How Religiosity Shapes Satisfaction with Democracy: A Mediterranean Case Study

Introduction

The number of democracies has continuously increased since the fall of the Berlin Wall. Today, more than half of all countries are indeed democratic (Pew Research Center, 2019). Paradoxically, the number of people dissatisfied with democracy keeps on increasing as well. As emphasised by Foa (2019) in a study of 154 countries, about 58% of people claim to be dissatisfied with democracy, which is its highest level since 1995. Even if the percentage of individuals satisfied with democracy differs depending on the countries chosen, studies constantly emphasise this state of affairs. For example, a recent study of the Pew Research Centre (2020) based on 27 countries also shows that a median of 52% of people is dissatisfied with democracy there. Precise numbers are debatable, but there indeed seems to be a structural phenomenon of dissatisfaction with democracy. Thus, various authors have considered that this represents a contemporary "crisis of democracy" (Crozier et al., 2012; Foa & Mounk, 2017; Held, 2006). Among them, Foa and Mounk think that this so-called "crisis" represents a danger for democracy. Indeed, they believe that individual perspectives on democracy, evaluated through satisfaction with it, influence its overall quality. This is because individuals who are less satisfied with democracy would be more prone to elect officials who may favour democratic backsliding (Foa & Mounk, 2017: 12). Democratic backsliding corresponds to a "deterioration of democratic governance in any regime". Such erosion can be brutal, such as with a military coup, or more incremental, through a progressive deterioration of the checks and balances that should control the executive, potentially leading to authoritarian regimes (Bermeo, 2016). Basing themselves on the case of countries like Poland, they show that such democratic deconsolidation can even happen in countries that were starting to be considered as consolidated (Foa & Mounk, 2017: p.11). In other words, they do not deny, as highlighted traditionally in academic literature, that consolidated democracies are less at risk of facing a phenomenon of backsliding than non-consolidated ones (Pérez-Liñán et al., 2019). However, they argue that even in consolidated democracies, discontent from the population towards democracy leads to its deconsolidation (Foa & Mounk, 2017).

Nevertheless, this idea of a "crisis of democracy" has been heavily contested. First, Kriesi shows that the current situation is not as alarming as pointed by Foa and Mounk. Despite a trend showing an overall decline in satisfaction with democracy, no determinism leads democracy towards its necessary deconsolidation (Kriesi, 2020). According to Kriesi, this idea of an intrinsic crisis of democracy is caused by a flawed use of statistical data. He specifically highlights issues of exaggerated conclusions "by cherry-picking cases", by using "improper visual presentations of the survey data" and by "interpreting the data superficially" (2020: 242). Second, Foa and Mounk's analysis –as well as the other authors supporting the idea of a crisis of democracy cited here- heavily relies on using a single variable that evaluates the satisfaction with democracy of the participants, without defining it properly. Indeed, it is often criticised as being interpreted as an opposition towards the legitimacy of democracyper se,whereas it is unclear what this variable expresses. One who is dissatisfied with democracy may be so because he opposes this regime as such. However, others may support it but find that their democracy does not fulfil what they regard as fundamental democratic standards (Linde & Ekman, 2003).

Nevertheless, this counter-argument to Foa and Mounk's work and to the use of the "satisfaction with democracy" variable is limited. Indeed, Kriesi convincingly highlights the flaws of Foa and Mounk's historical determinism conveyed by the idea of a supposed “crisis of democracy.” However, he eventually comes to a similar conclusion: that the lower the citizens are satisfied with democracy in general, the higher the risks of democratic deconsolidation (Kriesi, 2020). Linde and Ekman also recognise that, even if the "satisfaction with democracy" variable is not perfect, it is still an excellent way to analyse the health of a democracy. In their view, satisfaction with democracy is a good indicator of the performance of this regime. In the long run, if people are dissatisfied with democracy and its performance, they will have substantially more chances to support non-democratic forms of government, such as an authoritarian regime (Linde & Ekman, 2003). In summary, low satisfaction with democracy may not imply a rejection of democracy or always lead to the advent of an authoritarian leader. However, supporting an authoritarian leader is one of the outcomes that it has substantial chances to produce according to the literature, and which does represent a rejection of democratic regimes. Thus, despite being broad in scope, satisfaction with democracy remains an essential condition to prevent democratic deconsolidation from happening (Linde & Ekman, 2003).

Consequently, it is crucial to understand the factors that explain dissatisfaction with democracy. Previous studies have focused on different variables, such as economic inequality and (un)employment (Bauer, 2018), age (Bornand et al., 2017; Foa & Mounk, 2017), interpersonal tolerance (Sullivan & Transue, 1999). Nevertheless, yet, very little has focused on religiosity (Bloom & Arikan, 2013; Vlas & Gherghina, 2012). The relationship between this variable and satisfaction with democracy will be central to this paper. It is particularly interesting since Beckford and Luckmann have argued that the advent of modernity and secularisation in Western countries have removed the structural influence of religion both at a macro and micro levels (1992). Thus, if religion had indeed a statistically and substantially significant impact on individuals' satisfaction with democracy, it would tend to undermine their conclusion from this perspective. This topic can consequently help have a better understanding of the structuring role of religion in political behaviour.

We will rely on a case study of three Mediterranean countries: Italy, Spain and Portugal. To our knowledge, no study has focused on analysing the role religiosity plays in explaining satisfaction with democracy in this specific region, whereas it presents strong academic interests. Indeed, these countries are regularly pointed out as being very close from a cultural perspective, especially concerning religious values (Moreno, 2006). As a matter of fact, more than 80% of the Italian and Portuguese population and 70% of the Spanish identify themselves as catholic (Pew Research Center, 2018). Therefore, they constitute a fertile ground on which to do this research.

Thus, our main research question is: does religiosity favour or hamper satisfaction with democracy in Italy, Spain, and Portugal?

We will show that higher religiosity statistically leads to lower satisfaction with democracy in general. Building on a more specific example of dissatisfaction with democracy as well, we will also highlight that attending church regularly leads to higher approval for an authoritarian leader. These analyses will help better understand the strengths and the limits of using "satisfaction with democracy" as a variable and how to compensate for the difficulties it raises.

To do so, we will first focus on the theoretical framework and the literature that lead to our choice of hypotheses. Then, we will expose our methodology. Finally, we will analyse the results and conclude.

Theory

In order to understand whether religiosity favours or hampers satisfaction with democracy, it is first fundamental to define democracy. We will rely on Dahl's definition of democracy (1971). He regards democracy as a "government of the people, by the people and for the people". This implies substantial participation from the citizens, who can learn about policy alternatives and be active in defining and implementing them. At the same time, such a regime should promote principles of equality and tolerance through its institutions (Ibid). Today, democracy is mostly implemented under the form of representative democracy. One of the central characteristics of this regime is the expression of the power of the people through the election of officials that should theoretically act as "representatives" (Alonso et al., 2011: 23). In practice, it is difficult to assert that elected officials only act as "representatives" of the people, especially without imperative mandates (Roberts & Powell, 2005). Nevertheless, it remains a fundamental component of representative democracy. These definitions are essential to understand how satisfaction with democracy has been defined in academic literature.

As emphasised previously, satisfaction with democracy encompasses a wide scope of ideas. Linde and Ekman have shown that it is mainly composed of five elements that work both independently and jointly: 1) satisfaction with “political actors,” 2) with “regime principles,” 3) with “regime performance,” 4) with “regime institutions” and 5) with the “political community” (2003: 393).

Satisfaction with political actors could mean supporting or feeling represented by elected officials (2003: 401). Satisfaction with regime principles refers to an ideal of democracy, of how it should work in theory, such as considering the democratic principles evidenced by Dahl (1971).

Regime performance, in contrast, is the evaluation by the citizens of how democracy works in practice. From this perspective, citizens could be satisfied with democracy if it produces the outcomes they expect from the theory, favouring equality, tolerance, and political participation. In contrast, they could be dissatisfied with it if they subjectively believe it produces these outcomes insufficiently. Similarly, they could also be dissatisfied with democracy because of an opposition to what democracy promotes both in theory and practice. From a socio-psychological perspective, someone who is intolerant towards diversity would tend to be opposed to democracy (Sullivan & Transue, 1999).

Satisfaction with regime institutions relates to the practice of democracy as it is about the kind of institutions that implement these democratic principles (Linde & Ekman, 2003: 393). Dissatisfaction with it could consequently come from a preference for a direct democracy rather than a representative democracy (Ibid) or from an opposition to the separation of power and the will of having an authoritarian executive without checks and balances (2003: 398).

Finally, the notion of "political community", from a democratic perspective, refers to the citizens who participate in the functioning of democracy. That can be through "public participation" or helping fight against the exclusion and marginalisation of certain groups. Thus, participating in attaining common goals is at the core of the definition of democracy.Tolerance and willingness for equality are, once again, crucial here, as one who has a restrictive view of what the "political community is" –for instance, excluding migrants- would lean towards opposition to democracy as a system (Vlas & Gherghina, 2012).

In summary, dissatisfaction with democracy can consequently come from people who think their democratic system does not sufficiently fit with their democratic ideal and from people who oppose this regime and its values as a whole. This is why both these views influence the perception of democratic legitimacy in the long run, which is crucial to maintain a stable regime and prevent it from deconsolidation (Foa & Mounk, 2017). These elements are central to understanding how the relationship between religion or religiosity with democratic satisfaction has been studied in previous research.

First of all, it is necessary to define the concepts of religion and religiosity. Durkheim considers religion as a "solidary system of beliefs and practices related to sacred things". It establishes an opposition between what is profaned –and forbidden- and what is sacred and protected from desecration. The believers of a similar faith belong to a "moral community called Church". In other words, it is a social phenomenon that creates social interactions of which the basis is the sharing of some fundamental values related to the Beyond (Durkheim, 1912). These values have consequences on the believers’ relationship towards the society they live in, and as such, have political effects. Then, the concept of religiosity is twofold. It corresponds to the level of religious beliefs and practices of an individual (Bogdan et al., 2014).

Historically, the academic literature suggests that religion, and more specifically Christianity or even Catholicism, favours the establishment of democratic regimes and the development of democratic values. It should subsequently lead to compatibility between religiosity and democratic satisfaction as religion would have a structural influence on individuals, making them adopt democratic attitudes performatively (Foucault, 2004). Indeed, according to Moulin, Christianity has set the basis for the electoral democratic process. This is because the Church has been, in his view, the only institution to make its members choose their representatives by voting (Moulin in Jaffrelot, 2000: 20; see also Foucault, 2004). Furthermore, Tocqueville considers that Catholicism brings values of equality to individuals, which are crucial to implementing and supporting democracy (Tocqueville in Jaffrelot, 2000: 21). Similarly, for Huntington, if 39 out of the 46 democratic countries had a catholic culture in 1988, it is because of the historical diffusion of Catholicism that would have set the basis for these countries to become democratic (Huntington in Jaffrelot, 2000: 23). From this perspective, Catholicism would appear in stark contrast with Islam and Confucianism; the latter being considered by him as incompatible with democracy as they would supposedly not share these values (Ibid). There would thus be differences between religions in explaining the development of democratic values. This idea that Catholic values would favour democracy is also observable in more recent comparative politics works. Vlas and Gherghina, for instance, explain that Catholic values, after the Second Vatican Council, facilitate the adoption of human rights values by individuals (Vlas & Gherghina, 2012).

What is more, for every religion, being part of a community of believers who meet regularly by attending Church would strengthen individuals' "civicness" (Vlas & Gherghina, 2012). This civicness, which is crucial within democratic regimes, would increase their democratic attitudes and their satisfaction with democracy. From this perspective, religions, especially Catholicism, would positively influence the adoption of democratic values and behaviours, satisfaction with democracy, and, consequently, the maintaining of democratic regimes in the long run (Ibid).

Nevertheless, these arguments can be contrasted by emphasising how religion may lead to dissatisfaction towards democracy, undemocratic attitudes and potentially bring about democratic deconsolidation. Indeed, Bogdan et al. found that the higher the religiosity within one society and the more one religion is hegemonic within it, the fewer citizens are tolerant towards diversity, political pluralism and competition (Bogdan et al., 2014). It is thus not surprising that strong religiosity has been associated in the literature with higher degrees of homophobia (Laythe et al., 2001), with opposition to immigration –under specific circumstances- (Storm, 2018), and authoritarian views (Canetti-Nisim, 2004). In other words, it is not religion as such that would lead to democratic dissatisfaction and non-democratic views, but rather a high level of religiosity (Canetti-Nisim, 2004). However, these studies have mostly relied on the attitudinal part of religiosity, thus how regularly an individual attends his/her Church (Bogdan et al., 2014). Less focus has been put on the individual "belief" part of religiosity, while it is crucial to understand the relationship between religion and tolerance. Indeed, as emphasised by Portier and Willaime, high religiosity –regarded by them as the subjective importance of religion in one's life- may lead to higher intolerance because the higher one's religiosity, the higher the structuring role religion plays in his/her life (Portier & Willaime, 2021). To demonstrate this, they rely on the example of the French "Ancien Régime", thus before 1789.

During this period, religion could be considered an "englobing structure" within society (Beckford & Luckmann, 1992) as it was omnipresent in every part of it. Thus, there was no separation between sectors of activity, and religion was present in both the public and private levels of an individual's life. Consequently, religion was at the heart of a collective imaginary that made it virtually impossible to imagine other kinds of societies (Portier & Willaime, 2021), leading to intolerance and repressions of those who did not share these norms (Trousson & Vercruysse, 2003). Later on, processes of secularisation (Portier & Willaime, 2021) profoundly changed this state of affairs and provided religion with a more marginalised role within society. Nevertheless, Portier and Willaime argue that, nowadays, religion remains an englobing structure at the individual level despite being marginalised. In other words, individuals who have a strong religiosity would be more likely to have a vast majority of their life structured by religion and less acceptance of diversity of beliefs (Portier & Willaime, 2021), which is essential to sustain democratic regimes.

These elements show a possible relationship between high religiosity –in our case: within Catholicism- and democratic dissatisfaction. Our first hypothesis is consequently:

H1: Religiosity decreases the odds of being satisfied with democracy

Because the idea of "satisfaction with democracy" is very broad in scope, despite being useful to assess the health of a democracy, we will also develop another hypothesis to test a specific modality of dissatisfaction with democracy:

H2: Religiosity increases the odds of supporting an authoritarian leader

This element is critical as this literature review has shown that high religiosity can potentially lead to authoritarian views, which is a form of democratic dissatisfaction (Canetti-Nisim, 2004). This way, it will complement our first hypothesis by starting with a more general question on satisfaction with democracy before wondering about an element that substantially shows risks of democratic erosion by possibly electing leaders opposed to democracy.

In order to test these hypotheses, we will rely on a specific methodology.

Methodology

We have decided to rely on the European Values Study integrated survey 2008. The European Values Study is a cross national survey that studies values in contemporary society, such as religious beliefs, opinions on democracy and political preferences. This dataset is an integrated one comprising all the surveys that the European Values Study has done in 47 countries in 2008 and at the beginning of 2009. Initially, there were 477 variables and 66281 observations. For this research, a first step has thus been to only keep three countries: Italy, Spain, and Portugal. Nevertheless, there were still too many variables: 40 variables and 4572 observations. Consequently, we have selected 20 variables in this dataset that could be considered relevant by relying on the academic literature on the topic. We divided these variables into different categories: 1) the dependent variables, 2) the main independent variables evaluating religiosity, 3) the socio-demographic independent variables, 4) the independent variables relating to ethno-religious tolerance, 5) the independent variables relating to gender and sexuality tolerance, and 6) the independent variables relating to political and regime preferences. We also removed every line that contained a missing answer. In the end, we had 1751 observations. Here is a table with the descriptive statistics of each of the dependent and independent variables:

Dependent Variables

In order to test these hypotheses, we have decided to rely on two dependent variables:

-1) Satisfaction with democracy (named: satisfaction_democracy): it is intended to test H1. It is a categorical ordinal variable measured between 1 and 4. For clarity, we have reverse-coded it the following way: “1 = not satisfied at all with democracy,” “4 = very satisfied with democracy” (see appendix 1 for a comparison between the original variable and the recoded one).

-2) Support for a strong leader who does not have to bother with Parliament and elections (named: political_systems_strong_leader): it is intended to test H2 as it corresponds to supporting an authoritarian leader. It was initially a categorical ordinal variable, but we have recoded it as a binary variable with “1 = support for a strong leader who does not have to bother with Parliament and elections” and “0 = no support for a strong leader who does not have to bother with Parliament and elections” (see appendix 2 for a comparison between the original variable and the recoded one). We have recoded it because we were interested in whether the participant supported a strong authoritarian leader as such and not in the nuances of this support.

Descriptive statistics show that the answers to these variables are relatively equally distributed between the three countries.

Because satisfaction with democracy is an ordinal categorical variable, H1 will be tested thanks to ordinal logit models. Concerning H2, since supporting a strong leader who does not have to bother with Parliament and elections is a binary variable, it will be tested through binary logit models. Relying on logits seems relevant in this research, for it helps understand what the factors that make the current observation most likely -within this dataset- are. These models will be divided into four different thematics:

1) Socio-demographic factors

2) Ethno-religious (in)tolerance-related independent variables

3) Gender and sexuality (in)tolerance-related independent variables

4) Political and regime preferences-related independent variables

Independent Variable: Religiosity

In order to test whether there is a positive or negative relationship between religiosity and support with democracy from the one hand, and more specifically for an authoritarian leader on the other, we will divide religiosity into two main independent variables:

-1) The subjective importance of religion for an individual (named “how_imp_religion”)

-2) Religious attendance (named: “religious_attendance”).

These two main independent variables will be tested in each thematic, and at least one will be present in each model. These variables respectively correspond to the two facets of religiosity –both religious practices and individual faith. We have reverse-coded them in order to make greater scores account for greater religiosity –the order is between 1 and 4 regarding the importance of religion, and between 1 and 7 for religious attendance (see appendix 3 for a comparison between the original variable and the recoded one).

Other thematic independent variables

As per the other independent variables, they will be specific depending on the thematic of each model.

First, regarding the socio-demographic thematic, it will be composed of:

-age, which is a quantitative and continuous variable

-education, which is a quantitative and ordinal variable comprised between 1 and 20 (1 being the lowest level of education and 20 the highest). However, the maximum answered by respondents is 6 in this dataset.

-employment, which is a binary variable (1= employed, 0= unemployed).

-Female, which is the original “gender” variable recoded for clarity as “1 = female, 0 = male” (see appendix 4 for the comparison between the original variable and the recoded one).

-Vote is a binary variable assessing whether the individual would vote if there were an election (1 = would vote, 0 = would not vote).

-Interest in Politics (named: interest_politics), which is a categorical ordinal variable measured between 1 and 4 and which has been recoded as follows: “1 = not satisfied at all with democracy,” “4 = very satisfied with democracy” (see appendix 5 for a comparison between the original variable and the recoded one).

The last two variables are interesting in this thematic focusing on demographic factors since they have been identified in the literature as being strongly correlated with cultural and economic capital coming, for instance, from the level of education or the employment situation of the individual (Dormagen & Mouchard, 2015).

Some of these socio-demographic variables –especially “female” and age- may be present in other thematics as control variables.

Second, concerning the ethno-religious (in)tolerance-related independent variables, it will be composed of:

-Whether the participant would refuse to have muslims as neighbours or not (named: neighbours_muslims). It is a binary variable coded as “1= refusal, 0= non-refusal.”

- Whether the participant would refuse to have jews as neighbours or not (named: neighbours_jews). It is a binary variable coded as “1 = refusal, 0 = non-refusal”

-Whether the participant accepts or rejects immigration (1 = Acceptance of immigration, 0= rejection of immigration). It was initially a categorical ordinal variable, but we have recoded it as binary (see appendix 6 for a comparison between the original variable and the recoded one).

Third, the sexuality and gender (in)tolerance-related independent variables will be composed of:

-Whether the participant would refuse to have homosexuals as neighbours or not (named: neighbours_homosexuals). It is a binary variable coded as “1= refusal, 0 = non-refusal”).

-Whether the participant believes divorce can be justified (named: justified_divorce). It was initially a 10-scale variable (1= never justified, 10 = always justified), but we have recoded it as a binary variable (1 = can be justified, 0= cannot be justified) since we were more interested in knowing whether divorce had relatively strong chances of being justified for them or not (see appendix 7 for the comparison between the original variable and the recoded one).

-Whether the participant approves of abortion (named: abortion). Its coding has been slightly more complex. Indeed, two binary variables were initially asking whether they could approve of abortion for a married woman and a non-married woman respectively. We have merged these two variables to make an ordinal one with three possible answers (2 = always approves, 1 = depends on the context, 0 = never approves; see appendix 8 for the comparison between the original variable and the recoded one).

Fourth, regarding the regimes and political preferences-related independent variables, it will be composed of:

- Support for a strong leader who does not have to bother with Parliament and elections (named: political_systems_strong_leader) only when satisfaction with democracy is the dependent variable.

-Satisfaction with democracy (named: satisfaction_democracy) only when strong leader is the dependent variable.

- How the participant positions himself/herself on a left-right scale (named: lr_scale). It is a 10-scale variable assessing where the participant estimates his political positioning to belong (1 = left, 10 = right).

Dividing our models into thematics seemed to be the best way to consider various relevant variables to limit the risk of omitted variables –though it remains present. It was also a proper way to prevent ourselves from the risk of model overfitting –by having too many variables in one model.

Analysis and Results

As expressed previously, in each thematic, we have made two models, one composed of the independent variable assessing the importance of religion for the participant, and another with the independent variable assessing religious attendance. The results of each regression will be presented through odds ratios, which are useful to understand whether specific dependent variables increase or decrease the chances of an outcome –the dependent variable- to happen. In other words, it will help us know whether our main independent variables on religiosity –relating to the subjective importance of religion for an individual and religious attendance- increase or decrease the chances for these individuals to be satisfied with democracy in general on the one hand, and to support the leadership of an authoritarian leader on the other. We first regressed satisfaction with democracy against socio-demographic variables. The results are presented through the following odds ratios:

It is possible to observe that the importance of religion for an individual is statistically significant both at the 95% and 99% confidence rates. As the importance of religion increases, the more its odds ratios diminish. In other words: in these models, the more religion is important for an individual, the less he/she tends to be satisfied with democracy all else being equal. It also has a substantial impact on the dependent variable. As an example, Table 2 shows that when satisfaction with democracy is regressed against demographic variables, the fact that religion is important for an individual (how_imp_religion3) diminishes by about 30% his/her chances of being satisfied with democracy, and of 40% when it is very important (how_imp_religion4).

This variable is statistically significant with these demographic variables. But, it is necessary to verify that it is not only significant within this configuration because of possible omitted variables. Consequently, we have also regressed it against variables corresponding to the other thematics discussed previously –gender and sexuality (in)tolerance, ethno-religious (in)tolerance and political and regimes preferences (see appendix 9 to verify the content of these other thematic tables). It is worth noting that, in all these models, very few variables are statistically significant, and even less have a substantial impact. Besides, the observations we have made regarding the importance of the religion variable specifically are also observable in every other model we have made. As an example, when satisfaction with democracy is regressed against gender and sexuality (in)tolerance-related variables, the fact that religion is important or very important for an individual reduces respectively his/her chances of being satisfied with democracy respectively by about 29% and 34%. Similar magnitudes are observable when satisfaction with democracy is regressed against ethno-religious (in)tolerance-related variables and against political and regime preferences (see appendix 9).

However, the same does not completely hold true when looking at the relationship between religious attendance and satisfaction with democracy. Indeed, in all these models, religious attendance is only statistically significant at its maximum or extreme value (religious_attendance7). Even strong religious attendance (religious_attendance5 and religious_attendance6) prove statistically insignificant, while the repartition of respondents is relatively equilibrated between these three modalities. Consequently, it is difficult to draw any substantial lesson from the religious attendance variable here.

Our first hypothesis (H1) is consequently partially verified. Indeed, it is only verified regarding the importance of religion, not religious attendance, while it is a central component of religiosity.

We have relied on the same process concerning our second independent variable of interest: a strong leader who does not bother with Parliament and elections –meaning an authoritarian leader. First, we have regressed it against demographic variables. The results are the following:

Once again, very few variables are both statistically significant and have a substantial impact on the dependent variable. However, religious attendance consistently has a strong positive influence on supporting an authoritarian leader. Even in one of its lowest modalities (religious_attendance2) it increases the odds of supporting a leader who does not bother with Parliament and the executive by virtually 84%. In its highest statistically significant modality (religious_attendance6) it even increases it by about 120%. All these elements are also observable when regressing this dependent variable against the other thematic independent variables we have chosen (see appendix 10 to verify the content of these other thematic tables). For example, when regressed against gender and sexuality (in)tolerance-related independent variables, it is striking that the second modality of religious attendance (religious_attendance2) increases the chances of supporting an authoritarian leader by about 100%. Regarding its sixth modality, it is even an increase of virtually 150% (see appendix 10). It consequently seems to confirm our second hypothesis.

Nevertheless, the results regarding the importance of religion variable tend to infirm our hypothesis. As observable in table 3 -and the same goes on in every other model with the same dependent variable in which it has been tested- it is only statistically significant at its maximum modality. However, when doing a joint descriptive statistics of the importance of religion and the support for an authoritarian leader variables, it is possible to see that among those who consider religion as being very important in their life, only a minority (22%) supports an authoritarian leader. Thus, an increase in one unit of this modality of the importance of religion variable decreases the odds of supporting an authoritarian leader by 42%.

In other words, H2 is only partially confirmed as well. It is confirmed when looking at religious attendance since this variable makes it substantially more likely to support an authoritarian leader who does not bother with Parliament and elections. However, when looking at the importance of religion variable, it goes in the opposite direction.

We do not have a precise explanation of the reasons that led to these results that do not fully confirm our hypotheses. However, considering that the importance of religion for an individual leads to greater chances of not being satisfied with democracy, and of religious attendance to increasing the odds of supporting an authoritarian leader, seems to be in line with what has been exposed in the theoretical part of this research. The fact that when religion is very important for an individual it decreases substantially the odds of supporting an authoritarian leader is more counter-intuitive. Still, literature on the topic can be useful to try to have a first understanding of this observation.

As emphasized by previous research, religiosity is not only correlated to intolerance and to support for authoritarian regimes (Canetti-Nisim, 2004), but catholic values also seem to potentially favour democracy as a regime (Jaffrelot, 2020). It also appears that certain religious values could strengthen the tolerance of individuals having a strong religiosity, which is fundamental within democratic regimes (Ben-Nun Bloom et al., 2015). As a result, those individuals who subjectively consider religion as being very important to them may have strong democratic values. These values would lead them not to be satisfied with democracy, for it could be considered as not working sufficiently compared to their ideal of democracy. At the same time, despite being deceived of the way their democracy works, their strong religiosity would logically lead them not to support authoritarian leaders.

Besides, to understand the contrast between these latter individuals, for whom religion is very important and who would have strong democratic values, with those who attend Church regularly and support significantly an authoritarian leader, it must be emphasized that they may not be the same. As shown by Portier and Willaime, religiosity appears in different forms, and those who have a more “private” approach of religion, relying on a more independent faith, are not necessarily the ones who go very often to Church (2021). Regarding the latter, it is possible to hypothesize that, in opposition to the former, attending Church regularly makes them part of a community in which hierarchies differ from the ones induced by a democratic regime. This is viewable since the Pope is the leader of the Catholics and rules the Vatican with no separation between the executive, legislative, and judiciary powers as well as with his religious authority (Loi Fondamentale de la Cité du Vatican, 2000). Therefore, this socialization could reduce their attachment to democratic institutions and facilitate their support for authoritarian regimes.

Nevertheless, these possible explanations are only hypotheses at this stage, and they could be explored by further research on the topic.This could be done by relying on qualitative methods, especially semi-structured interviews and ethnography, to understand how their specific socializations influence the different ways the religiosity variables shape their satisfaction with democracy.

Conclusion

Our topic aimed at understanding whether religiosity favoured or hampered satisfaction with democracy in Italy, Spain and Portugal. We hypothesized that religiosity would both increase the odds of being dissatisfied with democracy and of supporting a strong authoritarian leader. However, these hypotheses have only been partially confirmed. We found that while the importance of religion –here: Catholicism within Mediterranean countries- significantly reduces the odds of being satisfied with democracy, it leads to opposite conclusions regarding supporting an authoritarian leader. In contrast, high religious attendance increases the chances of supporting an authoritarian leader substantially. These results led us to suggest that the importance of religion –and here Catholicism more specifically- could make individuals develop stronger democratic values, making them not satisfied with their democracy but also not supporting an authoritarian leader. We also thought that support for authoritarian leaders from those who attend Church could be explained by their socialization within institutions of which the governance is very dissimilar to democracies. This research consequently provided us with new hypotheses that could form the basis of further research on the topic. These results are complementing past research in interesting ways.

First, concerning the methodological debates about the use of satisfaction with democracy as an independent variable, our research demonstrates that it should be used –almost- only with another dependent variable corresponding to one of its specific modalities –such as, in our case, support for an authoritarian leader. We could not have reached these conclusions, especially the possible differences between the importance of religion and religious attendance in terms of the democratic values they may generate for an individual without using these two dependent variables. This is because using this second dependent variable, assessing whether individuals support an authoritarian leader, helped us make the results of the regression of satisfaction with democracy more accurate. In other words, complementing the use of satisfaction with democracy as a dependent variable by another variable that corresponds to one of its modalities is helpful to make clear when individuals are unsatisfied with democracy because they may have non-democratic values, or because they may have strong democratic values. From this point of view, we argue that, under these conditions, using satisfaction with democracy as a dependent variable can be useful to understand such social phenomena, which is a response to the criticisms made in the previous literature towards this variable.

Second, regarding previous research on religiosity, our study shows that its different components –religious beliefs and religious practices- may act as different social structures subsequently leading to different political behaviours. In other words, assessing religiosity as a whole with only one of these two components seems wrong. Using these two separate modalities of religiosity to assess the structural consequences of this variable seems to help reach more nuanced conclusions than previous works in the field. Narrower definitions of religiosity may have led researchers to focus on only one of these modalities, hence possibly explaining the opposite conclusions they have reached when analysing the relationships between religiosity and democracy. Further research could also consider the possible differences between religions, which has not been possible here because of the virtual absence of religions other than Catholicism in our dataset (see table 1).

---

References

Alonso, S., et al. (2011) The Future of Representative Democracy. Cambridge University Press.

Bauer, P. C. (2018). Unemployment, Trust in Government, and Satisfaction with Democracy : An Empirical Investigation. Socius, Vol 4, pp.1-14

Beckford, J-A., Luckmann, T. (1992). The Changing Face of Religion. SAGE Studies in International Sociology.  

Ben-Nun Bloom, P., et al. (2015). Praying For Both Teams: How Religion Both Facilitates and Dampens Anti-Immigration Sentiments. LSE Blog. Available at: https://blogs.lse.ac.uk/politicsandpolicy/praying-for-both-teams-how-religion-both-facilitates-and-dampens-anti-immigration-sentiment/#Author [Last access: 25/05/2021]

Bermeo, N. (2016). On Democratic Backsliding. Journal of Democracy, Vol 27(1), 5‑19.

Bloom, P. B.-N., & Arikan, G. (2013). Religion and Support for Democracy : A Cross-National Test of the Mediating Mechanisms. British Journal of Political Science, Vol 43(2), 375‑397.

Bogdan, D., Ciipriian, P., & Mariia, D. (2014). Religion and Democracy. Journal of Heterodox Economics, Vol 1(1), 31‑53.

Bornand, T., Biard, B., Baudewyns, P., & Reuchamps, M. (2017). Satisfaits de la démocratie ? Une analyse du soutien démocratique à partir de la comparaison de deux méthodes de classification des citoyens. Canadian Journal of Political Science, Vol 50(3), 795‑822.

Canetti-Nisim, D. (2004). The Effect of Religiosity on Endorsement of Democratic Values : The Mediating Influence of Authoritarianism. Political Behavior, Vol 26(4), 377‑398.

Crozier, M. J., Huntington, S. P., & Watanuki, J. (2012). The Crisis of Democracy. Report on the Governability of democracies to the Trilateral Commission. Sociología histórica, Vol 1, 311-329.

Dahl, R-A. (1971). Polyarchy: Participation and Opposition. Yale University Press.

Dormagen J-Y., Mouchard, D. (2015). Introduction à la sociologie politique. De Boeck Supérieur.

Foa, R. S., & Mounk, Y. (2017). The Signs of Deconsolidation. Journal of Democracy, Vol 28(1), 5‑15.

Foucault, M. (2004). Sécurité, Territoire, Population – Cours au college de France. 1977-1978. Seuil.

Held, D. (2006). Models of democracy. Stanford University Press.

Jaffrelot, C. (2000). Introduction. In Démocraties d'ailleurs. Démocraties et démocratisations hors d'Occident. Karthala, 6-57.

Kriesi, H. (2020). Is There a Crisis of Democracy in Europe? Politische Vierteljahresschrift, Vol 61(2), 237‑260.

Kroknes, V. F., Jakobsen, T. G., & Grønning, L.-M. (2015). Economic Performance and Political Trust : The impact of the financial crisis on European citizens. European Societies, Vol 17(5), 700‑723.

Laythe, B., Finkel, D., & Kirkpatrick, L. A. (2001). Predicting Prejudice from Religious Fundamentalism and Right-Wing Authoritarianism : A Multiple-Regression Approach. Journal for the Scientific Study of Religion, Vol 40(1), 1‑10.

Linde, J., & Ekman, J. (2003). Satisfaction with democracy : A note on a frequently used indicator in comparative politics. European journal of political research, Vol 42(3), 391‑408. Loi Fondamentale de la Cité du Vatican. (2000). Etat de la Cité du Vatican.

Moreno, L. (2006). The Model of Social Protection in Southern Europe. Revue francaise des affaires sociales, Vol 5, 73-95.

Pérez-Liñán, A., Schmidt, N., & Vairo, D. (2019). Presidential hegemony and democratic backsliding in Latin America, 1925–2016. Democratization, Vol 26(4), 606‑625.

Pew Research Center. (2018). Being Christian in Western Europe. Report, 1-168.

Pew Research Center. (2019). Despite global concerns about democracy, more than half of countries are democratic. Available at https://www.pewresearch.org/fact-tank/2019/05/14/more-than-half-of-countries-are-democratic/ [Last access: 25/05/2021].

Pew Research Center. (2020). Democratic Rights Popular Globally but Commitment to Them Not Always Strong. Report, 1-53.

Portier, P., Willaime, J-P. (2021). La religion dans la France contemporaine. Entre sécularisation etrecomposition, Armand Colin.

Roberts, A., & Powell, G. B. (2005). The Quality of Democracy. Comparative Politics, Vol 37(3), 357‑376.

Storm, I. (2018). When does religiosity matter for attitudes to immigration? The impact of economic insecurity and religious norms in Europe. European Societies, Vol 20(4), 595‑620.

Sullivan, J. L., & Transue, J. E. (1999). The Psychological Underpinnings of Democracy : A Selective Review of Research on Political Tolerance, Interpersonal Trust, and Social Capital. Annual Review of Psychology, Vol 50(1), 625‑650.

Trousson, R., Vercruysse, J. (2003). Dictionnaire général de Voltaire. Puf.

Vlas, N., & Gherghina, S. (2012). Where does religion meet democracy? A comparative analysis of attitudes in Europe. International Political Science Review, Vol 33(3), 336‑351.

---

Appendix

Click here to access the appendix.