The context of public attitudes towards the EUFootnote ¹

In the recent past, attitudes towards the EU have become problematic in many member states. Even those countries that traditionally were more optimistic have actually experienced important declines in their popular backing of the European integration process. This is particularly true where the economic crisis has hit the domestic scene more severely (Serricchio et al., Reference Serricchio, Tsakatika and Quaglia2013). Here, the strictness of austerity measures has created wide public discontent and blame towards the national government and the EU. In addition, there has been the growth of the protest vote and escalation of radical parties that have included euroscepticism in their rhetoric. This is the case in Italy, a country where in the past, public attitudes towards the EEC/EU were very optimistic and usually more positive than the European average (Isernia, Reference Isernia2005; Bellucci and Serricchio, Reference Bellucci and Serricchio2012). Here too, citizens have become more sceptical about the integration process (see Quaglia, Reference Quaglia2009; Serricchio, Reference Serricchio2011; Di Mauro, Reference Di Mauro2014) and they now vote for Eurosceptical parties more than ever before. After many years of wide consensus (Conti and Verzichelli, Reference Conti and Verzichelli2012), opposition to Europe has developed into a main dimension of party competition. It is also important to note that in this country there was a certain period of time where opposition occurred in a soft form mainly involving the mainstream parties (Conti and Memoli, Reference Conti and Memoli2014); however, more recently it has developed into a much harder fashion, which include exit proposals from the eurozone and rejection of the main policy pillars of the EU, two plans that are now proudly forwarded by popular radical parties such as the Five Star Movement and the Northern League (Di Virgilio et al., Reference Di Virgilio, Giannetti, Pedrazzani and Pinto2014).

In this article, through the analysis of public attitudes towards the EU in Italy, we explore the main features that currently inform different attitudes of public opinion, including the most Eurosceptical ones. Furthermore, by making reference to a dynamic case, we will test the main theoretical arguments that aim to explain these attitudes. Italy is one of the founding members of the EU and has participated from the very beginning in every phase of the integration process. Under these circumstances, citizens in this country qualify among those with the longest exposure to the supranational integration processes. For a long time, the Italian citizens were also among those most attached to the EU polity, but during the last decade the country has been going through a process of economic hardship unknown since its membership in the EEC/EU, a phenomenon that may have changed public awareness of the costs and benefits stemming from the integration process. At the end of 2013 (the year the data analysed for this article was collected), the World Economic Forum described Italy as a country with over 130% of public debt compared with its GDP, de-industrialization during the past 10 years was at a record high of 17.8%, and the unemployment rate was at 12%. Within this context, several competitive pressures within the EU have challenged the country. For example, after the recent waves of enlargement to new member states, Italy’s status changed from one of a net recipient to one of a net contributor to the EU budget, while at the same time the constraints of the eurozone imposed a tight discipline of its state finances. It is evident that public perceptions of the utilitarian gains of EU membership may have changed consequentially, particularly in the context of the economic crisis that only adds to the lasting economic decline of the country (Di Mauro, Reference Di Mauro2014: 157). Hence, Italy is a case where some of the most recognized determinants of public attitudes towards the EU (e.g. cognitive, affective, and utilitarian factors, see Sanders et al., Reference Sanders, Bellucci, Toka and Torcal2012a) take rather extreme forms; therefore, it is an interesting case to test theory.

Through original research techniques, the article assesses the relative influence of several factors on citizens’ attitudes towards the EU, within the dynamic context of a country whose connection with Europe has been more problematic in the recent past. For example, Figure 1 shows that support for membership declined dramatically over the past 20 years such that Italian public opinion has changed from being among the most supportive to levels of support that are below the EU average. This fall became particularly severe starting in the late 1990s, and now, today’s rates of public support for EU membership have halved as compared with the period before the Maastricht Treaty. In the following sections, we move the analysis beyond the generalized support for membership to examine the specific motivations behind citizens’ attitudes towards particular aspects of the EU process. In particular, we explore their attitudes towards the EU representation system, EU policy, and European identity. These are three fundamental dimensions that have been found to inspire the attitudes of mass publics in other countries (Sanders et al., 2012Reference Sanders, Bellucci, Toka and Torcala, Reference Sanders, Bellucci, Toka and Torcalb) and whose role we test in the Italian case. In doing so, we apply a methodological innovation in this research field based on rival models, a strategy that allows careful hypothesis testing together with accurate assessment of the impact of different factors on citizens’ attitudes.

Figure 1

Italy’s membership in the EU evaluated. Question: Generally speaking, do you think that (your country’s) membership of the European Community (Common Market, EU) is …? Answers of those who consider membership a good thing (Italy and EEC/EU average). Source: Eurobarometer (different years).

The multifarious nature of citizens’ attitudes towards the EU

The first pioneering studies on public opinion and the EU argued that citizens were broadly favourable about the Common Market and that they gave their permissive consensus to the elites to pursue the final goal of Europe’s integration (Lindberg and Scheingold, Reference Lindberg and Scheingold1970). At the time, citizens were described as not very informed and not very interested in this process, presumably because it did not impact their daily lives in tangible ways. Their consensus was broad and so attitudes were not contingent on specific aspects of integration. When the public is not very (or at all) informed about a given issue, they can be more easily influenced by their political representatives and prove ready to align with the political elites (Lupia and McCubbins, Reference Lupia and McCubbins1998). For a long time, this appeared to be the case with the issue of Europe’s integration. However, after the Maastricht Treaty in 1992, popular views changed substantially (Eichenberg and Dalton, Reference Eichenberg and Dalton2007). In the last 20 years, the EU has produced an impact on the member states that has certainly become more evident to citizens. Since then, it has also become more relevant to assess public attitudes towards the EU empirically, especially in those countries where clear signs of public opposition (that were more clearly exemplified at the time of ratification of EU treaties and during the European elections) made the permissive consensus no longer obvious.

However, in the crucial years immediately after the Maastricht Treaty, the analyses of attitudes towards the EU still tended to be broad and confined to the concept of underlying support (Gabel, Reference Gabel1998a). Where opposition was more visible, the scholarly tendency was to consider negative attitudes as country specific and dependent on the circumstances at work in the domestic arena (Benoit, Reference Benoit1997), including popularity of the incumbent and specific national identity patterns. However, the scale of citizens’ euroscepticism has grown so much over time that the scholarly interest for these attitudes and for their comparisons across countries has increased as well as become more issue specific. The most recent studies show that even though a sense of identification with Europe has gradually developed among citizens (Duchesne and Frogniere, Reference Duchesne and Frognier1995; Carey, Reference Carey2002; Herrmann et al., Reference Herrmann, Risse and Brewer2004; Brueter, Reference Bruter2005), identification with their own nation has certainly not vanished (Herb and Kaplan, Reference Herb and Kaplan1999; Citrin and Sides, Reference Citrin and Sides2004; Hooghe and Marks, Reference Hooghe and Marks2009). This interplay could actually represent an obstacle for the development of a post-national layer of identity, especially in the context of the economic crisis when, on account of increased European policy constraints upon national governments, popular dissatisfaction more often takes the colour of a nationalistic/identity defence (Hooghe and Marks, Reference Hooghe and Marks2009). Italy is indeed one of those countries more severely hit by the crisis and where the sense of popular attachment to Europe has also fallen more dramatically (Serricchio, Reference Serricchio2011; Bellucci and Serricchio, Reference Bellucci and Serricchio2012; Di Mauro, Reference Di Mauro2014).

Recent research shows that, in general, attitudes towards the EU are multi-dimensional as they pertain to different aspects of the EU process (Beaudonnet and Di Mauro, Reference Beaudonnet and Di Mauro2012; Sanders et al., Reference Sanders, Bellucci, Toka and Torcal2012a). Some authors took a clear Eastonian perspective (Easton, Reference Easton1965, Reference Easton1975) and incorporated several dimensions in their definition of support for the EU. For example, some defined support as broad and specific: both are relevant for public attitudes, but the former has to do with attachment to Europe, whereas the latter pertains to the evaluation of the outputs of the EU system based on institutional performance and actual policies (see Kopecky and Mudde, Reference Kopecky and Mudde2002; Memoli and Vassallo, Reference Memoli and Vassallo2009; Boomgaarden et al., Reference Boomgaarden, Andreas, Schuck, Elenbaas and de Vreese2011). The same applies to the study of (negative) attitudes. For example, Taggart and Szczerbiak (Reference Taggart and Szczerbiak2004) distinguished between hard and soft euroscepticism. According to the two authors, the former category implies outright rejection of the entire project of political and economic integration, as well as opposition to the own country joining or remaining a member of the EU, whereas the latter involves contingent or qualified opposition to specific aspects of European integration usually related to policy. Lubbers and Scheepers (Reference Lubbers and Scheepers2010) separated political from instrumental euroscepticism: the former type concerns opposition to delegation of policy competence to the EU, whereas the latter refers to negative assessment of the benefits stemming from (own country) membership in the EU. De Vries and Edwards (Reference De Vries and Edwards2009) conceptualized euroscepticism as a continuum of stances ranging from outright rejection of their own country membership in the EU to contingent opposition against some of its policies.

Moving from the general acknowledgement of the multifarious nature of the phenomenon, some authors conceptualized attitudes towards the EU as based on the three dimensions of identity, representation, and policy scope. These dimensions nurture the quality and legitimacy of every democratic system (Benhabib, Reference Benhabib2002) as well as that of the EU, a supranational organization that challenges feelings of attachment and identification with a polity, the mechanisms of political representation and democratic control over political authorities, as well as the scope of many policies (Bartolini, Reference Bartolini2005: 211). For these reasons, in the past, these three dimensions have been proposed for the empirical assessment of citizens’ attitudes towards the EU (Sanders et al., 2012Reference Sanders, Bellucci, Toka and Torcala, Reference Sanders, Bellucci, Toka and Torcalb), as well as of the elites (Best et al., Reference Best, Lengyel and Verzichelli2012), the parties (Conti, Reference Conti2014), and the media (Bayley and Williams, Reference Bayley and Williams2012). From these theoretical premises, we start our analysis of the Italian case. Do citizens conceive the multi-dimensional nature of the EU process as it was conceptualized by scholars? The tripartition based on identity, representation, and policy scope certainly offers a valid starting point that allows to encompass many aspects of the EU process, but do Italian citizens think of the EU in the same way? We address this problem before making use of the above three dimensions as indicators of public attitudes towards the EU. We consider it important to first establish whether these three dimensions really are distinct for Italian citizens, thus if they require specific examination as it was theorized in the comparative literature.

In order to answer the above questions, we analysed recent data on Italian public opinion (Eurobarometer 80.1, November 2013), not only to depict a situation that is as close as possible to reality, but also because they refer to a period of intense politicization of the issue of European integration. On the one hand, the general elections were quite recent at the time (they occurred in February of the same year) and the memory of the politicization of Europe during the electoral campaign (Di Virgilio et al., Reference Di Virgilio, Giannetti, Pedrazzani and Pinto2014) was still alive. On the other, the European elections of May 2014 were approaching and salience of EU-related issues was increasing in view of this event. The imminent Italian Presidency of the EU (starting in July 2014) contributed as well by making Europe more central in the domestic discourse. For these reasons, we maintain that this was a period where citizens were extraordinarily exposed to debates on the EU and it is therefore ideal for the analysis of their attitudes.

For our analysis, we first selected from the data set a group of variables pertaining to the above three dimensions. Then, we analysed whether the answers of the respondents co-vary in a way that allowed grouping these variables consistently with the above three-dimensional model. From the operational point of view, in order to assess the dimensionality of citizens’ conceptions of the EU, we applied a factor analysis to the data (results are reported in Table 1).Footnote ² The evidence showed that Italian citizens were able to discern different aspects of the integration process and that the way they spontaneously grouped them fits the three-dimensional model. The clustering of items under the three dimensions of identity, representation, and policy scope worked well even when citizens were asked to express their views on policies that currently have different levels of integration (i.e. the single currency, as opposed to the foreign and security policies that are instead much less integrated), or institutions that have different functions and represent different stances (e.g. the Parliament as opposed to the Commission and the European Central Bank). Thus, beyond the theoretical significance of the above three dimensions, we could also confirm their empirical validity for the analysis of public opinion and the EU. From these premises, we then analysed citizens’ attitudes by referencing precisely to these three dimensions (i.e. the dependent variables of our analysis). The results showed that citizens’ attitudes were different across these dimensions and that they varied in the distinct ways of European representation, identity, and policy scope.

Table 1

Factor analysis of citizens’ responses

Source: Eurobarometer 80.1 (2013).

The theoretical foundations of public attitudes towards the EU

In the previous section, we defined the dependent variables of our study (i.e. public attitudes towards identity, representation, and policy scope in the EU). In this section, we draw from the literature and define the independent variables of our investigation. Research shows that the most relevant determinants of citizens’ attitudes towards the EU pertain to a mix of socio-demographics, partisanship, symbolic motivations, and utilitarian cost-benefit analyses. Along these lines, some authors (Bellucci et al., Reference Bellucci, Sanders and Serricchio2012) grouped all these factors under the following four dimensions: cognitive mobilization, utilitarian calculations, political heuristics, and polity identification. The multiple motivations informing these attitudes can be considered evidence of progress in the integration process, from functional mainly addressing economic cooperation, to multi-faceted encompassing political and symbolic aspects as well. As to cognitive mobilization, in the 1970s, Inglehart (Reference Inglehart1970, Reference Inglehart1977), Inglehart et al. (Reference Inglehart, Rabier and Reif1991) argued that information and knowledge of the EEC/EU, as well as a higher level of education, positively influences citizens’ attitudes towards the integration process. Following a utilitarian approach, other authors argued that citizens make their own calculations about the costs and benefits stemming from EU membership based upon their personal interests (egotropic utilitarianism) and those of their community (typically the national polity, sociotropic utilitarianism) (Eichenberg and Dalton, Reference Eichenberg and Dalton1993; Gabel and Palmer, Reference Gabel and Palmer1995; Gabel and Whitten, Reference Gabel and Whitten1997; Marks and Steenbergen, Reference Marks and Steenbergen2004; Loveless and Rohrschneider, Reference Loveless and Rohrschneider2011). Some other scholars maintained that the political orientations of citizens towards national politics filter their attitudes towards the EU. In this perspective, identification with a party and its stance, as well as information on domestic politics (Gabel, Reference Gabel1998b) and attachment/trust in the national institutions (Anderson, Reference Anderson1998) are considered influential factors. However, no consensus has been reached in the literature between those who support the argument of a positive influence and double allegiance between national and European levels of politics (van Kersbergen, Reference Van Kersbergen2000) and the advocates of a negative influence (Sánchez-Cuenca, Reference Sánchez-Cuenca2000; Di Mauro, Reference Di Mauro2014). As such, the direction of the relationship remains a matter of debate. In more recent times, identity was proposed as another powerful source of support for and opposition to the EU, but whereas some authors argued that a strong national identity is an obstacle to the development of a truly European identity (Carey, Reference Carey2002; McLaren, Reference McLaren2002), others maintained that there is a positive correlation between these two layers of identification (Duchesne and Frogniere, Reference Duchesne and Frognier1995; Citrin and Sides, Reference Citrin and Sides2004; Bruter Reference Bruter2005). According to Hooghe and Marks (Reference Hooghe and Marks2005), it is the political context, particularly how divided national elites are on EU issues, that determines the nature and direction of the interplay between the European and other layers of identity. However, according to Bellucci et al. (Reference Bellucci, Sanders and Serricchio2012), it is the character of identity as mainly civic or ethnic – nations emphasize either one or the other aspect – that determines the direction of the relationship between national and European identity.

Starting from these theoretical premises, we inserted our analysis of the Italian case in a multi-dimensional explanatory framework that allowed us to test the validity of some of the most recognized theoretical arguments. For this purpose, we selected from the Eurobarometer data set a group of indicators that allowed us to measure the impact of the above hypothesized determinants on citizens’ attitudes (see Table 2). Some of these indicators consider the current situation of Europe in the context of the crisis.Footnote ³ What are the most influential factors that inform public attitudes towards the EU in Italy? Considering that the literature has documented a negative trend in these attitudes, what are the main causes of this widespread malaise?

Table 2

The operationalization of three explanatory keys of citizens’ attitudes

^a Question: What does the EU mean to you personally? (Multiple answers possible).

Analysis and discussion

In order to estimate the relative impact of several explanatory components on public attitudes towards the EU, we made use of rival models, a strategy that was pioneered in applied econometrics (Granger, Reference Granger1999). Despite its advantages, rival models are still of limited use in political science with only a few exceptions (see Whitely and Seyd, Reference Whitely and Seyd1996; Franzese, Reference Franzese2002; Franchino, Reference Franchino2012). By proposing a method that, in the future, could be also replicated in other national contexts as well as in comparative works, a methodological goal of the article was to enhance the level of accuracy of the results generated in this research field. Other works normally make use of measures such as goodness of fit or R ² to indicate how well data fit a statistical model in the absence of clearly specified alternative models; or they make generic reference to alternative or complementary models (McLaren, Reference McLaren2002, Reference McLaren2004, Reference McLaren2006; Hooghe and Marks, Reference Hooghe and Marks2004; Sanders et al., Reference Sanders, Bellucci, Toka and Torcal2012b; Di Mauro, Reference Di Mauro2014; on the Italian case see Bellucci and Serricchio, Reference Bellucci and Serricchio2012; Serricchio, Reference Serricchio2012) in the absence of a valid test of the best model and mutual relationships among rivals. All of these solutions may affect the accuracy and the overall reliability of findings or they may cause loss of information (Charemza and Deadman, Reference Charemza and Deadman1997). These are problems that the use of rival models can properly address. In particular, their use allows greater accuracy in assessing to what extent different theoretical perspectives co-exist and contribute with different intensity to explain a phenomenon, or if one theory prevails over others.

Any analysis based on rival models starts with the theoretical construct of a data-generating process (DGP) that is the hypothesized multi-dimensional causal mechanism informing the analysis (Granger, Reference Granger1990). As reported by Patty et al. (Reference Pattie, Seyd and Whiteley2004: 172), in reality ‘the DGP is usually very complex and can only be approximated in practice by models which are estimated. In this case, the best methodological strategy involves testing rival models against each other to see which provide the best approximation to the DGP. If one model is better than another it should encompass the alternative, that is it should be able to predict or account for the results of an alternative model, as well as predict phenomena that its rival is unable to predict’. According to Charemza and Deadman (Reference Charemza and Deadman1997: 250), model A encompasses model B if it can explain the variance of model B and also the variance that model B is unable to explain. In order to test the rivals, we first defined the independent variables of our DGP (pertaining to the utilitarian, cognitive, and identification theories; Table 2).Footnote ⁴ Second, we applied a simple OLS multiple regression to test the impact of each theory and to control for the effects of the other theories. Third, because the models in the DGP are non-nested, we adopted a J test (see Davidson and MacKinnon, Reference Davidson and MacKinnon1981) again based on simple OLS multiple regression to derive predicted values from a given equation and then substitute them into a rival equation. In this new equation, if one model was able to predict the results of the alternative model and other aspects that its rival was unable to predict, then it could be considered as encompassing the rival. One model can encompass another, but models can also encompass each other if they are submodels of some more general model. Hence, the accuracy of this kind of analysis consists not only of its capacity to document the relative impact of every causal component with respect to the others, it also reveals the limits of the explanatory capacity of different components when they belong to a more general unidentified model.

As we show in Tables 3–5 (where attitudes towards the EU have been broken down into the three dimensions of representation, identity, and policy scope that have emerged from the analysis in Table 1), none of the rival models encompassed the others. Several components were significant when the rivals were taken into account (see a2, b2, c2 in Tables 3–5). This confirmed that the utilitarian, identity, and cognitive components, all played an impact on attitudes towards the EU and that none encompassed the others. Hence, any reduction of the causal mechanism to a single or two sets of components would result in loss of information as the data showed that reality is, in fact, more complex.

Table 3

Non-nested model: representation

Source: Eurobarometer 80.1 (2013).

The dependent variable is a representation index built with the indicators shown in Table 1.

βs with P<0.05 in bold.

Table 4

Non-nested model: identity

Source: Eurobarometer 80.1 (2013).

The dependent variable is an identity index built with the indicators shown in Table 1.

βs with P<0.05 in bold.

Table 5

Non-nested model: policy scope

Source: Eurobarometer 80.1 (2013).

The dependent variable is a policy index built with the indicators shown in Table 1.

βs with P<0.05 in bold.

More specifically, with respect to the dimension of representation (Table 3), by comparing the base models (a1, b1, c1) with their rivals (a2, b2, c2), we found that the utilitarian model (a2) had a better fit than the other two (identity and cognitive) models. As a matter of fact, the four independent variables pertaining to utilitarianism were all significant even when controlled for the rivals (predict identity components and predict cognitive components), whereas the (identity, cognitive) variables of the other two models (b2 and c2) were not significant when the predicted values of the two rivals were considered. Only one predictor of the identity components was significant (model b2), whereas none of the predictors of the cognitive components contributed to explaining the dependent variable (model c2). After having introduced the control with rivals, we concluded that the attitudes towards representation in the EU, notably towards the EU institutions, were mainly the result of utilitarian calculations about the perceived costs and benefits produced by the European governance on their own personal interests and on those of their own polity. There was a positive relationship between a positive assessment of the current/future European economic situation, optimism with respect to the role played by the EU in the management of the economic crisis and attitudes towards representation in the EU. However, there was a negative relationship between those who held optimistic views about the very meaning of the integration process (i.e. it promotes economic prosperity, social protection, freedom to travel, study, and work anywhere in the EU) and representation: this result showed that in Italy the malaise with the EU institutions and their outcomes was strongly associated with those who held a broad pro-European posture. Those who were more unhappy with the current trajectory of the EU were also those who had greater expectations about the gains stemming from the integration process (in their view, a positive sum game). Under this perspective, euroscepticism in Italy does not appear a principled stance but one that is contingent on a cost/benefit analysis: those who are more Eurosceptical tend to make a negative evaluation of the EU institutions because of their impact on interests, but, at the same time, they hold positive views about the integration process.

With respect to Table 4, where identity was considered as a dependent variable, none of the models fit particularly well. As a matter of fact, after having introduced the rivals, we found that for each cluster of components, only one variable was statistically significant. In sum, our test based on rivals showed that the independent variables selected for the models (that find justification in theory and solid terms of comparison in empirical research) had only a limited explanatory power; differently from utilitarianism in the case of representation, none of the components was more influential than the others. The fabrics of identity in Italy appeared rather uncertain and the determinants of the attitudes towards European identity were ambiguous and to a large extent unidentified (as it is exemplified by the low R ²). The analysis based on rival models allowed us to accurately assess the contribution of each component affecting the attitudes towards each dimension of the EU process. Here, none of the components really emerged as a strong predictor of attitudes towards European identity.

As to the policy scope dimension (Table 5), we found that identity played a more relevant influence than the utilitarian and cognitive components. There was a positive relationship between attachment to one’s own country/feeling Italian and attitudes towards EU policy competence. This finding contradicts the argument that the policy constraints upon national governments, especially in a country so much hit by the crisis and compelled by the austerity measures as Italy, create tensions with the EU that take the form of a nationalistic/identity defence (Hooghe and Marks, Reference Hooghe and Marks2009). We discuss this finding more carefully below.

In sum, the results presented in the Tables 3–5 show that for the three dimensions of the EU process that were analysed in the article (representation, identity, and policy scope), no single explanatory model prevailed on the others (as it is testified by the statistical significance of the rivals predict utilitarian components, predict identity components, and predict cognitive components). However, we also found that the influence of utilitarianism was more apparent in representation; identity was more influential in policy scope; utilitarianism, identity, and cognitive mobilization were all influential for identity (but they have limited explanatory power). In the end, no model entirely encompassed its rivals, whereas their explanatory power varied across the dependent variables.

Moving from the above three models, we produced a parsimonious encompassing version, namely the best model stemming from all those that have been tested before. In order to create this model, we included in a single model (Table 6) only the set of variables that were statistically significant in models a2, b2, and c2 of Tables 3–5. We then applied an OLS multiple regression to the data. This new encompassing model that synthesized cognitive, utilitarian, and identity theory explained more than one-quarter of the variation at the individual level, but only with respect to some aspects of EU integration (representation and policy scope), whereas its explanatory capacity was much lower with respect to the other aspect (identity). Again, utilitarianism emerged as the most influential factor, a result that confirms its relevance found in recent work (Bellucci and Serricchio, Reference Bellucci and Serricchio2012).

Table 6

Cumulative model of representation, identity, and policy scope

Source: Eurobarometer 80.1 (2013).

βs with P<0.05 in bold.

The results obtained suggest three different conclusions. First, utilitarian, (less) cognitive, and identity predictors are useful explanations – they are rival, not alternative or complementary – but they are not sufficient to explain public attitudes towards the EU. This is important because in order to be able to recommend remedies against increasing discontent with the EU and mounting euroscepticism, it would be necessary to have a comprehensive understanding of the determinants of those attitudes. Our study shows that, at least with reference to the Italian case, our understanding of the phenomenon is still limited [in a comparative work Guerra and Serricchio (Reference Guerra and Serricchio2014) came to a similar conclusion] and that theory should be refined to identify factors that explain in more comprehensive ways public attitudes towards the EU in this country. Second, in the models with representation and policy scope considered as dependent variables, the utilitarian and cognitive components were statistically significant. However, a different situation applies when identity was considered as a dependent variable (Table 6) as none of the predictors were statistically significant. This result reinforces our findings on the limited capacity of existing theory to explain attitudes towards EU identity in Italy, a problem that had not emerged in previous works (Bellucci and Serricchio, Reference Bellucci and Serricchio2012). Third, the explanatory power of identity considered as an independent variable was very limited and was not significant in the general model shown in Table 6. This result contradicts Hooghe and Marks’ (Reference Hooghe and Marks2009) argument on the political consequences of identity affecting attitudes towards the EU. All member states do not politicize the EU through an identity drive based on nationalism. This is certainly not the case in Italy where identity (both considered as explanandum and explanans) and the EU are not necessarily opposed concepts.

Final remarks

In the article, we analysed citizens’ attitudes towards the EU in Italy. The analysis was based on recent data and novel research techniques that confirmed some of the results of past research. The findings also correct some interpretations and help specify theory. The Italian context emerged as one of increasing disenchantment, where the number of citizens positively considering membership of their country in the EU declined over time considerably. Within this context, we analysed in greater detail public attitudes towards the EU. We first confirmed that attitudes are multi-dimensional and that the Italian citizens think of the EU in its different aspects pertaining to identity, representation, and policy scope. These dimensions of the EU integration process may interact with each other as was stated by Sanders et al. (Reference Sanders, Bellucci, Toka and Torcal2012b), but they also exist with some degree of distinctiveness in the mind of citizens. Therefore, it seems accurate to analyse popular attitudes taking these three dimensions into careful consideration.

In order to understand public attitudes towards the EU in Italy, we applied a multi-dimensional test synthesising cognitive, utilitarian, and identity theory. We found that the three explanatory models stemming from these theories are rivals and that none encompasses the others. However, the utilitarian model is the one showing the greatest explanatory capacity, whereas the other two models have more limited influence. This result suggests that Italian citizens evaluate the EU primarily in relation with the costs and benefits for the country and upon their personal interests. Although in this country membership in the EEC/EU has been represented for a long time as a meta-level choice (for the Western camp, for the Atlantic Community, for democracy, see Isernia, Reference Isernia2008), nowadays citizens tend to approach it (particularly its institutions) in a pragmatic way that is more focussed on the gains that membership is able to produce, possibly in the same way they would consider other levels of government. This is the sign of progress reached by European integration, with the EU more pragmatically embedded in mechanisms of popular assessment of its institutional performance and policy output. We showed that those who expect more from European integration are also those who are more critical about the EU institutions. Those citizens who are more optimistic about the mission and the broad meaning of European integration are also dissatisfied with its outcomes and institutional working system. In Italy, euroscepticism does not generate from principled opposition to Europe and to delegation of sovereignty to the EU, surprisingly, it is more present among the Europhiles who think that European integration can make reality more prosperous. In light of the Eastonian perspective discussed previously, the growing euroscepticism of the Italians could be defined as specific and mainly interest driven, in presence of a broad backing of EU integration.

However, our study also shows that our understanding of public attitudes towards the EU in this country is still limited and that theory should be refined to propose more comprehensive explanations of the phenomenon. Our multi-dimensional causal mechanism based on comparative theory only explains a fragment of the variation of attitudes in Italy. In particular, the cognitive and identity components play only a very limited influence, which is different from what has been hypothesized and found in other national contexts. These results are also interesting because they contradict a popular interpretation in recent political analyses based on the assumption that identity plays a fundamental role in shaping a constraining dissensus on the EU. In this article, we were able to provide evidence of the counterargument that this is not the case everywhere and specifically not in Italy. Ultimately, thanks to our research strategy based on rival modelling, we were able to assess the overall explanatory power of theory with a great level of accuracy, an example that could serve as a point of reference for future research even of a comparative kind.