Content uploaded by Alvaro Martinez Perez
Author content
All content in this area was uploaded by Alvaro Martinez Perez on Oct 24, 2014
Content may be subject to copyright.
1
Università degli studi di Milano
Dipartimento di studi sociali e politici
Working Papers
del Dipartimento di studi sociali e politici
25 / 01/ 2006
Eadem sed Aliter :
Religious Voting in Portugal and Spain
Kerman Calvo, Álvaro Martínez, José R. Montero
(University of Essex / Instituto Juan March)
www.sociol.unimi.it
2
Eadem sed Aliter: Religious Voting in Portugal and Spain *
Kerman Calvo
University of Essex
Álvaro Martínez
Instituto Juan March and University of Essex
and
José Ramón Montero
Universidad Autónoma de Madrid and Instituto Juan March
* We would like to thank Carlos Jalali for his comments to a previous version of this
paper, co-authored with him in January 2005 at a Conference in Lisbon on
“Comportamento electoral e attitudes políticas: Portugal no contexto europeo”.
Versions of this paper have been also presented at the Third General European
Consortium for Political Research (ECPR) Conference, Budapest, September 2005,
and at the 7th Congreso, Asociación Española de Ciencia Política, Madrid,
September 2005. We would also like to express our gratitude to the Centro de
Estudios Avanzados en Ciencias Sociales, Instituto Juan March, Madrid, for his
excellent facilities, and to the Universidad Autónoma de Madrid and the Spanish
Ministerio de Educación y Ciencia (SEC2002-03364) for their financial support.
Paper presented at the Dipartimento di Studi Sociali e Politici, Facoltà di Scienze Politiche, Università di
Milano, 25 January, 2006
3
Introduction
The political impact of religiosity has for long been a neglected issue, both in Europe in general, and in the
Iberian Peninsula in particular. Yet the intensity of conflict generated by religious issues in many European
countries over their recent past suggests religious issues can play an important role in the public sphere, and
particularly in the realm of voting behaviour. In order to clarify the links between religiosity and voting, this
chapter studies the bearing of religiosity on electoral behaviour in two specific countries: namely, Spain and
Portugal.
The paper aims at disentangling some puzzles regarding party competition in two countries that exhibit a
number of remarkable similarities. To begin with, Spain and Portugal are both essentially Catholic countries.
This should be taken to imply both a particular configuration of the country’s religious map, as well as the pre-
eminence of an influential culture based on Catholic values and ideas. Also, both countries have seen their
politics deeply interwoven with religion at crucial historical junctures. Paraphrasing Juan Linz’s (1993: 1)
dictum, in the 20th century the somewhat parallel history of religion in Portugal and Spain is certainly full of
dramatic events. In both countries, the outburst of authoritarian ruling was grounded – among on other things –
in a deeply entrenched conflict around clericalism. And, in both, the architects of the new democratic regimes
faced a long list of unresolved religious issues with the potential of shattering the new democratic order into
pieces. Lastly, in both countries the conflict around religious issues seems to be gaining new momentum --
notably, over the issues of religious education or gay marriage in Spain, or the controversy over the placing of
crucifixes in public schools in Portugal.
So, does a common Catholic culture lead to commonalities in the ways religiosity links up with political
behaviour? After the longest period of democratic policy making in the Iberian Peninsula, and in a context of a
seemingly unstoppable process of secularization, is still religiosity a factor that shapes the electoral fortunes of
major political parties? From the perspective of the relationship between religiosity and party choice, in what
ways are Spain and Portugal different? More generally, in observing the interweaving of religiosity and
political behaviour in two Catholic countries, what are the lessons to be drown for the understanding of
religious voting in European democracies?
We argue that, despite a shared tradition and a common history, the relationship between religiosity and
politics is taking different shapes in Portugal and Spain. The results of a multivariate analysis of post electoral
data in both countries reveals that the strength of the relationship is not even between the two cases: it is
stronger in Spain than in Portugal. Also, we defend that to understand the bearing of religiosity on voting in
Spain, a close look at the interacting effects between religion, ideology, and voting needs to be carried out.
That, however, is not necessary in the case of Portugal, where the effect of religious identities on voting is
apparently not significantly mediated by ideology.
4
We organise the paper as follows. In the first place, some paradoxes coming from the theoretical treatment
assigned to religion in most standard accounts of voting are considered. If religiosity is on the verge of
disappearing due to the secularisation process, how the persistence of the religious factor in many western
European countries can be explained? Secondly, the religious maps of Spain and Portugal are sketched.
Thirdly, the specification of the model is briefly discussed. Fourthly, the results of the “full” model composed
for the four surveys are examined. This is meant to discuss the magnitude of the direct effects of religiosity on
the vote. Fifthly, we move to the discussion of the indirect effects, considering in this section the bearing of
religious value orientations on ideological identities. Sixthly, an explanation for the differences between the
two cases is presented. Lastly, we conclude.
Religious cleavage versus religious voting
The attention paid to the religious cleavage has been uneven, to say the least. Of course, the “State-Church
cleavage” was among the few Lipset and Rokkan (1967: 50 ff.) privileged as the ones becoming frozen after the
crystallisation of mass politics in the inter-war years. But after their seminal work, the religious cleavage has
merited only sporadic interest, and even then, it has largely been the object of sweeping generalisations that
have done little to clarify issues of causality (Knutsen 1995: 463). To begin with, it has received far less
attention than the class cleavage. In contrast to the alleged ubiquity of class conflicts, the religious cleavage has
often wrongly been assumed to be limited to the few cases in which confessional parties occasionally compete
with class-based parties, and/or in which serious conflicts have sometimes broken out for religious reasons lato
sensu. But this obviously is far from true: in many countries, religious conflicts exist in the absence of
confessional parties. Moreover, the scarce attention addressed to the religious cleavage contradicted its apparent
relevance in many comparative analyses undertaken in the 1970s (Converse 1974: 733-734; Lijphart 1971: 7-8;
1980: 287; Rose and Urwin 1969: 12).
The standard accounts of the religious cleavage in the New Politics of the 1980s and afterwards revolved
around the framework of “decay”: if for some the religious cleavage has, as a minimum, been dramatically
reduced, for others the presence of religious factors in the voting choice has vaporised. As stated for instance by
Franklin, Mackie, et al. (1992: 40), “the decline in the political saliency of religion should certainly have hurt
the electoral prospect of religious parties, and may also have hurt right wing parties in general”. This decline
reaches the apex in many empirical analyses of voting behaviour, where religious variables are simply not taken
into consideration and thus not included in any of their otherwise extremely sophisticated statistical models
aiming at explaining voters’ choices.
Thanks to the rise of theories pointing to the weakening of social cleavages in Western Europe, the
intensifying processes of partisan dealignments, and their combined effects on electoral change (Dalton,
Flanagan, and Beck 1984), the analysis of religiosity gained again new momentum in the mid-1990s (Dalton
1996: 331). Paradoxically, the new interest in religious voting emerged from the confirmation of the limits of
5
the secularization argument. For one, secularization seems to adopt context-specific meanings, implying
distinctive social and political processes in different countries. Likewise, the rise of new form of religious
fundamentalism, and the resurgence of political conflicts around a number of “religious-issues” compose a
scenario where religious divisions retain a potential for divisions at the level of politics.
We claim in this paper that, in spite of the alleged decline of the religious cleavage, in many Western
European countries there is a religious voting. Religiosity is surely lessening its political significance and has
consequently much lower impact on voting. Party leaders, building strategically on the outcomes of
secularisation and social change processes, as well as on increasing levels of education and information, have
decided to maximise their electoral appeals by downplaying the conflictive ladders of religious divisions. In
most countries dominated by only one denomination, leaders of both confessional and secular parties have come
to accept neither to incorporate religious conflicts back into the political agenda, nor to mobilise voters with
religious or anti-religious flags in electoral contests. Yet, religiosity appears as a relevant variable for explaining
voting in a number of countries (Calvo and Montero 2002; Kotler-Berkowitz 2002). This variable has different
doses of explanatory power and shows also distinctive relationships with other independent variables,
depending upon the contingent combination of the social, cultural, and political features of a given country. But
it does matter. In other words, despite the fact that society might be less religious as a whole, those that remain
religious are growing politically committed, developing intense preferences in a broad spectrum of issue
connected with social and political life. Moreover, in some other countries a growing number of indicators point
at a renewed radicalisation of those segments of society that resist secularization. To the extent that these two
different religious situations can impinge on voting choices, the case for further attention to religious voting is
clearly a pressing one.
Religious voting implies some type of systematic association between religiosity and vote in a number of
citizens. In this sense, it is roughly equivalent to class voting, or ethnic voting, or race voting. Take, for
instance, the first. If we substitute class by the pertinent indicator of religiosity, the concept of religious voting
is immediately apparent in the definition of class voting provided by Evans (2000: 401). In his terms, “at first
glance, the idea of class [religious] voting appears straightforward. It refers to the tendency for voters in a
particular class [denomination, or level of religiosity] to vote for a specific party, political candidate (or
groupings of these), rather than an alternative option, compared with voters of another class or classes
[denominations, or levels of religiosity]. In other words, class [religious] voting describes a pattern of
association between class [religiosity] and vote”. In this sense, religious voting seems intuitively to be less
strong than the religious cleavage. But the repeated existence of a clear connection between some relevant
empirical indicators of religiosity, on the one hand, and voting for a given party, on the other, needs to be
explained by a research strategy other than the non-inclusion of religious variables in models of electoral
behaviour.
Religious maps compared
6
Very broadly, the Portuguese are more religious than their Spanish counterparts. According with recent data
from the European Social Survey (ESS), if in Spain the percentage of those belonging to a religion has
descended to 75 per cent of the population, in Portugal the figure remains at an overwhelming 86 percent, the
third highest in Europe.1 This suggests that the secularization process has been met differently in the two
Iberian countries, an impression that is confirmed by the examination of further indicators of religiosity.2 For
instance, Table 1 reports a classification of European countries according to their average level of subjective
religiosity. The contrasts are apparent: whereas Portugal belongs to the group of highly religious societies
(formed by Greece, Poland, Italy, and Ireland), Spain displays levels along the lines of the more secularized
societies. The indicators assembled in Table 2 further confirm that Portugal is more religious a country than
Spain. Just to mention a few, note that while in 1999 82 percent of Portuguese respondents considered
themselves religious, only 56 percent affirmed so in Spain. The importance of religion seems to be much more
important for the lives of the Portuguese: the gap between the two countries as far as the capacity of religion to
bring comfort and strength is an astonishing 27 percentage points. Lastly, remarkable differences can be found
in relation to the public role attributed to the Church. If on the whole Spaniards seem reluctant to define the
Church as a trustworthy institution, the Portuguese trust their Church firmly, capable of satisfying the people’s
spiritual needs.
Tables 1 and 2
Contrasting the levels of religious practice will also contribute to set out the differences in the religious map
in the two countries compared. Not only is Portugal more devout a country than Spain: it also exhibits rates of
religious practice that outmarch virtually any other in Europe. Among the Catholic countries, only the Polish,
Irish, and Italians go to church more frequently (Table 3). Spain, in contrast, appears at the bottom of the list of
Catholic countries. It is noteworthy that, with the new picture, important dissimilarities emerge among
European confessions. While a sizeable segment of the population in Catholic countries is still willing to
comply with the mandates of their religion, the rates of religious practice have dramatically dropped in
countries with Protestant or Lutheran majorities.
Table 3
1 European Social Survey (2002-2003) [ESS]; Questions C9 and C10. Of course, Catholicism is by far the religion that the
majority declares to belong to in both countries: 97 percent of those belonging to a religious defined themselves as Roman
Catholics in Spain and Portugal alike. Only in Italy and Poland (99 percent) we find a higher proportion of Catholics. See
Calvo and Montero (2005).
2 Religiosity is a multi-faceted phenomenon that consists of three easily identifiable dimensions: namely, belonging (also
called “denomination”), beliefs, and behaviour, or, religious practice (Jagodzinski and Dobbelaere 1995). Focusing on each
of the different dimensions largely depends on the particular object of research, as well as on the available data. Note,
however, that we do not agree with Knutsen (2004), who uses religious denomination to asses the impact of religiosity on
party choice in eight western European countries. Denomination has been proved to be only a cultural element, incapable of
stirring differences among voters that could end up with discernible voting patterns.
7
In short, Portugal is more religious a country than Spain. The data presented in this section define
Portuguese citizens as clearly more faithful and more eager to participate in religious services than their Spanish
peers. To what extent does this distinctiveness lead to differences in the relationship between religiosity and
electoral behaviour? In the following section we will present the results of a multivariate analysis with party
choice as the dependent variable, and religiosity, measured as church attendance, as the main independent
factor. Whether or not this relationship adopts a similar outlook in both countries will be the prime aim of this
effort.
Model specification
We have considered two general elections for each country: on the one hand, the most recent general elections
in Spain (2004), and the last Portuguese legislative election with a post electoral study available (Portugal-2002)
and, on the other hand, the first elections organized after the successful completion of the transition period
(Spain-1982; Portugal-1983).3 We seek to find out now whether the strength of religious voting has changed
across time, and, if so, whether it has changed differently in Portugal and Spain. A logistic regression model
has been assembled for the four surveys. Because small sample sizes badly affect the efficacy of these models
for the smaller parties, we have opted for concentrating on the two large parties in each country. These are the
Partido Socialista Obrero Español (PSOE, Socialdemocrat) and Alianza Popular and later Partido Popular
(AP-PP, conservative) in Spain, and the Partido Socialista (PS, Socialdemocrat) and Partido Social Democrata
(PSD, conservative). Considering this, the dependent variable takes value 1 when the respondents have voted
for any of the two large parties, and value 0 when they have voted for any other party in the opposed ideological
direction. Accordingly, in explaining the vote for, say, the PSOE, Izquierda Unida (IU) voters – generally
located further at the left - are not included in value 0 of the variable; likewise, nationalist voters –often situated
further to the right of the PP - are not included in value 0 in the models for AP-PP. Of course, this applies as
well to the Portuguese case: when discussing the vote for the PS, Bloco de Esquerda and Partido Comunista
Português (PCP) voters will not be included in value 0, while Centro Democrático Social (CDS) voters are
excluded in the models for the PSD.
Religiosity is our key independent variable. Following the standard indicator in the literature (see
Jagodzinski and Dobbelaere 1995), we use church attendance – i.e., religious practice – to measure individual
religiosity. This is possible for Spain-2004 and for the two Portuguese studies.4 As for Spain-1982, the model
3 For Portugal-1983, we have used the data from the Four Nation Study (“The Political Culture of Southern Europe: A Four
Nation Study"), hold in the data bank of the Spanish Centro de Investigaciones Sociológicas (CIS). The Portuguese survey
(CIS study # 1.459) has a sample of N= 2,000. For Spain-1982 we have used the post-electoral survey run by DATA
(N=5,463). As for Portugal-2002, data come from the postelectoral survey undertaken by the Instituto de Ciências Sociais
(ICS), Universidade de Lisboa (N=1,303). Lastly, for Spain-2004, the Demoscopia postelectoral survey has been employed
(N=2,929). The limitations of this study should not be concealed. First and foremost, PP vote is clearly underrepresented.
For reasons that are still unclear, a large part of PP voters do not manifest this option before the pollster.
4 Note that religious practice correlates very highly with the more direct measures of the religious phenomenon, such as
religious beliefs as well as subjective evaluations of religious commitment.
8
includes a mixed indicator of practice and identity, where the respondent is asked to locate herself in a 6-points
scale where feeling “very good catholic” ranks in one extreme, and feeling “an atheist” dwells at the other; the
intermediate values use religious practice to define different intensities of religious commitment. In order to
enhance the efficiency of the models, we have ungrouped the variable, introducing it in dummy format. Also
following the standard recommendations, we have selected as the reference category (RC) a group that is large
in size and meaningful in relation to the dependent variable (Hardy 1993: 10). Thus, the item “never attending
to religious services” is taken as the reference category in the cases of Spain-2004 and Portugal-2003. For
Spain-1982, however, the comparison group – thus omitted from the model – is being “a not very practising
catholic”. Lastly, the item of attending to church “two times per month” is the reference category in the case of
Portugal-1983. In interpreting the results, we must think about the significance of the transit from the reference
category to any other religious category, thus having to focus the attention both to the direction of the
coefficients as well as the level of significance.
We also seek to asses the bearing of morality on voting choices. While moral views have been traditionally
associated with religiosity, the transformations in the religious phenomenon brought about by the secularization
process are casting new doubts as to whether or not moral divisions still reflect religious ones. But it is striking
the scarce attention paid to the electoral consequences of societal moral divides, as well as to the voting
preferences of moral conservatives and moral liberals. For these reasons, a variable measuring attitudes
towards abortion has been included in the models for Spain-1982 and 2004, and Portugal-2003. To the best of
our knowledge, no such variable can be found in electoral studies in Portugal during the 1980s. It is important
to note that the construction of this indicator varies. For Spain-1982 we have employed a variable measuring the
attitudes towards abortion, but in a very specific situation: namely, when pregnancy could cause fatal damage
either to the mother or to the baby herself. However, for Spain-2004 and Portugal-2003 the indicator reports
attitudes towards “abortion on demand” (in both cases a scale ranging from value 1 to value 10). Moral
evaluations are space and time-bounded. What appears perfectly acceptable now was the subject of much moral
hazard in the past. At the beginning of the 1980s, abortion on demand was simply not a possibility either in
Spain or in Portugal. Even more, women taking clandestine abortion and physicians helping them could be by
law criminally prosecuted, and many of them were indeed prosecuted and sentenced to jail. As a maximum, the
debate revolved then around the justification of some scenarios where the general ban on abortion could be
specifically lifted. At present time, however, the discussion is on whether or not women should have the right to
abort freely, regardless the concurrence of extenuating circumstances (rape, risk of malformations, or physical
and mental illnesses).
As for the list of political and social controls, two important political variables as ideology and leadership
evaluation escort religious practice in the models. Whether or not candidates count on voters’ support has
proved to be a decisive explanation for party choice. Particularly in the case of catch-all parties, previous work
shows that a positive evaluation of candidates increases the odds of supporting the party he or she represents
(Gunther and Montero 2001: 129-131; Costa Lobo 2002). As for ideology, a wealth of previous work has
9
identified the self-positioning in the ideological scale as the best predictor of the vote in Spain and Portugal
alike (Torcal and Medina 2002; Gunther and Montero 2001; Sánchez-Cuenca 2003). Although the role of
ideology in predicting party choice is not as stronger in Portugal, it still appears as a formidable variable with
ample capacity to explain electoral preferences (Freire 2001).
Ideology is a particularly sensitive topic in any discussion of religiosity and electoral behaviour. Because it
occurs causally later than most of the independent variables included in conventional regression models, it very
often robes much of the effect of some independent variables on the vote. This has been documented to take
place, for instance, in the case of class voting; in that case, ideology assumes a great deal of the direct effect of
social class on party choice (Bartle 1998). Class, however, exerts an indirect, yet powerful still, effect on
voting (Evans 1999; García de Polavieja 2001). Of course failing to acknowledge the distinction between direct
and indirect effects can give way to misleading interpretations of the results. Particularly in those cases where
religious identities are strongly correlated with ideological positions, the effect of ideology can lead us to
believe that religiosity is not influential when, in reality, it is so (Calvo and Montero 2002). Precisely as a way
to highlight the disruptive effect of ideology, we will show the models in a twofold format: firstly with ideology
included (“full model” henceforth), then without.5 Also, ideology, this time as a dependent variable, will be
regressed on a number of variables, including religiosity. With that we hope to demonstrate that the individual
location in the left-right scale has much to do with his or her religiosity.
The following equations summarize our models:
SPAIN
Equation (1) Y(Voto PSOE1982) = f(Attendace2, Attendance3, Attendance4, Attendance5, Abortion1 Ideology,
Gonzalez’sTher, control variables, ε).
Equation (2) Y(Voto AP1982) = f(Attendace2, Attendance3, Attendance4, Attendance5, Abortion1 Ideology,
Fraga’sTher, control variables, ε).
Equation (3) Y(Voto PSOE2004) = f(Attendance2, Attendance 3, Attendance4, Attendance5, Attendance6,
Abortion1, Ideology, Zapatero’sTher, EconomicPerform, control variables, ε).
Equation (4) Y(Voto PP2004) = f(Attendance2, Attendance 3, Attendance4, Attendance5, Abortion1
Attendance6, Attendance7, Ideology, Rajoy’sTher, EconomicPerform, control variables, ε).
PORTUGAL
Equation (5) Y(Voto PS1983) = f(Attendace2, Attendance3, Attendance4, Attendance5, Ideology, Soares’Ther,
control variables, ε).
Equation (6) Y(Voto PSD1983) = f(Attendace2, Attendance3, Attendance4, Attendance5, Ideology,
Machete’sTher, control variables, ε).
Equation (7) Y(Voto PS2002) = f(Attendace2, Attendance3, Attendance4, Attendance5, Attendance6, Abortion1
deology, Rodrigues’Ther, control variables, ε).
5 The models also include a number of control variables. These are evaluation of economic performance (for Spain-1982 and
2004 and for Portugal-2002), occupational status, family income, education, age, gender, and community size.
10
Equation (8) Y(Voto PSD2002) = f(Attendace2, Attendance3, Attendance4, Attendance5 Attendance6, Abortion1
Ideology, Barroso’sTher, GovernmentPerform, control variables, ε).
Religiosity and party support: observing direct effects
As we introduced before, not every explanatory variable occurs at the same causal moment (Bartle 1998: 501-
502). Some of them, notably political variables such as ideology or party attachment, take shape “later” than
the so-called structural variables (class, religiosity) or the socio-demographic ones (age, gender, etc). This can
create a great deal of problems. In many occasions, intermediate political variables “suck” the explanatory
power of a number of variables that occur causally at an earlier stage, making the observer believe that there are
not causal effects when, perhaps, they actually exist. According to this basic argument, a distinction must be
operated between the direct, and the indirect effects of any given factor on the chosen dependent variable(s). A
variable’s total effect is thus composed of the sum of direct and indirect effects. “Full” models, including
ideology as a control variable, are discussed in this section. Thus, the discussion will revolve around the
magnitude of the direct effects of religiosity on party choice. Clearly, the expectation is that religiosity
performs worse in the full models than in scenarios where the disruptive effect of ideology is eliminated; these
indirect effects will be addressed later.
We begin with the models for Spain-1982 (Table 4; Model A columns). In line with previous findings
(Montero and Calvo 2000: 125), we find a religious voting of moderate strength for the PSOE in the context of
the 1982 general elections. The observation of the second column of Table 4 shows that the transit from being a
“not a very practising Catholic”, the reference category, to the two categories of highest religiosity was
statistically significant, and in the correct direction. According to the expectations, stepping up in the religious
scale jeopardized the odds of PSOE voting in 1982. But the same cannot be said as to the transits towards
categories of lower religiosity. Particularly notorious is the transit towards atheism, which does not take the
expected direction (and is only mildly statistically significant).6 So, at the beginning of the 1980s, and in the
context of a particularly distinctive electoral race (Linz and Montero, 1986), what Jagodzinski and Dobbelaere
(1995) termed “Church integration” operated as a deterrent of socialist support. What about the bearing of
moral views on voting? In the case of PSOE voting in 1982, whether or not the respondent justifies a restricted
legalization of abortion (the hot topic at a time when “abortion on demand” was clearly not an issue), does not
seem to exert an autonomous effect on party choice. While the direction of the coefficient suggests that PSOE
voters tended to hold permissive moral views, the variable has no statistically significant effect. As we will see
6 We have attempted to improve the model specification displayed in a previous analysis of the same election (Montero and
Calvo [2000: Table 8.4]). Apart from the inclusion of the abortion variable, the variable “occupational status” distinguishes
now between “employed”, “unemployed”, “retired”, “student”, and “housewife” (“self-employed” being the reference
category). “Trade union membership” has been eliminated; in contrast, “income” is now in our models. Finally, instead of in
the continuous format, “education” and “age” have been fragmented into dummy variables. Thanks to these changes, the
overall R2 for the PSOE model has improved from 0.41 (N=1,346) to 0.46 (N=962); 85.86 percent of the cases have been
now correctly classified. As for AP vote, the overall explanatory power of the model remains largely the same (0.72 to 0.70).
11
right now, this stands in contrast with the case of AP voting in that same year, when the abortion variable was
significant.7
Table 4
Moving now to AP voting in 1982, it must be noted that, like in the case of the PSOE, a change in the
reference category has revealed some aspects in the relationship between religiosity and conservative voting
that were not obvious in our previous research (see Montero and Calvo 2000).8 Most interesting of all is the
certain symmetry that we have found between the models for this Spanish election. The model for the PSOE
tells the story of those that were more determined to resist the wave of socialist support that swept the Spanish
electorate in 1982: religious people appeared less eager to opt for the PSOE than any other social group.9 A
similar argument applies to the case of AP voting: while no clear orientations are found as to whether or not
religiosity determined party support, the model helps to identify those that would be hardly ready to vote for
this political party. Observing the third column of Table 4, the transit from the reference category to two of the
less religious categories is statistically significant and in the correct direction. Non-religious people did not like
AP in closely the same way that religious people were wary of the PSOE. However, the transit towards the
highest religious categories did not look determinant for the prospects of AP support. Moreover, finding a
negative coefficient in the transit towards the highest religious category is perplexing, to say the least.
Sticking to a principled condemnation of abortion was beneficial for AP voting. These results introduced
one of the defining features of Spanish politics as far the issue of morality is concerned: while the right
galvanises the support of the morally conservative, the left seems unable to do so. Conservative voting attracts
a relatively much higher share among moral conservatives than leftist voting does in its quest to garner the
support of moral liberals. Interestingly, moral liberals seem to be more flexible at the time of casting the vote,
displaying an observable tendency towards supporting several political parties, including those towards the left
of large social democratic parties, i.e., the Communist party, or Izquierda Unida (its new organizational format
since 1986).
In short, at the level of direct effects, we have found a mild, but important, effect of religion on the vote in
Spain-1982. Religiosity seems to be linked with a form of “negative voting”. In other words, if religiosity was
not strong enough to determine the vote, it, however, showed a capacity to set limits in the range of plausible
alternatives. Let us move to the models for Portugal-1983 (Model A columns in Table 5). The probability of
voting for the PS (against the probability of voting for any other party further to its right) was not very much
7 In this model, both income and education demonstrate a high predictive capacity for PSOE voting (higher levels in both
variables representing a deterrent of PSOE support).
8 There, the reference category was “attending to church every Sunday”.
9 Of course this does not mean that all religious people, at the time close to 35 per cent of the population, voted conservative.
In the 1982 “earthquake” electoral context, the PSOE swept the board virtually in every constituency considered. What we
claim is that religious people were among the more immune to an otherwise overwhelmingly powerful trend.
12
affected by religiosity in the 1983 context. Meeting the expectations, the coefficients organize a scenario where
PS voters tend to be less religious than those voting for rights parties. However, the effect is not strong in terms
of statistical significance, as only the transit from the comparison group to the least religious category is
statistically significant (and only at the confidence level of 90 per cent). In contrast, climbing up the religious
lather increased the odds of PSD voting in 1983, as the significance of the transits from the reference category
towards the highest religious categories demonstrates. This relationship, however, is not a robust one. While
higher levels of religiosity favoured the electoral fortunes of these parties, the model also tells that a decreasing
religiosity might also help the PSD vis-à-vis any other party towards its left. These latter transits are not
statistically significant; still, the surprising direction of the coefficients makes us suspect that something is
distorting the effect of religiosity on this kind of voting.
Table 5
In any case, the overall explanatory capacity of the model is constrained in the Portuguese case, particularly
as far as PSD voting is concerned. In both countries, religiosity failed to bring about strong consequences on
voting choices during the early 1980s. Neither in Portugal, nor in Spain electoral competition revolved around
religiosity for reasons either contextual or structural. While in Spain the relevance of religiosity was
overshadowed by the ability of the PSOE to receive a sort of universal vote in the extremely realigning and
volatile 1982 elections (Linz and Montero 1986; Gunther and Montero 2001: 121-122), in Portugal voters of
every religious group made roughly similar choices between the two larger parties (Bacalau 1994: 65 ff).
Religious identities did not lead to strong electoral alignments, and hence the competition to gain the specific
support of religious constituencies was not stiff. Notwithstanding this, we have found meaningful differences
between the two cases compared. If in Spain religiosity showed some capacity to set limits on the plausible
alternatives of voter, we have come to the conclusion that, in Portugal, religiosity was simply unimportant as a
predictor of voting decisions. We comment further on these differences in the last section of the paper.
Let us move now to the most recent elections with post-electoral surveys available: namely, Spain-2004 and
Portugal-2002 (Model A columns in Tables 6 and 7). A highly significant religious variable commands a model
for PP-vote – then the incumbent party - that shows great efficacy in explaining the probability of voting for this
party. Even when controlled by ideology, virtually every religious category shows an autonomous effect on the
dependent variable.10 Of course this invites further questioning as to what has changed in the context of
conservative voting in Spain, which explains such a remarkable transformation in the salience of the religious
variable. What has not changed is the strong effect of moral views on conservative voting in Spain. In 2004, as
in 1982, holding negative views about the legalization of abortion increases the likelihood of PP voting. Moral
conservatives, who happen to be the most religious, persist in their loyal support to the right.11 Yet the same
10 Note that in these elections, the RC is the lowest religious category.
11 Economic performance, gender, and, of course, ideology, and Rajoy’s evaluation contribute also to explain the decision of
voting for the PP in 2004. Given that in 2004 the PP was the incumbent party, positive evaluations of the economy
reinforced the tendency of supporting the continuity of the government.
13
does not hold true in the other side of the continuum: replicating the situation at the beginning of the 1980s, the
PSOE still fails to conquer the minds of the moral liberals. In contrast with the notorious influence of religion
on the vote for the PP in 2004, the decision to choose the PSOE vis-à-vis any other electoral choice further to its
right seems to have a weak religious component. Considering the full model, the data presented in Table 6
confirms the existence of only a mild relationship between religiosity and PSOE vote (only one transit is
statistically significant). As a matter of fact, PSOE voting is better explained by looking at other variables, such
as the assessment of the economy, which is again highly significant, or, of course, ideology. Similarly, income
is negatively related with this kind of vote, and so education is; higher educational status lead voters towards
more conservative orientations.
Tables 6 and 7
In sharp contrast, a quick perusal at the results of Table 7 confirms that, in contemporary Portugal,
religiosity continues to be a weak predictor of the vote. At least when the question is why Portuguese citizens
opt among the two largest political parties, in a context of a party system that is more fragmented than the
Spanish one, the answer still is not to be found within the confined of societal religious divides. While in Spain
the PP and the PSOE have electoral constituencies more or less well defined in religious terms, (something that,
undoubtedly, impinges on policy making), in Portugal none of the religious categories show any statistical
significance. Both PSD and PS voting appear immune to the bearing of religiosity, composing a picture where
religiosity has lost any capacity for shaping the electoral preferences of Portuguese voters. Equally relevant, but
perhaps not that surprising, is the confirmation that moral views are neither able to predict the vote for the two
largest parties in Portugal. Neither the PS, nor the PSD are garnering the support of the moral liberals. In fact,
both parties compete for the support of the conservatives, who happen to outclass moral liberals in
contemporary Portugal. Thus, we believe that morality, as religiosity, does not organize the competition among
the two main Portuguese political parties. Both issues matter, but as a way to organize the voting decision
between large, centrist political parties (PS, PSD) and the smaller, more extreme-oriented ones (Bloco on the
left, CDS on the right).
Religiosity and party support: indirect effects
Religiosity is a textbook example of a variable apparently incapable of bring about direct effects. Being so
closely related with key political controls, such as ideology (see below), it often gives the impression of being
unimportant, when it is not so. Conscious of the potential magnitude of this problem (see Montero and Calvo
2000), we have devised a twofold strategy to reveal the indirect effects of religiosity on voting that remained,
perhaps, hidden in the discussion of direct effects. Firstly, we proceed with a simple exercise that consists of
deleting ideology from Tables 4, 5, 6 and 7. If the indirect effects hypothesis holds, we should witness
noticeable changes in the behaviour of the religious categories as soon as ideology is expelled from the models.
Naturally, our interpretation goes in the direction of insisting on the changes produced by this action in the other
14
explanatory variables. We do not consider whether or not the models perform better or worse as a whole.
These results are offered in the Model B columns in Tables 4 (Spain-1982), 5 (Portugal-1983), 6 (Spain-2004)
and 7 (Porgual-2002). Secondly, a linear regression model has been assembled to explain ideology (Table 8).
From a certain theoretical standpoint, ideology is understood as a composite variable that summarizes a number
of individual’s characteristics (Knutsen 1995, 1997, 1998). It is expected that the more religiosity translates
into ideological positions, the lower its capacity to exert direct effects. Thus, our chief aim is to find out
whether being “leftist” or “rightists”, is connected with the values of the respondent (or, alternatively, whether
ideology depends on other factors).
What happens if ideology is not included in the models as a political control? For one, the models collapse.
Given the proven strength of ideological voting in both countries (Gunther and Montero 2001), it is not
surprising that the overall explanatory capacity of the models drops as a consequence of this decision. Indeed,
the R2 falls dramatically in the Spanish and the Portuguese surveys alike. Nevertheless, it is the impact on the
significance of the independent variables what should interest us now. As for Spain-1982, as soon as ideology
departs, religiosity is recuperated as an important predictor of PSOE voting (Table 4). Without ideology,
religiosity gains new strength to explain not only negative voting preferences, as we saw before, but also
positive ones. For instance, the transit from the comparison group towards the less religious categories becomes
significant, confirming that a decreasing religiosity helped PSOE voting in 1982, but only when translated into
leftist ideological positions. Equally significant is the transformation in the model for AP in the same year:
without ideology, religiosity emerges as a very powerful predictor of its vote, satisfying the expectation of AP
as the natural destination of religious voters (and an unlikely one for non-religious). In the case of Portugal,
however, no indirect effects of similar magnitude are revealed (Table 5). Despite the positive consequences of
the elimination of ideology on the significance of some factors – occupation and even religiosity itself for the
PS -, the results as to the bearing of religiosity do not change very dramatically. Particularly surprising is the
limited effect in the significance of the religious variable in the vote for the PSD. Being ideology expelled, the
coefficients do take now the expected direction. Also, the transit from the reference category to the highest
religious category gains in statistical health. However, scaling down in the level of religious commitment
continues to be unimportant as regards PSD voting.
As for Spain-2004, the elimination of ideology still stirs spectacular consequences on the performance of
religious categories (Table 6). Note for instance the significance, at the highest level, of every religious
category in the explanation of PP voting. Also, abortion seems to be gaining new vigour as political controls
are temporarily eliminated. All this confirms that the interweaving of religious and ideological identities is, at
least in the case of conservative voters, a defining feature of the Spanish politics. Conservative voting has a
distinctive religious element that, however, often fails to shine in due regard as it takes the face of conservative
ideological positions. Leftist voters, on the contrary, are coming to display a less cohesive religious profile.
The results presented in Table 6 confirm that, in the case of PSOE voting, the problem of hidden effects has
softened somehow with the passing of time. While the departure of ideology clearly benefits the performance
of religion as a factor to explain PSOE vote, the variable fails to acquire a profile as a good predictor of this
15
kind of voting. In any case, if ideology is not present in the models, higher levels of religiosity decrease the
likelihood of PSOE voting. This is a statistically significant effect.
Does the elimination of ideology transform the role of religiosity in the explanation of voting choices in
Portugal-2002? The answer goes clearly in the negative. The striking continuity in the (reduced) significance of
the religious variables after eliminating religiosity applies to religion and morality alike. Thus, we must
conclude that the absence of discernible direct effects of religiosity on the vote in Portugal is substantive, rather
than technical. While in Spain there are reasons to believe that religiosity matters, but through the definition of
ideological identities that, in turn, shape voting choices, in Portugal the story is a different one: at least as far as
the vote for the largest parties is considered, religiosity is simply not a factor anchoring this type of vote.
Having completed half of the task, we proceed now with the second strategy introduced at the beginning of
the section. Table 8 shows the result of a model where ideology is regressed on a number of variables: namely,
the respondent’s recollection of his or her mother’s ideology, church attendance (again in dummy format),
gender, age, education (measured as number of years in education), occupation (in dummy format), and family
income (taken as a continuous variable).12 On any account this is an incomplete model specification,
particularly given the emphasis that the literature is placing on ideology as determined by a shifting and
evolving political environment (for a review see Torcal and Medina 2002). Yet we are not interested in the
partisanship element of ideology. Much to the contrary, our interest points at the relationship between ideology
and variables that occur causally at an earlier stage. In short, when voters locate themselves in the ideological
continuum, to what extend are they reflecting religious value orientations?13
Table 8
The model performs nicely. Firstly, it shows a high overall predictive strength (with an R2 of 0.46).
Secondly, most of the coefficients included are statistically significant, religiosity being one of them. At the
maximum level of statistical significance, every transit from the reference categor towards higher religiosity
pushes the respondent towards the right. Such a result clearly confirms that religiosity and ideology are closely
related. To a considerable extent, ideological identities respond to previously conformed religious value
orientations. Particularly in the case of AP/PP voting in Spain, high levels of religiosity lead voters to rightist
ideological positions that, in most cases, will represent PP voting. But the same applies to the choices of non-
believers. Feelings of disassociation from religion will be converted into leftist ideological positions that, again,
suggest leftist voting.
12 Having found only very weak indirect effects in the Portuguese case, we have decided to limit this exploration to the
Spanish case.
13 As has been said, the model includes a variable measuring the respondent’s recollection of his or her mother’s ideology.
That is meant to tap the well-known socialization argument, which sees ideology (and party identification) as the
consequence of a socialization process taking part within the family. Previous work in this subject has found that, as far as
the transmission of ideological views is concerned, the bonds are closer between sons and their mothers, rather than between
sons and their fathers (Jaime 2000).
16
The conclusion of this section is twofold. Firstly, in order to capture the true magnitude of the bearing of
religiosity on party choice, the analysis must go beyond the observation of direct effects. And secondly, the
importance of this kind of effects is uneven. In Portugal, the weakness of religiosity as a predictor of the vote
does not conceal indirect effects of considerable magnitude. In Spain, however, religiosity and ideology trace a
complex interaction that distorts the influence of the former on party choice.
Equal but different
In this paper both important similarities as well as striking differences as regards the relationship between
religion and politics have been identified. Perhaps the single most important feature that Spain and Portugal has
in common is that, in spite of a tradition of internecine disputes around religion, the potential for a pervasive
religious cleavage has never been activated. In the late 1970s, the social climate in both Portugal and Spain
contained some potential for pervasive religious conflicts (Gunther and Montero 2001; Montero 1997). This is
particularly seen in the virulent anti-clericalism associated with the Republican experience in both countries,
which in part explained their eventual demise and replacement by enduring authoritarian regimes that enjoyed
Church support. It could have been perfectly possible for the new political elites to politicise a number of
religious issues that simply lay dormant.
Religious peace, however, replaced conflict, and the foundations for a softer relationship between
religiosity and political behaviour were established in both countries already in late 1970s. We have previously
explained the de-activation of religious conflicts during these crucial times in terms of the role of political elites
as cleavage (de)activators (Montero and Calvo 1999). As it is commonly known, the impact of processes of
social change on cleavage consolidation is usually framed by the widely-accepted threefold conceptual schema
by Bartolini and Mair (1990: 212 ff.). In terms of the religious cleavage, the most immediate impact of social
change is said to be the elimination of religious differences at the societal level. This blurring of religious
divisions at the societal level has, according to this theory, an immediate impact on the formation of political
identities. As there are no clearly defined religious social groups, religiosity can hardly be expected to
determine people’s understanding of politics. Consequently, and this brings us to the third dimension of this
schema, confessional and religious parties (that is, the organisational expression of existing religious divides at
the societal level) virtually eliminate their programmatic references to religious issues. This strategy eventually
explains why these issues have ceased to be an object of confrontation in the partisan arena. In short, the erosion
of religious subcultures, paralleling the process that affected the traditional working class, should lead to party
choices being increasingly based on individual preferences rather than on collective identifications (Dogan
1996; Gallagher, Laver, and Mair 1995: 225).
Convincing as the threefold sequence appears to be, it says little about the mechanisms that are supposed to
set these dynamics in motion. Moreover, in the view of existing assumptions, cleavages are said to solely
reflect objective interests and identities, implying that the strength of their correlation with individuals’ electoral
17
choices is merely a function of particular individuals’ position in the social structure. We, in contrast, believe
that political agency matters. In more cases than it is usually recognised, given a certain distribution of religious
traits among the citizenry, political actors (the complex web of organisations, parties, elites, as well as their
preferences, strategies, exchanges, decisions, discourses) are in a position to shape from above, with differing
but nonetheless discernible degrees of autonomy, the religious cleavage (Sartori 1969: 89). Paraphrasing the
argument originally made by Przeworski and Sprague (1986: 10-11, 143) and developed for the analysis of the
formation of Christian democratic parties by Kalyvas (1996: 8-9), we claim that the religious conflict is salient
in a given society if, when, and only to the extent to which it is important to political parties which mobilise
religious or secular citizens. During the consolidation of democracy, Portuguese and Spanish elites opted not to
centre on the religious cleavage. This explains the mild effects of religious voting on the early elections
considered in this paper. Using the Portuguese case as an illustration, during the transition period the Church
was tainted by the association with the authoritarian regime of Salazar, even if the role of Catholic groups in the
opposition to Salazar partially restored legitimacy. As Pasquino (1990: 49) points out, “Catholics were
unwilling to engage, qua Catholics, in politics”, and we should add that Catholics leaders were even less willing
to do so. On the other hand, the religious scab was less fresh than across the border – inter alia, due to the
length of the Salazarist regime, its aversion to radicalism (which precluded a virulent backlash and reversal of
the republican project), and notably the relative peaceful demise of the Portuguese First Republic, when
compared to the Civil War of its Iberian neighbour. Established as a moribund cleavage, religiosity only but
receded as an explanatory factor for party choice in both countries during the 1980s. The religious peace
insinuated during the transition years evolved into a quasi-structural definition of the political role of religiosity.
Without any salience in the electoral competition, the soft approach that governed the relationship between the
Church and the governments in Portugal and Spain simply eliminated the need to think in religious terms at the
time of casting the vote.
Nevertheless, that religion is prevented from representing a major cleavage does not preclude religiosity
from influencing party choice in a significant fashion. In short, religious voting can be found in contexts of
weak, or even non-existent religious cleavage. This is the terrain where the comparison between Spain and
Portugal shines as an illuminating example of the different intensities that religious voting can adopt in context
of institutional religious peace. For one, as we have seen, religious voting is more salient in Spain than in
Portugal. With different intensities according to political parties as well as to time point, we have nonetheless
found a consistently strong association between religiosity and, particularly, conservative voting in Spain. In
Portugal, however, neither conservative, nor leftist voting seems to be linked with religion in any meaningful
way. To explain these differences, the founding processes by which the party systems in both countries, as
well as the initial alliances between voters and parties, should be considered. Despite the expressed
determination on the part of both political and religious elites not to politicise religion (Linz, 1993), Spanish
voters were offered a limited range of electoral options (at least narrower than in Portuguese case) that,
moreover, could be roughly classified along religious terms. Building on the manifested ties with the recent
Francoist past, and helped by the ideological and moral profiles of the party’s governing elite, voters could
attribute a certain religious identities to AP that was different from the one conferred on the PSOE or the
18
Communist Party. Portuguese voters, however, faced a more complex task. Exposed to a much diversified
party system, and in a context where the largest political parties only seemed to differ in the issue of regime
type (Jalali, 2002), the possibility of classifying the PSD and the PS according to religion was largely
impossible. The evolution of political events in both countries are only but exacerbating these differences.
Whereas in Spain, a number of policies relating to religious education, moral issues or Church-state relations
raised the profile of the PP as a morally conservative, religious political party (against the morally liberal stance
of the PSOE), in Portugal both the PS and the PSD continue to share basic views in relation to crucial religious
and moral problems.
It is interesting to note that all this has resulted in notorious differences in the religious profiles of the
different political parties. According to the data provided in figures 1 and 2, the two largest Portuguese political
parties soon exhibited very similar religious profiles. In both cases those that we call “nuclear Catholics”
represented more than fifty per cent of their voters. In a country that has systematically ranked among the more
religious in Europe, the main political parties compete for a very large part of the electorate that, however, is
largely uniform in religious terms. A different argument, of course, should be elaborated to explain the choice
between the large and the extreme Portuguese political parties. In Spain, however, the religious profiles of AP
and the PSOE were very different from the onset. In spite of the fact that in 1982 the PSOE outdid AP in every
religious category, it was among the more religious where the latter party managed to perform better.
Figures 1 and 2
A second difference between our cases compared relates to the definition and evolution of the triangular
relationship conformed by religiosity, ideology, and party choice. One key finding of this analysis is that
whereas in Portugal ideology does not obstruct the bearing of religiosity on party choice, in Spain it clearly does
so. It is the fact that ideology is not obstructing the effect of religiosity what accounts for the absence of
indirect effects in the Portuguese case. Tables 9 (Spain) and 10 (Portugal) display three-way contingency tables
with religiosity, ideology and party choice as variables. The third and forth columns summarize the voting
preferences of the different religious groups. Such basic information reveals again interesting differences
between Spain and Portugal. While in the former case, religious voters have adopted more distinctive voting
intentions as time has passed by, in the latter the irrelevance of religion has become even more apparent: in
2002, every religious group split evenly between the two large options. But nothing of the sort has happened in
the march 2004 general elections in Spain; in this case, nuclear Catholics (around 20 per cent of the total
population) opted for the PP, while Non-believers (around 40 per cent) preferred the PSOE.
The tables also confirm that religiosity is not bounded to ideology in Portugal in the same way that it is so
in Spain. For each of the percentages resulting after cross-tabulating religion and party choice, the table
displays a distribution according to ideological categories. For instance, the percentages shown in the fourth
row of Table 9 represent the ideological distribution of those nuclear Catholics that voted for AP in 1982 (28
per cent), and for the PP in 2004 (40 per cent). We are interested to see whether or not, for each different party,
19
the distribution adopts a different shape according to religion. Should this be the case, we could conclude that
ideology is closely linked with religion. Otherwise, the conclusion will be that ideological identities are formed
regardless religious identities, thus suggesting a definition of ideology free of value orientations. As expected,
the contrasts are striking. While in Portugal ideological distributions for each political party replicate the same
pattern across the different religious groups, in Spain the shape of this distribution seems to depend highly on
the degree of religiosity. An example based on PP voting on 2004 will illustrate the point. In the case of those
nuclear Catholics who voted for the PP, ideological conservatism (right and extreme right) amounted to 60 per
cent of the category. In contrast, only 47 per cent of those non-believers had voted for the PP were
ideologically conservative. Variations of this kind cannot be found in the Portuguese case.
Tables 9 and 10
Concluding remarks
Portugal and Spain present an interesting counterpoint in terms of electoral behaviour. Despite evident
similarities anchored in history, culture, and geography, these two countries have come to develop different
scenarios for the development of religious voting. Indeed, our models seem to challenge the perception of
Iberian countries as split along religious lines. In the context of a recently completed transition period, these
elections took place in an environment were religious conflicts had been deliberately silenced. Moreover, the
identification with strong religious identities were also explicitly excluded in the strategies followed by party
leaders when building ex novo new party systems in both Iberian countries. Consequently, religious identities
were doomed to play only a secondary role in the definition of electoral alignments at the beginning of the
1980s. All that notwithstanding, our data confirm the existence of a religious voting of considerable strength in
Spain. Perhaps without becoming determinant, we have seen that religiosity, particularly in the case of the
diminishing group of nuclear Catholics, still help voters to opt among political parties.
In this paper we have shown that the differences between Spain and Portugal operate at different levels. On
the one hand, the different role of religiosity on voting is linked with “substantive” realities, such as diverging
party systems or distinctive religious maps. On the other hand, however, we have also found interesting
“operational” differences in the mechanisms that translate religious identities into party choice. In Spain,
ideological identities incorporate value orientations to a great extent, in Portugal the location in the left-right
scale appears to be less value-dependent. While a careful examination of why both “left” and “right” means
different things in different places escapes the limits of this paper, we have come to the conclusion that a great
deal of the differences between the cases has to do with varying definitions of ideological identities. In Spain,
to a considerable extent, religious identities translate into ideological identities. In Portugal this does not
happen.
20
21
References
Bacalhau, Mário. 1994. Attitudes, Opinioes e Comportamentos Políticos de los Portugueses: 1973- 1993.
Lisboa: Fundaçao Luso Americana para o Desenvolvimento.
Bartolini, Stefano, and Peter Mair. 1990. Identity, Competition, and Electoral Availability. The Stabilisation of
European Electorates, 1885-1985. Cambridge: Cambridge University Press.
Bartle, John. 1998. "Left-Right Position Matters, But Does Social Class? Causal Models of the 1992 British
General Election." British Journal of Political Science 28, no. 3: 501-30.
Calvo, Kerman, and José Ramón Montero. 2002. “Cuando ser conservador ya no es un problema: religiosidad,
ideología y voto en las elecciones generales de 2000”. Revista Española de Ciencia Política 6: 17-57.
Calvo, Kerman, and José Ramón Montero. 2005. “Valores y religiosidad”, en España: sociedad y política en
perspectiva comparada, ed. Mariano Torcal, Laura Morales and Santiago Pérez-Nievas. Valencia: Tirant
lo Blanch
Converse, Philip E. 1974. “Some Priority Variables in Comparative Research”. In Electoral Behavior: A
Comparative Handbook, ed. Richard Rose. New York: The Free Press.
Costa Lobo, Marina. 2002. “O Impacto de los Lideres Partidários: Una Escolha entre Candidatos poco
Populares”. In Portugal a Votos. As Eleicoes Legislativas de 2002 ed. André Freire. M. Costa Lobo, and
Pedro Magalhaes. Lisbon: Impresa de Ciaencias Sociais.
Dalton, Russell J. 1996. “Political Cleavages, Issues, and Electoral Change”. In Comparing Democracies.
Elections and Voting in Global Perspective, ed. Lawrence LeDuc, Richard G. Niemi, and Pippa Norris.
London: Sage Publications.
Dalton, Russell J., Scott C. Flanagan, and Paul A. Beck, eds. 1984. Electoral Change in Advanced Industrial
Democracies: Realignment or Dealignment? Princeton: Princeton University Press.
Dogan, Mattei. 1996. “Erosion of Class Voting and of the Religious Vote in Western Europe”. International
Social Science Journal 146
Evans, Geoffrey, ed. 1999. The End of Class Voting? Class Voting in Comparative Perspective. Oxford:
Clarendon Press.
Evans, Geoffrey. 2000. “The Continued Significance of Class Voting”. Annual Review of Political Science
2000: 401-417.
Franklin, Mark, Thomas T. Mackie, Henry Valen, et al. 1992. Electoral Change. Responses to Evolving Social and
Attitudinal Structures in Western Countries. Cambridge: Cambridge University Press.
Freire, André. 2001. Mudança Eleitoral em Portugal. Cleavages, Economia e Voto em Eleiçoes Legislativas, 1983-
1999. Oeiras: Celta Editora.
Gallagher, Michael, Michael Laver, and Peter Mair. 1995. Representative Government in Modern Europe. New
York: McGraw-Hill, 2nd edition.
García de Polavieja, Javier. 2001. “¿Qué es el Voto de Clase? Los Mecanismos de Voto de Clase en España”.
Zona Abierta 96/97: 173-213.
Gunther, Richard, and Montero, José Ramon. 2001. “The Anchors of Partisanship: A Comparative Analysis of
Voting Behaviour in Four Southern European Democracies”. In Parties, Politics, and Democracy in the
New Southern Europe, ed. P. Nikiforos Diamandouros and Richard Gunther. Baltimore: Johns Hopkins
University Press.
22
Jagodzinski, Wolfgang, and Karel Dobbeleare. 1995. “Secularization and Church Religiosity”. In The Impact
of Values, ed. Jan W. van Deth and Elinor Scarbrough. Oxford: Oxford University Press.
Jaime Castillo, Antonio M. 2000. "Familia y Socialización Política. La Transmisión de Orientaciones
Ideológicas en el Seno de la Familia Española." Revista Española de Investigaciones Sociológicas 92: 71-
93.
Jalali, Carlos. 2002. “The Evolution of the Portuguese Party System in Comparative European Perspective since
1974”. Oxford: University of Oxford, D. Phil. thesis.
Kalyvas, Stathis N. 1996. The Rise of Christian Democracy in Europe. Ithaca, NY: Cornell University Press.
Knutsen, Oddbjorn. 2004. “Religious Democratization and Party Choice in Western Europe: A Comparative
Longitudinal Study from Eight Countries, 1970-97.” International Political Science Review 25 (1): 97-128.
Knutsen, Oddbjorn. 1995. “Party Choice”. In The Impact of Values, ed. Jan van Deth and Elinor Scarbrough.
Oxford: Oxford University Press.
Knutsen, Oddbjorn. 1997. "The Partisan and the Value-Based Component of Left-Right Identification: A
Comparative Study." International Political Science Review 18: 191-225.
Knutsen, Oddbjorn. 1998. "Experts Judgements of the Left-Right Location of Political Parties: A Comparative
Longitudinal Study." West European Politics 21: 63-94.
Kotler-Berkowitz, Laurence A. 2001. “Religion and Voting Behaviour in Great Britain: A Reassessment”. British
Journal of Political Science 31: 523-554.
[0]Lijphart, Arend. 1971. “Class Voting and Religious Voting in the European Democracies: A Preliminary
Report”. Glasgow: University of Strathclyde, Occasional Paper 8.
Lijphart, Arend. 1980. “Language, Religion, Class and Party Choice: Belgium, Canada, Switzerland and South
Africa Compared”. In Electoral Participation. A Comparative Analysis, ed. Richard Rose. Beverly Hills
and London: Sage Publications.
Linz, Juan. 1967. “The Party System of Spain: Past and Future”. In Party Systems and Voter Alignments: Cross-
National Perspectives, ed. Seymour M. Lipset and Stein Rokkan. New York: The Free Press.
Linz, Juan., and José Ramón Montero, eds. 1986. Crisis y Cambio: Electores y Partidos en la España de los
Años Ochenta. Madrid: Centro de Estudios Constitucionales.
Linz, Juan. 1993. “Religión y política en España”. In Religión y sociedad en España, ed. Rafael Díaz-Salazar
and Salvador Giner. Madrid: Centro de Investigaciones Sociológicas.
Lipset, Seymour M., and Stein Rokkan. 1967. “Cleavage Structures, Party Systems, and Voter Alignments: An
Introduction”. In Party Systems and Voter Alignments: Cross-National Perspectives, ed. S. M. Lipset and
S. Rokkan. New York: The Free Press.
Montero, Jose Ramón. 1997. "Secularization and Cleavage Decline: Religiosity, Electoral Behaviour, and
Generacional Change in Spain". Manuscript.
Montero, José Ramón, and Kerman Calvo. 2000. “An Elusive Cleavage? Religiosity and Party Choice in Spain”, en
David Broughton y Hans-Martien ten Naipel, eds., Religion and Mass Electoral Behaviour in Europe.
Londres: Routledge.
Pasquino, Gianfranco. 1990. “Party Elites and Democratic Consolidation: Cross-national Comparison of
Southern European Experience”. In Securing Democracy: Political Parties and Democratic
Consolidation in Southern Europe, ed. Geoffrey Pridham. London: Routledge.
23
Przeworski, Adam, and John Sprague. 1986. Paper Stones: A History of Electoral Socialism. Chicago:
Chicago University Press.
Rose, Richard, and Derek Urwin. 1969. “Social Cohesion, Political Parties and Strains in Regimes”.
Comparative Political Studies 2: 7-67.
Sánchez-Cuenca Ignacio. 2003. "How Can Governments Be Accountable If Voter Vote Ideologically?" 191.
CEACS, Estudios/Working Papers, Instituto Juan March, Madrid-
Sartori, Giovanni. 1969. “From the Sociology of Politics to Political Sociology”. In Politics and the Social
Sciences, ed. Seymour M. Lipset. New York: Oxford University Press.
Torcal, Mariano, y Lucía Medina. 2002. “Ideología y voto en España 1979-2000: los procesos de
reconstrucción racional de la identificación ideológica”. Revista Española de Ciencia Política 6: 57-96.
24
Table 1. Subjective religiosity in Europe, 2002-2003 a
Average
Median
Mode
Standard
deviation
N
Greece 7.7 8 10 2.2 2,566
Poland 6.6 7 8 2.4 2,110
Italy 6.1 6 6 2.5 1,207
Ireland 5.8 6 5 2.4 2,046
Portugal 5.7 6 5 2.5 1,511
Finland 5.6 6 7 2.6 2,000
Austria 5.2 5 5 3 2,257
Switzerland 5.2 5 5 2.8 2,038
Netherlands 5.1 6 7 2.9 2,364
Belgium 5.0 5 5 3.0 1,899
Slovenia 4.9 5 5 2.8 1,519
Israel 4.7 5 5 3.1 2,499
Denmark 4.4 5 5 2.5 1,506
Hungary 4.4 5 5 3.0 1,685
Spain 4.4 5 5 2.7 1,729
United Kingdom 4.3 5 5 2.8 2,052
Germany 4.2 5 0 2.9 2,919
Luxemburg 4.1 4 0 3.1 1,552
Norway 4.1 4 5 2.5 2,036
Sweden 3.7 3 0 2.8 1,999
Czech Republic 3.1 2 0 2.9 1,360
European Average 5 5 5 2.9 40,574
a Figures are means in scales of self-definition of religiosity, where value 0 is “not being religious at all”, and
value 10 implies being “very religious”. Countries are sorted according to the means.
Source: European Social Survey (ESS), 2002.
25
Table 2. Selected indicators of religiosity in Spain (1981, 1990, and 1999) and Portugal (1999) (in percentages)
Spain Portugal
Indicator 1981 1990 1999 1999
Religiosity
As a person you are…
Religious 63 63 56 82
Not religious 30 28 33 14
Atheist 4 4 6 3
You consider yourself as…
Very good catholic + practising catholic 37 30 29 -
Not very practising catholic 27 28 28 -
Not practising catholic 19 26 25 -
Indifferent 10 12 12 -
Not believer, atheist 4 4 6 -
Other religion 1 1 1 -
Importance of religion in your life…
A lot of + quite - 54 42 71
Little + not any - 45 58 29
Beliefs
Belief in…
God 87 81 81 93
Life after dead 55 42 40 47
Heaven 50 48 42 63
Hell 34 27 27 35
Sin 58 57 44 73
Importance of God (means in scale 1 to 10) 6.39 6.25 5.97 7.53
You find comfort and strength thanks to religion 57 53 49 76
You have moments of praying and meditation 60 61 61 71
Religious practice
Attending to church
Once per week or more 41 33 25 40
Once per month 12 10 10 12
Sometimes 10 17 9 8
Never or almost never 36 40 56 40
Church
You have confidence in Church
A lot of + quite 50 53 41 73
No much or none at all 49 47 57 27
You think Church is giving proper answers to...
Moral and individual problems 39 39 33 58
Problems in family life 34 38 29 52
Spiritual necessities of the people 45 49 48 69
Social problems in the country - 33 23 41
(N) (2.305) (2.637) (1.200) (1.895)
Sources: For Spain, World Values Surveys (WVS), 1981, 1990, and 1999. For Portugal, WVS, 1999.
26
Table 3: Frequency of church attendance in Europe, 2002 a (in percentages)
Country
At least once a
week
At least once a
month
Occasionally
Almost never or
never
Poland 56 19 17 8
Ireland 54 13 10 23
Italy 31 12 20 37
Portugal 29 15 8 48
Greece 27 28 28 16
Spain 21 9 17 53
Slovenia 20 10 31 39
Israel 19 7 27 47
Austria 18 15 19 48
Luxembourg 15 10 17 58
United Kingdom 13 6 12 68
Netherlands 12 9 13 66
Switzerland 11 12 27 50
Hungary 11 7 22 59
Belgium 11 8 15 66
Czech Republic 9 5 16 70
Germany 8 10 21 60
Norway 5 6 24 65
Sweden 5 6 17 72
Finland 5 7 25 64
Denmark 3 7 21 69
Total 18 11 20 51
a The question is as follows: “Apart from special occasions, such as weddings and funerals, about how often do
you attend religious services nowadays? Every day; More than once a week; Once a week; At least once a month;
Only on special holy days; Less often; Never”. Countries are ordered by the highest frequency.
Source: ESS, 2002.
27
Table 4: Logistic regressions to explain party vote in Spain 1982a
Model A: With ideology Model B: Without ideology
Variables Vote PSOE Vote AP Vote PSOE Vote AP
Religious attendance (RC: Not very practising
Catholics)
Atheist -0.4299* -2.5264*** 0.3682 -2.7601***
(0.2573) (0.8576) (0.2483) (0.7008)
Indifferent 0.5403** -1.2621** 1.0482*** -1.7279***
(0.1911) (0.5625) (0.1784) (0.4584)
Not practising Catholics 0.1310 -0.4838 0.3917*** -0.7039***
(0.1542) (0.3809) (0.1471) (0.2676)
Practising Catholics -0.7101*** 0.2824 -0.8921*** 0.7558***
(0.1495) (0.3017) (0.1421) (0.2221)
Very good Catholics -0.9271*** -0.7859 -1.2933*** 0.4538
(0.2799) (0.5388) (0.2577) (0.3270)
Attitudes towards abortion -0.1225 0.6790*** -0.0966 0.7075***
(0.1091) (0.2490) (0.1029) (0.1824)
Ideology -4.3837*** 13.4816***
(0.3828) (1.1561)
Manuel Fraga’s (AP) thermometer 7.6804*** 9.5461***
(0.7218) (0.5889)
Felipe Gonzalez’s (PSOE) 7.6053*** 7.8633***
(0.3767) (0.3555)
Governmental performance 0.3449* -0.6002 0.4554*** -0.8379**
(0.1916) (0.4357) (0.1767) (0.3385)
Occupational status (RC: Employed)
Retired -0.6317** 0.1190 -0.4330* -0.1653
(0.2785) (0.6497) (0.2642) (0.4146)
Unemployed -0.4165* -0.1685 -0.2091 -0.5200
(0.2181) (0.5958) (0.2094) (0.4446)
Student -0.1457 -0.1534 -0.0192 -0.1299
(0.2036) (0.4347) (0.1910) (0.3227)
Housewife 0.1947 0.0624 0.2705 0.0419
(0.1861) (0.3988) (0.1723) (0.2933)
Family income (RC: Up to 10,000 ptas.)
10,001-20,000 ptas. -2.0060*** 0.4197 -2.0467*** 2.4033*
(0.6665) (1.1060) (0.6582) (1.3071)
20,001-30,000 ptas. -2.6796*** 0.2084 -2.6369*** 2.0545*
(0.6058) (0.8293) (0.6166) (1.2229)
30,001-50,000 ptas. -2.4396*** 0.3904 -2.4682*** 2.0702*
0.5854) (0.8014) (0.5995) (1.2157)
50,001-70,000 ptas. -2.8763*** 0.6406 -2.9024*** 2.4477**
(0.5870) (0.7883) (0.6006) (1.2146)
70,001 –100,000 ptas. -3.6230*** 0.7152 -3.6181*** 2.7004**
(0.5961) (0.8149) (0.6069) (1.2316)
100,000 – 250,000 ptas. -3.8277*** 0.9340 -3.9168*** 2.9970**
(0.6031) (0.8365) (0.6147) (1.2447)
28
More than 250,000 ptas. -3.3927*** 2.2570 -3.7737*** 4.3648***
(0.7122) (1.4292) (0.7477) (1.3633)
Education (RC: Primary school)
Secondary education -0.4601*** 0.8502*** -0.4559*** 0.8872***
(0.1470) (0.3163) (0.1374) (0.2302)
University (undergraduates) -0.6415*** 1.1362** -0.6625*** 1.1378***
(0.2244) (0.4853) (0.2095) (0.3705)
University (postgraduates) -0.3322 0.3252 -0.4450** 0.5585
(0.2387) (0.5355) (0.2215) (0.4090)
Gender (RC: female) 0.0795 0.6989** 0.1039 0.5882**
(0.1526) (0.3356) (0.1416) (0.2521)
Age (RC: 18-36 years)
37-41 -0.2229 0.0928 -0.3437* 0.1364
(0.1976) (0.4583) (0.1921) (0.3312)
42-51 -0.2235 0.2666 -0.3031** 0.5562**
(0.1663) (0.3629) (0.1579) (0.2518)
52-66 -0.3584** 0.4917 -0.4778*** 0.7056**
(0.1764) (0.3735) (0.1671) (0.2828)
More than 67 years -0.4438 0.0049 -0.6671** 0.5952
(0.2884) (0.7184) (0.2766) (0.4465)
Community size 0.0651*** -0.0811 0.0710*** -0.0088
(0.0251) (0.0595) (0.0237) (0.0450)
Intercept 0.3058 -14.7033*** -1.9394*** -10.7755***
(0.6822) (1.2128) (0.6589) (1.3454)
Number of cases 2,851 2,106 2,900 2,146
Prob. 0.00 0.00 0.00 0.00
Pseudo R2 0.45 0.79 0.40 0.65
Log likelihood -1089.99 -248.64 -1197.82 -423.50
Correctly classified (in %) 83.76 95.44 82.00 92.64
a Robust standard errors in parenthesis. Levels of statistical significance are ***at 1%; ** at 5%; * at 10%.
Source: DATA survey, 1982
29
Table 5: Logistic regressions to explain party vote in Portugal 1983 a
Model A: With Ideology Model B: Without Ideology
Variables Vote PS Vote PSD Vote PS Vote PSD
Religious attendance (RC: Two times per month)
Never 0.7926* 0.8688 1.023*** -0.0410
(0.4630) (0.6305) (0.328) (0.4907)
Sometimes 0.2549 0.1358 0.1625 0.0662
(0.4326) (0.6045) (0.3399) (0.4535)
Every Sunday -0.6219 1.0475* -1.3372 0.8584*
(0.4015) (0.5963) (0.3314) (0.4623)
More than once per week -0.7054 2.0330*** -0.7858 2.0067***
(0.6129) (0.8029) (0.5415) (0.7493)
Ideology -6.8926*** 8.2448***
(1.2045) (1.3714)
Rui Machete’s (PSD) thermometer 2.8327*** 3.2250***
(0.7605) (0.6144)
Mario Soare’s (PS) thermometer 5.0089*** 4.3767***
(0.6733) (0.522)
Occupational status (RC: Self-employed)
Employed 1.1357*** -0.8528 0.9917*** -1.1578***
(0.4403) (0.5392) (0.383) (0.452)
Unemployed 1.7595** -1.6615 1.2369* -1.4069
(0.7334) (1.0333) (0.661) (0.902)
Retired 1.4146** -0.4128 1.2826*** -0.6077
(0.622) (0.6686) (0.498) (0.593)
Student 1.2130 -0.1199 0.5214 0.5840
(0.9469) (1.1392) (0.754) (0.822)
Housewife 1.4039*** -1.1065 1.0547** -1.0943**
(0.5360) (0.6772) (0.423) (0.518)
Family income (RC: Less than 20.000 escudos)
20,000-29,999 escudos 0.5004 -0.2657 0.5216 -0.1611
(0.5202) (0.5998) (0.400) (0.551)
30,000-39,999 e. -0.4342 0.7069 -0.1907 0.6179
(0.557) (0.5787) (0.384) (0.525)
40,000-49,999 e. -0.5789 0.1026 -0.6259 0.5314
(0.599) (0.5925) (0.445) (0.546)
50,000-59,999 e. -0.6653 0.8017 -0.5075 0.9522*
(0.605) (0.6307) (0.469) (0.561)
60,000-69,999 e. -1.3695** 0.4875 -1.3229** 0.7224
(0.714) (0.8792) (0.543) (0.645)
79,999 e. -1.1893 1.5110 -0.8898 1.1437
(0.747) (0.9882) (0.619) (0.903)
Education (RC: Primary school)
Professional training -0.8572 n.d. -0.7644 n.d.
(0.729) (0.759)
Secondary school -0.8181** -0.3706 -0.6735* -0.3748
(0.375) (0.4920) (0.353) (0.392)
University -2.4130*** 0.7217 -1.7621*** 0.4968
30
(0.596) (0.7175) (0.509) (0.588)
Gender (RC: Female) -0.1001 0.1135 -0.2826 0.4228
(0.364) (0.4715) (0.288) (0.386)
Age (RC: 18-36 years)
31-45 -0.4252 0.1165 -0.4552 -0.0827
(0.419) (0.4889) (0.351) (0.384)
46-65 -0.2608 -0.2542 -0.2304 -0.5178
(0.412) (0.5684) (0.359) (0.439)
More than 65 years -0.9528 0.6825 -1.1027** 0.1885
(0.599) (0.7507) (0.497) (0.656)
Community size -0.0650 0.3648** -0.1084 0.2641**
(0.140) (0.1624) (0.1078) (0.1367)
Intercept 2.2957** -7.7827*** -1.3607** -2.7400***
(0.9370) (1.3115) (0.6924) (0.8232)
Number of cases 457 335 531 358
Prob. 0.00 0.00 0.00 0.00
Pseudo R2 0.41 0.39 0.25 0.19
Log likelihood -175.27 -127.80 -256.24 -179.06
Correctly classified (in %) 83.81 85.07 74,95 76,54
a Robust standard errors are in parenthesis. Levels of statistical significance are ***at 1%; ** at 5%; * at 10%.
Sources: The Four Nation Study, 1985.
31
Table 6: Logistic regressions for explaining party vote in Spain, 2004 a
Model A: With ideology Model B: Without ideology
Vote PP Vote PSOE Vote PP Vote PSOE
Religious attendance (RC: Never)
With less frequency 1.1691* -0.0984 1.2953*** -0.1985
(0.7170) (0.2982) (0.4768) (0.2452)
Just in special religious celebrations 1.9472*** 0.1117 2.0659*** -0.0585
(0.6814) (0.2698) (0.4814) (0.2367)
At least once per month 0.7711 -0.3307 1.9053*** -0.6233**
(0.8375) (0.3720) (0.5452) (0.3224)
Once per week 1.6476** -0.9617*** 1.8492*** -1.1301***
(0.7552) (0.3594) (0.5487) (0.3000)
More than once per week 2.2026** -0.5868 2.4462*** -1.1184**
(0.9777) (0.5753) (0.7875) (0.5253)
Attitudes towards abortion -1.7930** 0.1543 -2.1464*** 0.4013
(0.9139) (0.3855) (0.6002) (0.3331)
Ideology 12.8934 -4.3564***
(2.0479) (0.6955)
José L. Rdz. Zapatero’s (PSOE) thermometer 6.9745*** 7.5044***
(0.6516) (0.5887)
Mariano Rajoy’s (PP) thermometer 8.8716*** 9.9116***
(1.0053) (0.9748)
Economic performance -2.8272*** 0.4380*** -2.6900*** 0.5887***
(0.4303) (0.1448) (0.3901) (0.1191)
Occupational status (RC: Self-employed)
Retired 0.6483 -0.1032 0.6856 -0.1054
(0.7553) (0.3773) (0.5244) (0.3387)
Unemployed -1.2206 0.0804 -0.9665 0.2332
(0.8696) (0.4947) (0.7398) (0.4189)
Student -2.9846 0.6074 -0.9254 0.6627
(1.6989)* (0.5899) (1.0589) (0.5016)
Housewife 0.5350 0.3313 0.3649 0.4692*
(0.5537) (0.3144) (0.4570) (0.2787)
Family income (RC: Less than 600 Euros)
600-900 Euros -0.4498 -0.1551 -0.2604 -0.3174
(0.7989) (0.3624) (0.5290) (0.3158)
900-1,500 Euros -0.6816 -0.3120 -0.1846 -0.2884
(0.8004) (0.3620) (0.5162) (0.3134)
1,500-2,100 Euros -0.0714 -0.5041 -0.1010 -0.3388
(0.8807) (0.3908) (0.5976) (0.3414)
2,100-2,700 Euros 1.2241 -1.9925** 1.3502* -0.8323**
(0.9911) (0.4524) (0.7269) (0.3945)
More than 2,700 Euros -0.7680 -0.2982 0.7350 -0.3145
(1.1248) (0.8060) (0.9442) (0.6346)
Education (RC: Primary school)
Secondary school and professional training -0.2117 -0.0151 -0.1220 -0.1525
32
(0.4925 (0.2540) (0.4186) (0.2275
University (3 years) -0.3868 -0.1237 -0.1557 -0.1485
(1.0254) (0.5471) (1.2335) (0.5475)
University (4 or more years) 0.6030 -0.9026*** 0.3046 -0.8055***
(0.6043) (0.2855) (0.5434) (0.2455)
Gender f 0.8779** -0.1996 0.7038* -0.1198
(0.4365) (0.2329) (0.3787) (0.1971)
Age (RC: Female)
37-41 1.8287*** 0.0957 1.1358** 0.0958
(0.6386) (0.3276) (0.4871) (0.2904)
42-51 0.2739 -0.2980 0.2735 -0.2183
(0.6214) (0.2752) (0.4888 (0.2401)
52-66 0.3560 0.0423 -0.0093 -0.1156
(0.6651) (0.3062) (0.4438) (0.2771)
67-81 -1.0854 0.0831 -1.0107* -0.0050
(0.8974) (0.4307) (0.6263) (0.3790
More than 81 years 1.7257** -0.7297 1.5276* -1.3087*
(0.7883) (0.7962) (0.8845) (0.7304)
Community size 0.0254 -0.0276 0.1008 -0.0461
(0.1382) (0.0627) (0.1036) (0.0546)
Intercept -6.9775*** -2.3014*** -1.2933 -5.0911***
(1.6626) (0.8707) (1.1844) (0.6396)
Number of cases 859 971 955 1,100
Prob. 0.00 0.00 0.00 0.00
Pseudo R2 0.79 0.43 0.68 0.36
Log likelihood -113.36 -376.23 -176.80 -476.07
Correctly classified (in %) 95.23 82.70 91.83 79.73
a Robust standard errors in parenthesis. Levels of statistical significance are ***at 1%; ** at 5%; * at 10%.
Source: Demoscopia survey, 2003.
33
Table 7: Logistic regression explaining party vote in Portugal, 2002 a
Model A: With ideology Model B: Without ideology
Variables Vote PSD Vote PS Vote PSD Vote PS
Religious attendance (RC: Never)
One per year -0.7458 0.1630 -0.5598 0.1367
(0.7180) (0.8366) (0.7054) (0.6865)
Two to eleven times per year -0.3518 -0.2132 -0.4377 0.1022
(0.6413) (0.7077) (0.6902) (0.5826)
Once per month 0.6947 -1.0312 0.7725 -0.8973
(0.7604) (0.8163) (0.7289) (0.6887)
Two or more times per month -0.4189 -0.2760 -0.2706 0.0978
(0.7179) (0.7826) (0.7361) (0.6639)
Once per week or more -0.3829 -0.2616 -0.0660 -0.2792
(0.7301) (0.7902) (0.7291) (0.6379)
Attitudes towards abortion -0.4507 -0.0644 -0.3581 -0.0667
(0.5825) (0.5290) (0.4881) (0.4862)
Ideology 5.1656*** -5.953***
(1.2768) (1.0689)
José M. Durâo Barroso’s (PSD) thermometer 0.6163*** 0.7056***
(0.0807) (0.0786)
Ferro Rodrigues’ (PS) thermometer 0.5415*** 0.5886***
(0.0892) (0.0757)
Government performance 1.1239*** -0.9527*** 1.2159*** -1.0869***
(0.2838) (0.3651) (0.2758) (0.3190)
Occupational status (RC: Self-employed)
Unemployed 0.7538 -0.1549 0.7797 -0.2499
(0.9063) (0.9893) (0.7015) (0.7802)
Student -0.2183 -0.0900 -0.2006 0.0421
(0.9194) (0.7682) (0.9727) (0.9190)
Retired -0.3144 0.1539 -0.2188 0.1191
(0.7702) (0.5811) (0.7303) (0.4892)
Housewife -0.6598 0.0408 -0.2071 -0.0805
(0.5859) (0.5067) (0.5677) (0.4687)
Family income (RC: Less than 300 Euros)
301-750 Euros 0.5663 -0.1230 0.0569 0.0803
(0.7798) (0.6885) (0.7300) (0.4677)
751-1,500 Euros 0.7612 -0.5524 0.3295 -0.4074
(0.8066) (0.6772) (0.7510) (0.5087)
1,501-2,500 Euros 0.7800 0.0267 0.5652 0.1035
(0.8008) (0.8336) (0.7550) (0.6625)
More than 2,500 Euros 0.8497 -0.7379 0.8799 -0.6076
(1.1569) (1.0737) (0.9999) (0.7693)
Education (RC: Primary school)
Basic level completed -0.3184 0.3080 0.1352 0.1319
(0.5370) (0.5053) (0.4714) (0.4435)
Secondary school 0.8220 -0.9618** 1.0740** -0.9087*
(0.5720) (0.5266) (0.4749) (0.4766)
University -0.0186 -0.3750 0.2431 -0.4224
34
(0.6352) (0.6218) (0.5758) (0.5858)
Gender (RC: Female) -0.3119 -0.0546 -0.3205 -0.0236
(0.3947) (0.3756) (0.3484) (0.3310)
Age (RC: 18-36 years)
31-45 1.0712** -0.6596 0.8233** -0.5928
(0.5375) (0.4936) (0.4285) (0.4862)
46-65 1.6338*** -1.0990* 1.5415*** -0.9726*
(0.6178) (0.5456) (0.5609) (0.5346)
More than 65 years 1.1983 -1.3495* 0.8763 -1.2381*
(0.9594) (0.8114) (0.8319) (0.6633)
Community size -0.0796 0.0599 -0.0700 0.0882
(0.0875) (0.0781) (0.0809) (0.0675)
Intercept 10.1787*** 4.0077** -8.6096*** 0.3803
(1.8382) (1.7189) (1.4500) (1.2999)
Number of cases 377 370 395 386
Prob. 0.00 0.00 0.00 0.00
Pseudo R2 0.56 0.49 0.49 0.38
Log likelihood -115.22 -128.93 -139.53 -165.01
Correctly classified (in %) 87.80 86.22 86.08 82.64
a Robust standard errors in parenthesis. Levels of statistical significance are ***at 1%; ** at 5%; * at 10%.
Source: Instituto de Ciências Sociais (ICS) survey, 2002.
35
Figure 5. Religiosity and Party Support in Spain, 1983-2004
10%
13% 14%
46%
30%
36%
90%
62%
65%
56%
18%
34%
40%
53%
25%
46% 47%
53%
5%
29%
33%
37%
72%
54%
45%
29%
24%
17%
5%
8%
2%
7%
34%
14% 12%
33%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1983 1990 1999 2004 1983 1990 1999 2004 1983 1990 1999 2004
AP/PP PSOE PCE/IU
Non-Believers
Nominal Believers
Nuclear Believers
Figure 1, Religiosity and Party Support (Spain)
Sources: Four Nations Survey, 1985, WVS 1990 and 1999, and Demoscopia survey, 2003.
36
Figure 5. Religiosity and Party Support in Portugal, 1983-2002
26%
18%
11%
52%
43%
38%
22%
9% 10%
17%
15%
21%
3%
13%
26%
51% 51%
53%
30%
50%
46%
54%
24%
33%
37%
50%
26%
28%
35%
34%
48%
32%
37%
17%
7%
17%
54%
58%
53%
33%
59%
51%
61%
53%
36%
20%
10%
27%
0%
10%
20%
30%
40%
50%
60%
70%
1983 1990 1999 2002 1983 1990 1999 2002 1983 1990 1999 2002 1983 1990 1999 2002
PS PCP PSD CDS
Non-Believers
Nominal Believers
Nuclear Believers
Figure 2, Religiosity and Party Support, Portugal
Sources: Four Nations Survey 1985, WVS 1990 and 1999, and ICS survey, 2002.
37
Table 8: Regression linear model to explain ideology in Spain, 2002*
Ideology
Coefficients Standard Errors
Mother´s ideology
Religious attendance (RC: Non-believer)
Almost never
Sometimes a year
Sometimes a month
Almost every Sunday
More than once a week
Occupation (RC: Worker)
Pensionist
Unemployed
Student
Housewife
Family income
Education
Age
Gender
Constant
5.2343***
0.6662***
1.0612***
1.0433***
1.3436***
1.9716***
0.3913***
-0.1527
-0.0615
0.1974*
0.0178
-0.0907***
-0.0132***
0.0910
1.9638***
(0.2215)
(0.1030)
(0.1174)
(0.1404)
(0.1296)
(0.2629)
(0.1360)
(0.1262)
(0.1162)
(0.1060)
(0.0251)
(0.0249)
(0.0031)
(0.0740)
(0.2137)
Number of cases
F
Prob>F
R2
1,848
92.76
0.00
0.46
* Levels of significance are ***at 1%, ** at 5%, * at 10%
Source: Centro de Investigaciones Sociológicas (CIS) # 2,384, 2002.
38
Table 9. Spain, 1982 and 2004: Religiosity, party choice, and ideology
Ideology b
Religiosity a Totals
Left
Center-left
Center 1
Center 2
Center-right
Right
Vote 1982 2004 1982 2004 1982 2004 1982 2004 1982 2004 1982 2004 1982 2004
AP-PP 28 40 -- 1 1 2 6 18 14 18 64 44 14 16
UCD-CDS c 10
-- 3 44 29 23 2
PSOE 29 33 6 7 47 52 35 31 7 6 4 3 -- 1
Nuclear Catholic
PCE-IU 1 2 70 28 20 71 10 -- -- -- -- -- -- --
AP-PP 16 21 -- -- 2 1 13 24 16 27 59 40 10 8
UCD-CDS 6-- 11 40 38 11 --
PSOE 56 47 9 14 64 58 21 22 3 3 2 2 -- --
Nominal Catholic
PCE-IU 1 4 50 34 37 54 12 10 -- -- -- -- -- 2
AP-PP 5 8 1 -- 5 7 7 25 26 21 54 27 7 20
UCD-CDS 2-- 28 53 17 -- 3
PSOE 53 47 15 22 68 57 13 16 2 4 1 -- -- 1
Non-believer
PCE-IU 6 9 67 39 29 52 4 5 -- 2 -- -- -- 2
a This is a three-way table. Religiosity has been grouped into three basic categories: Nuclear Catholics, Nominal Catholics and Non-believers. For Spain 1982, Nuclear Catholics are those who consider themselves either as “very
good Catholics” or as “Practising Catholics”; Nominal Catholics are those who feel themselves to be “not very practising Catholics”. Lastly, Non-believers are those who are “not practising”, “indifferent” or “atheists”. For Spain
2004, the Nuclear Catholics category is formed by those attending to church either “every day”, “more than once a week” or “once a week”; Nominal Catholics and those who attend to church “once a month”, “just in special religious
celebrations” or “less frequency”; lastly, Non-believers are those who “never” attend to church.
b Left represents those individuals located at values 1 and 2 in the self-placement scale; Center-left corresponds with values 3 and 4; Center-1 corresponds with value 5; Center-2 with value 6; Center-right with values 7 and 8, and
Right with points 9 and 10.
c Unión de Centro Democrático (UCD) and Centro Democrático y Social (CDS) only for 1982.
Sources: DATA survey, 1982, and Demoscopia survey, 2003..
39
Table 10. Portugal, 1983 and 2002: Religiosity, party choice, and ideology
Ideology b
Religiosity a Totals Extreme left Left Center-left Center Center-right Right Extreme
right
Vote 1983
2002 1983 2002 1983 2002 1983 2002 2002 1983 2002 1983 2002 1982 2002
APU/BE c 4 346 17 31 -- 15 33 50 8 -- -- -- -- --
PS 48
37 3 5 12 19 47 13 42 19 5 9 9 9 5
PSD 31 46 -- -- 3 3 16 -- 15 25 8 40 41 17 31
Nuclear
Catholic
CDS d 16 13 2 -- -- 4 10 -- 8 14 11 42 35 32 42
APU/BE c 19 9 33 28 51 30 10 19 16 4 3 1 3 -- --
PS 51 44 4 5 18 15 43 32 30 20 8 11 5 4 4
PSD 21 42 -- 1 2 1 20 1 22 34 16 32 45 12 13
Nominal
Catholic
CDS d 9 5 -- -- 6 6 11 -- 31 8 -- 57 25 17 37
APU/BE c 32 16 42 48 37 30 15 9 9 2 4 2 -- 1 --
PS 44 42 8 5 24 17 50 33 29 13 5 1 9 3 2
PSD 18 37 -- -- 2 2 17 2 16 45 18 28 53 8 10Non-believer
CDS d5 5 -- -- 7 -- -- -- 28 20 -- 47 28 27 43
a This is a three-way table. For Portugal 1983, Nuclear Catholics are those who attend to church “more than once a week” and “every Sunday”; Nuclear Catholics, “two times a month” and “sometimes”; and Non -believers are those
who “never” attend to church”. For Portugal 2002, Nuclear Catholics attend to church “once a week or more”; Nominal Catholics grouped those who attend to church “two or more times a month”, “once a month” and “two to eleven
times a year”; finally the category Non-believers are those who attend to church “once a year” or “never”.
b For Portugal 1983, Ideology is constructed in the same way that for Spain. However, for Portugal 2002 there are slightly differences because the original self-placement variable has 11 values ranging from 0 to 10 and being the 5
value the modal. For that reason, we have introduced a new category called Center for this particular location. Thus, this variable remains as follows: 0-1, Extreme left; 2-3, Left; 4, Center-left; 5, Center; 6, Center-right; 7-8, Right;
9-10, Extreme right.
c For the 1983 elections, Aliança Povo Unido (APU) was an electoral coalition including the Movimento Democrático Português (MDP) and the Partido Comunista Português (PCP). For 2002, the Bloco de Esquerda, a new leftist
party far away from PCP, ran together in coalition with the communists.
d Centro Democrático Socia (CDS), the most conservative party in Portugal, also formed an electoral coalition with other smaller conservative and rightist parties in the 2002’s elections.
Sources: Four Nations Study, 1985, and ICS survey, 2002.
