration attitudes (odds ratio)

To go deeper into the relative weight of the two components in explaining voting behavior, we also present some results coming from an analysis of mediation (see Nguyen et al., 2021 for details). By this technique, we can precisely disentangle the total effect (TE) of religiosity on party choice between a direct (DE) and an effect which is mediated by specific attitudes (IE).

No matter the attitudes we focus on, all the three models confirm that a direct effect of religious belonging on party choices exists. What Table 4 adds to this knowledge is that, in the case of homosexuality and, even more strongly, in that of immigration, there is also a strong component (10 over 27 p.p. and 13 over 27 p.p. respectively) of the total effect which is mediated by attitudes. This suggests that party choices can also be explained by the ability of parties to gain a position on these specific attitudes to attract voters.

Table 4. Mediation analysis of Catholic belonging on right-wing parties voting

The different weights of attitudes are also worth of attention. Immigration has the heaviest impact on voting for the right-wing, which is a legitimate outcome given that its greatest share of votes is taken by RRPs that found their ideological outlook on nativism and construct a crucial part of their political discourse on immigration stances. The greater importance of immigration is also made very visible when we compare the explained variance of the different models reported in Table 3. So, not only Tables 3 and 4 suggest a double effect of religiosity on voting behavior (both direct and mediated) but they also hint at crucial differences in the relationship between religiosity and different issues.

Conclusion

The empirical evidence provided by this article shows the continuing importance of religiosity in shaping voting behavior. On the one hand, we found evidence to claim that voting behavior during the Italian national elections of 2022 was correlated with cultural elements of religiosity, and, on the other, we also showed how religiosity proved to be crucial beyond participation and practice. If we expected a grounded and robust evidence that nominal Catholics tend to vote massively for the radical right – more so than practicing Catholics – then this expectation is not confirmed. Religiosity still plays a crucial role in shaping voting behavior, but in such a way that it reduces the salience of participation, giving prominence to the element of identity provided by religious affiliation. In short, the sense of identity and belonging provided by religion seems to be more important than the observance and celebration of religious duties, which do not hold the same discriminating effect they had in the past.

We rather expected a non-linear relationship between religiosity and voting behavior, in which practicing Catholics would cast their vote for more moderate parties and nominal Catholics would significantly do otherwise; in a nutshell, we expected we would place a demarcation line between those who practice and those who do not practice, while this line eventually had to be moved between those who identify as Catholics and those who do not. This also reflects on parties: if we supposed heterogeneity within right-wing parties, expecting nominal Catholics to vote massively for RRPs (such as FdI and Lega), we witnessed a clear separation between the left and the right dominated by religious identity. This may also be further evidence of the increasingly studied phenomenon of a-religious voting, which refers to the emergence of a divisive moral-ethic agenda that fuels non-religious voting for socialist parties especially in Catholic countries (Ignazi and Tuorto, 2023). Within this framework, voting behavior is framed by one's distance from religion, rather than one's adherence. Therefore, our results may also be read backward: it is not-being religious that may predispose one to vote for the left, not the other way around. This suggestion may even fit the Italian context, where declaring as Catholic is the standard, while it needs more effort to move away from religiosity (cf. Garelli, 2012).

Further results also support the care we address to the essential role of identity. Specifically, we showed two mechanisms through which religiosity is relevant to voting behavior: a direct effect of religiosity, in the shape of religious affiliation; and another effect mediated by values and attitudes, which is particularly prominent in the case of immigration issues.

This mediated effect seems to be carved into the conflict between traditionalists and liberals that is well documented (and debated) in the literature. Our results tell that religious affiliation (namely to identify as Catholics) is correlated with the embracement of specific values and attitudes, and both hold robust effects on voting behavior. Still, we cannot state the existence of a causal mechanism between affiliation and anti-abortion, anti-gay marriage, and anti-immigration attitudes. If attitudes toward abortion and homosexuality may quite convincingly be an effect of the level of religiosity, the literature tells a more sophisticated story about immigration by giving evidence of non-linear – and multidimensional – relationship (Ladini et al., 2021).

A second conclusion we cannot trace is whether these cultural orientations exert a noticeable effect on voting behavior only if defined by religiosity. Both attitudes and religiosity may stand under a wider shade, namely a broader conflict that may encapsulate them. Could it be the old post-materialist vs. materialist conflict well documented by Inglehart (1981) or a newer conflict that builds its core into conflicting identities: one liberal and secular, one traditionalist and religious, among other characteristics. This seems to be a much broader topic that needs further analysis. What we do know on the other hand, is the different robustness of the effect among attitudes. Specifically, the effect of attitudes toward immigration is greater than moral and ethical attitudes, which is consistent with the specificities of the Italian right-wing. The effect of attitudes could be greater or weaker in other contexts depending on the structure of the party supply. However, the impact of attitudes and their relationship with religiosity proved to be robust also within the Italian scenario.

Our first hint for further research comes accordingly: given the cultural and religious evolution happening in Western Europe, the interlock between attitudes and religiosity needs to be constantly taken into consideration in both single case studies and in comparative perspective. Second, the focus on the mediated effect of religiosity should also be paired with a more profound operationalization of religiosity itself. We think that future research should head toward more complex and complete measurements that encompass all the components of religiosity, including the one that seems to be holding more explicative power in contemporary Western European societies: identity.