- 1 -
Democratic perception
and European attitudes
10 July
2019
THE INFLUENCE OF PERCEIVED VOICE ON PUBLIC OPINION FOR
EUROPEAN INTEGRATION ACROSS COUNTRIES
DAAN MARK VANHOUTEN
- 2 -
Abstract
This thesis studies the effect of perception on national and European policy on public opinion towards European integration. Public opinion plays an important role in contemporary discussions surrounding the European Union. European integration is one of the main topics discussed. Former studies focussed on rational economic behaviour, institutional evaluations, or identity explanations. This thesis focusses on perception as an explanation for the variation of public opinion on European integration across countries, respecting previous research. This thesis finds that perception of voice in European policy is possibly an important predictor of the level of support for European integration. This counts also for perception of voice in national policy in some countries, but not in all. This suggests there is a more complex relationship of perceived voice in national policy, which should be further investigated in future research.
Key words: Perceived voice, democratic representation, European policy, national policy, public opinion, support, Euroscepticism, European integration.
- 3 -
Table of content
Introduction ... - 4 - Theoretical substantiation ... - 9 - Methodology ... - 13 - Results ... - 16 - Discussion ... - 21 - Conclusion ... - 27 - References ... - 28 - Appendices ... - 31 -- 4 -
Introduction
The dawn of the ideal of a unified Europe
A hundred years ago, Europe was a continent at war. Since World War II, a lot has happened. We moved from a continent at war and a divided Europe towards a Europe that cooperates on the world stage. Ever since we stopped disputing each other on the battlefield, Europe has come to realise that a united Europe is in the best interest of all Member States. On the international stage, it was stuck between two great powers, the United States and the Soviet Union, and tensions started rising between both rivals. Both the aftermath of the Second World War and the beginning of the Cold War signalled the beginning of Europe as an influential player internationally (CVCE, July 2016). Much has been said about the unification process of Europe. Conventionally the process was seen as elite driven (Eichenberg & Dalton, 1993). These elites may have disagreed whether Europe should be federalist or intergovernmentalist, but they both wanted a bigger role for the European interests (Baldwin & Wyplosz, 2015). For a period of time, the European integration process stagnated. The Single European Act (SEA) of 1986 brought the goal of European integration back to life. The SEA may have been the initial work of the elites, but from this point onwards, public opinion started to exert incremental influence on the policymakers and the institutions of Europe itself. This becomes most clear from the referenda on the accession of Ireland, Norway, the United Kingdom and Denmark in the period from 1972-1992 (Eichenberg & Dalton, 1993).
Public opinion matters
There has been lots of research in the field of public opinion. Eichenberg and Dalton (1993) tried to reveal what factors influenced public sentiments towards Europe in the period 1973-1988. They did so by tracking changes in support over time and they tried to explain cross-national differences in support. They found that European integration is influenced by a combination of domestic and international factors. Levels of support are linked to success and failure in subsequent international negotiations. Institutional innovations, such as the SEA and the Treaty of Maastricht, mark the increases in support for integration (Eichenberg & Dalton, 1993). Gabel (1998b) says that through its impact on mass behaviour, public opinion shapes and constrains the process of European integration. Public opinion can modify institutional reforms as has happened during the Danish referendum on the Maastricht Treaty and public attitudes can influence politics through lobbying, protests and elections. Hence, public opinion forms the foundation for integration (Gabel, 1998b).
Why do opinions differ?
Several studies have identified systematic differences in the explanation of public opinion towards European integration. Some individual-level differences can be related to partisanship, age, income,
- 5 -
occupation, cognitive skills and political values (Gabel, 1998b). Anderson and Reichert (1996) try to study how differences in public opinion towards European integration can be explained amongst the EU-12. They hypothesise that support is the result of direct and indirect economic benefits, also called the benefit hypothesis (Sánchez-Cuenca, 2000). Individuals living in countries that experience more net benefit, support European integration more. This counts for individuals that directly benefit from integration too (Anderson & Reichert, 1996). Sánchez-Cuenca (2000) proposes a second hypothesis, the institutional hypothesis. The better the opinion of citizens is towards European institutions, the stronger their support for European integration is, and the better their opinion of national institutions is, the less support there will be for more integration (Sánchez-Cuenca, 2000). These two hypotheses are drawn from the instrumental approach in understanding public opinion towards European integration. Instrumental means the consequence of a cost-benefit evaluation. Gabel (1998b) sums up a few popular explanations for the variation in public opinion towards European integration: Cognitive mobilisation (Inglehart, 1970a), political values (Inglehart, 1970b), Utilitarian appraisals of integrative policy (Gabel, 1998a), class partisanship (Inglehart, Rabier & Reif, 1991), and support for government (Franklin, Van der Eijk & Marsh, 1995; Ray, 2003). De Vreese and Boomgaarden (2005) add a few additional explanations: considerations of national identity such as national pride and territorial identity (Carey, 2002), national attachment and exclusive national identity (Marks & Hooghe, 2003), fear of loss of national identity (Christin & Trechsel, 2002), perceived cultural threat (McLaren, 2002), and anti-immigration sentiments (McLaren, 2002).
Explanations of public opinion over the years
The first explanations of variance in public opinion towards European integration over different individuals in Europe have been related to age, income, cognitive skills and political values (Gabel, 1998b). Individual-level support for integration is related to individual demographic characteristics such as age, gender, education, occupation and income. These variables are expected to measure individual benefits, as higher status individuals and males are expected to benefit more from further integration than lower status individuals and females. In addition to individual benefits, indirect economic benefits, such as the percentage of EU trade and the net benefits from the EU, can also explain why integration is a good thing (Anderson & Reichert, 1996).
Next, Inglehart (1970a; Janssen, 1991) describes cognitive mobilisation as an aspect of the process of social mobilization, where countries and individuals transcend their previous networks and expand their horizons. In the heart of the process of social mobilization lays cognitive mobilisation, which is the “increasingly wide distribution of political skills necessary to cope with an extensive political community” (Inglehart, 1970a: 47). Because of the distance between the average citizen and the European institutions and the complexity of the relationship, it is likely that a relative high level of
- 6 -
cognitive mobilisation is required to mobilise public support for European integration. The development of mass support for EU integration is a result of a two-step process. The first step is the occurrence of cognitive mobilisation, proxied by education-levels, and the second step is internalization of the values, where the dominant message regarding remote, political integration is internalised. This also means that negative messages lead to a decrease in EU integrational support from the more educated citizens (Inglehart, 1970a).
Janssen (1991) also describes Inglehart’s Silent Revolution, where a second individual-level explanatory variable is distinguished, political value-orientations. Inglehart (as described in Janssen, 1991) distinguishes two different value-orientations: materialist and postmaterialist. Materialists give priority to economic and physical security, whereas postmaterialists prioritise the fulfilment of individual needs. These two different orientations lead to two hypotheses. First, integration fulfils the intellectual needs of postmaterialists, and materialists have less time for such abstract issues. Postmaterialists are therefore believed to be more pro integration and materialist more against. Secondly, postmaterialists have more need to belong as opposed to materialist and are therefore more likely to attach themselves to the European cause (Janssen, 1991).
Gabel and Palmer (as described by Gabel, 1998a; b) theorised a utilitarian model for explaining public opinion towards EU integration. This theory states that appraisal of costs and benefits from integrative policy differ in different socioeconomic situations. These different experiences lead to different attitudes towards EU integration. This in turn leads to the prediction that citizens with larger human capital, meaning higher levels of education and occupational skills, are more likely to adapt more easily to the occupational competition and are therefore more likely to endorse the call for more European integration than citizens with smaller human capital. Furthermore, they predict that citizens with larger financial capital are provided with more investment opportunities, which leads to more potential benefits, are more likely to gain from more integration. Citizens with lower wages face higher costs due to competition and welfare spending retrenchments. Finally, citizens living closer to a neighbour country face potential benefits from integration (Gabel and Palmer, 1995, as described in Gabel, 1998a; b). However, Anderson (1998) argues that the requirement for this perspective is not always met, because of the lack of specific information.
Several studies investigated the role of partisanship in public opinion (Franklin, et al., 1995; Gabel, 1998b; Inglehart et al., 1991). Previously named individual characteristics, such as age, occupation, income and political values, influence individuals’ opinion towards integration, and besides they influence partisanship. At the same time, parties shape supporters’ attitudes towards integration independent of their personal characteristics. Inglehart and colleagues (1991) discovered that Left
- 7 -
class partisanship leads to less support for European integration than Right class partisanship. This effect varies with different levels of disagreement among parties, party unity, issue salience and party attachment (Ray, 2003).
Support for government has been an explanatory variable in studies regarding public opinion towards European integration (Franklin et al., 1995). When public opinion is badly informed, opinions can change more easily under salient factors. The most salient factor in Parliament is the government, so Franklin and colleagues (1995) state that support for government or government popularity in interaction with party allegiance influences public opinion. Not only party allegiance is important. Hooghe and Marks (2005) claim that as national elites are more divided, citizens are more cued to oppose European integration. This could attenuate the role government plays in the cueing of the public.
Finally, research over the past few years added immigration variables to the explanation of public opinion towards EU integration (Carey, 2002; Christin & Trechsel, 2002; Marks & Hooghe, 2003; McLaren, 2002). National identity can be conceptualised as intensity of feelings towards one’s country, level of attachment to the nation and other territorial entities, and fear of other identities and cultures. Strong feelings of national identity lead to lower support for European integration (Carey, 2002). Christin and Trechsel (2002) studied public opinion in Switzerland. They found that among the Swiss electorate, national political, economic and cultural dimensions have played a role in explaining their support for EU membership. National interest was a strong predictor in public opinion, next to perceived economic gains and losses. In this thesis, national interest was conceptualised as institutional attachment and image of neighbouring countries (Christin & Trechsel, 2002). This corresponds with the benefit and institutional hypothesis of Sánchez-Cuenca (2000) and adds a national, cultural factor. McLaren (2002) argues that prior research ignores a major explanation of public opinion towards European integration, the notion of perceived threat. She rejects the benefit hypothesis and even posits questions at the institutional view, since both perspectives assume rationality (McLaren, 2002). Hooghe and Marks (2005) test both the benefit hypothesis and the identity hypothesis and find that both are influential, with the identity hypothesis being the most influential.
This thesis
This thesis concentrates on the effect of perception of voice in national and European policy on public opinion towards European integration and across all countries of the European Union. These concepts are related to the democratic representation citizens experience. Two different hypotheses are expected. First, when perceived voice of European citizens on European policy is high, it is expected that European citizens are more likely to support European integration. Secondly, when
- 8 -
perceived voice of European citizens on national policy is high, it is expected that it is less likely to support for European integration. Why these hypotheses are drawn, and why this research question is asked, the next chapter will explain. Thereafter, a sample of the European population will be studied and analysed to formulate an answer to the main question of this thesis: “What is the effect of perceived voice of European citizens on national and European policy on public opinion towards European integration across countries?”
- 9 -
Theoretical substantiation
Rohrschneider (2002) already argued that perceptions of representation in a regime are key predictors in predicting mass support for the EU. He claims that representational performance confluences EU support in two different tracks, an output-based track and a procedure-based track. In the output-based track, he endorses the influence of economic evaluations. Besides this economic calculus approach, the procedural track focusses on the “citizens’ commitment to the idea of democratic representation. Additionally, McLaren (2002) points out that previous research may have been too focussed on rational calculations too much when it comes to explaining the public opinion towards European integration. In this thesis, perception is used to assess what people may consider when trying to make a rational choice. Just like the study by McLaren (2002), where perceived threat to national identity is used, rather than rational economic calculus alone. Franklin and colleagues (1995) posit that support for government influences support for integration. If there is a supportive basis for the government, government’s view towards complex matters, such as the call for European integration, is used to formulate an individual view on the matter. Therefore, it is believed that support for government and perceived voice in policy are to some extent intertwined. This could be attenuated by the level of dispersion of the national elites (Hooghe & Marks, 2005). National governments play a role in treaty negotiations and therefore domestic political evaluations should be involved when studying voting behaviour (Lubbers, 2008). Trust in national and international institutions are variables used by De Vreese and Boomgaarden (2005) in order to evaluate these domestic politics. Perception is believed to be important in forming an opinion and making rational decisions. Moreover, perception on the level of voice in policy, or democratic representation, might influence support for the government and may also influence the level of trust of institutions, which both in turn are believed to influence public opinion towards European integration (De Vreese & Boomgaarden, 2005; Rohrschneider, 2002). Therefore, the effect of perceived voice in national and European policy on public opinion towards European integration is tested in this thesis.
Mechanisms for the hypotheses
European policy hypothesis
Perceived voice in European policy is thought to possibly predict the differences in public opinion towards European integration. If citizens view their voice as valued, this should lead to an increased support towards the European Union. McLaren (2002) states that the institutions of the European Union are likely to be a contributor of change. If this change is the consequence of political choices supported by the individuals voting, personal opinion or general attitude, it would be trivial to expect a positive relationship between the perceived voice in European policy and the attitude towards European integration. This relationship should also be seen in the light of democratic representation. Rohrschneider (2002) argues that democratic representation is a condition of mass support for
- 10 -
Europe’s political integration. This can be viewed in economic terms as the output of publicly desired goods, but also as democratic procedures for expressing mass preferences (Dahl, 1989, as described in Rohrschneider, 2002). Perceived voice in European policy can be a proxy of the level of faith people have in the democratic system. Therefore, it is expected that a high perceived voice in European policy leads to more support for European integration.
National policy hypothesis
A more complex mechanism underlies the hypothesis of the perceived national voice and its effect on public opinion towards European integration. It can be believed that a feeling of valuation of an individual’s voice can be transferred to their feeling of voice in the EU (Anderson & Reichert, 1996; De Vreese & Boomgaarden, 2005; Franklin, Marsh & Wlezien, 1994). These concepts are likely to correlate to some extent. However, more probable is that a high perceived voice in national politics predicts a lesser support for European integration. The idea behind the hypothesis’ mechanism can be divided into two different categories of scholars. First, Sánchez-Cuenca (2000) describes the mechanism of low support for European institutions when support for national institutions is high. This institutional hypothesis may point towards a mechanism of high-perceived voice in national policy, leads to higher support for national government and governmental institutions, may lead to lower support for the EU and European institutions. The second stream of thoughts stem from Franklin and colleagues (1995). They argue that voters’ support for government is important. Voters with high support for government are more likely to follow the reasoning of their national government. If this national government is supporting European integration, this group will support integration as well. However, if support for national government is high and the national government is not supporting European integration, then support for integration is also lower in this group.
These two insights are not mutually exclusive. It is possible to follow both streams of thought. When perceived voice of national policy is high, people’s valuation of national institutions is possibly higher. With higher valuation of national institutions, the first theory says that the valuation of European institutions is lower. The concepts of valuation of European institutions and support for European integration are not the same, but it can be expected that low valuation of European institutions is related to low support for European integration. The second theory states that when support for government is high, which is most likely to follow from higher perceived national voice, voters are more likely to follow the line of reasoning of their government regarding European integration. Therefore, national governments’ view on European integration might reinforce the low support for European integration if governments’ view is unsupportive and it might attenuate the first reasoning if governments’ view on European integration is positive. If perceived voice in national policy is low, valuation of national policy is low and valuation of European is high. In this scenario, there is
- 11 -
probably no mediating effect of national governments’ view, because low perceived voice is believed to be affiliated with low support for government. With low support for government, voters will not evaluate European integration the same way the government does.
Figure 1 shows a proposed schema of the different potential scenarios. The first two mechanisms would predict that a higher perceived voice in domestic policy, could lead to a lower support for European integration and vice versa. This outcome would be in line with the hypothesised mediating role national institutions play in the evaluation of Europe’s ability to deliver on citizens’ interests (Rohrschneider, 2002). The last mechanism would predict that lower perceived national voice would lead to a higher support for government.
Figure 1 Theoretical schema of the mechanism of perceived national policy on public opinion for European integration
Mechanism for the research question
There are multiple mechanisms possible. When the hypothesised mechanisms are correct, there are four options. The first option consists of an expected high European voice and a low domestic voice. In this case, high European voice is hypothesised to lead to a higher support for European integration and low domestic voice would be expected to lead to higher support. On aggregate, this would lead to higher support for European integration. In the second case, perception on both European and domestic voice are high, which leads to a higher support vis-à-vis the higher European perceived voice and an unclear response from the higher domestic voice, since it is dependent of governmental stance on the subject. In case of a supportive government, there will be a weaker effect of perceived domestic voice on support for European integration. In case of a non-supportive government, there will be a clear counterfactual effect from perceived domestic voice on the relationship between
- 12 -
perceived European voice and support for European integration. This mechanism leads to an ambivalent outcome in which the strength of both hypothesised relationships is important to understand the complete mechanism. In the third case, European perceived voice is low, but national perceived voice is high. Support for European integration would be low because of the perceived lower European voice. At the same time, like in the previous case, the perceived domestic voice has an unclear effect on the outcome. Idem, the strength of both hypothesised relationships is instrumental to assess the outcome. In the last case, perception on both European and domestic voice are low. This would lead to a clear negative effect of perceived European voice towards EU integration. Besides that, low domestic support would have a positive effect on support for EU integration. The latter could cancel out the effect of the first relationship of European voice or could make a supportive base for European integration.
Conceptualization
Before the research question can be analysed, first the concepts must be conceptualised. The independent variable can be split up into two separate independent variables, perceived voice in European policy and perceived voice in national policy. Perception of influence in policy given the democratic structure is the underlying concept what will be measured. The dependent variable needs to be conceptualised too. Public opinion towards European integration is a term used by several different scholars (Boomgaarden et al., 2011; De Vreese & Boomgaarden, 2005; Eichenberg & Dalton, 2007; Hobolt & De Vries, 2016), but is not operationalised the in the same way across studies. In addition, there are concepts, such as Euroscepticism, mass support for Europe’s political integration, and support for membership in the EU that describe the same concept (Boomgaarden et al., 2011; Lubbers, 2008; Rohrschneider, 2002). Public opinion towards European integration is used, because that concept is used in the most studies and describes the underlying concept best and it is also a neutral term. Support for European integration would suggest pro-European feelings and Euroscepticism would suggest the opposite. In this context those two terms are sometimes used.
- 13 -
Methodology
Former studies have pointed out a few theoretical controversies and common methodological problems (Gabel, 1998b). Empirical studies relied primary on bivariate analyses, which may conceal intervening and spurious relationships. Therefore, this thesis controls for numerous alternative explanations and individual-level differences. The next problem Gabel (1998b) addresses is that of scholars using all kind of different dependent variables. This makes research on public opinion towards European integration hard to unite. Therefore, this thesis tests the hypotheses on different operationalisations of the dependent variable in order to justify the results of this thesis.
Operationalization of the variables
Dependent variableBoomgaarden, Schuck, Elenbaas and De Vreese (2011) call for a multi-dimensional approach to attitudes towards the EU, because of growing diversification. Terms like EU support and Euroscepticism are too one-dimensional. They discriminated between five dimensions of EU attitudes: performance, identity, affection, utilitarianism, and strengthening. This multi-dimensional approach is attempted to approach by selecting several dependent variables, which are arguably using parts of the same constructs as Boomgaarden and colleagues (2011) did. This thesis is also used to select questions for the factor analysis. Questions used were restricted to the Eurobarometer questions for replicability purposes and practical considerations.
Several questions from the Eurobarometer are entered in an explanatory factor rotated principle component factor analysis in Table 1. After recoding, three factors can be distincted from the factor analysis, which in total explain 56,75 percent of the variance. The first factor consists of five questions and is labelled as ‘General European evaluation and performance’. The second factor consists of three questions and is labelled ‘Sense of EU identity’. The questions within this label are related to perceived citizenship, knowledge of European rights, and attachment to the European Union. The third factor is labelled ‘Stance on immigrants’ and can be compared to the negative affection label by Boomgaarden et al. (2011). This immigration aspect of support for integration is a crucial addition to the discussion over the last years (Carey, 2002; Christin & Trechsel, 2002; De Vreese & Boomgaarden, 2005; Marks & Hooghe, 2003; McLaren, 2002).
Crohnbach’s Alpha, reflecting the reliability of the scale, was for label ‘General European evaluation and performance’ (.743), for label ‘Sense of EU identity’ (.740), and for ‘Stance on immigrants’ (.712). Also weighted scales are created and the correlations between these scales vary from .353 to .511. The biggest correlation is between the factors ‘General European evaluation and performance’ and ‘Sense of EU identity’ (r = .511) and the smallest correlation is between the factors ‘General European evaluation and performance’ and ‘Stance on immigrants’ (r = .353). The factors ‘Sense of
- 14 -
EU identity’ and ‘Stance on immigrants’ correlate a bit higher than the lowest correlation, but still low (r = .370). In the further analysis, the different factors are tested against established explanatory factors and of course the hypothesised explanatory variables.
Table 2 Factor analysis on public opinion towards European integration
Factors General European evaluation and performance Sense of EU citizenship Stance on immigration
At the present time, would you say that, in general, things are going in the right direction or in the wrong direction, in the EU? .827 On the whole, are you very satisfied, fairly satisfied, not very satisfied or not at all satisfied with the way democracy works in the
EU? .785
Would you say that you are very optimistic, fairly optimistic, fairly pessimistic or very pessimistic about the future of the EU? .730 Please tell me if you tend to trust or tend not to trust the European
Parliament .664
In your opinion, in terms of shared values, are EU Member States
close of not? .530
For each of the following statements, please tell me to what extent it corresponds or not to your own opinion. You know your rights as a
member of the EU .883
For each of the following statements, please tell me to what extent it corresponds or not to your own opinion. You feel you are a citizen of
the EU .805
Please tell me how attached you feel to the EU .677
Please tell me whether each of the following statements evokes a positive or negative feeling for you. Immigration of people from
outside the EU .877
To what extent do you agree or disagree with each of the following
statements? Immigrants contribute a lot to <OUR COUNTRY> .823
Please tell me whether each of the following statements evokes a positive or negative feeling for you. Immigration of people from
other EU Member States .657
Crohnbach's Alpha .743 .740 .712
Initial Eigenvalue 4.128 1.380 1.064
% Variance explained 37.53 12.55 9.67
Rotation Sums of Squared Loadings 3.46 2.97 2.52
Note: Principal components analysis with promax rotation (Kaiser normalization). Rotation
converged in five iterations. Factor loadings below .5 and factors with a Crohnbach's Alpha below .7 were omitted from the table. Loadings are sorted by size.
Independent variable
The main objective of this thesis is to find out whether perception in voice in policy, nationally as well as in European perspective, has an influence on the level of support for European integration. This is best operationalised by the questions: “Please tell me to what extent you agree or disagree with each of the following statements; (1) My voice counts in the EU and (2) My voice counts in <OUR COUNTRY>.”. Answers ranged from ‘Totally agree’ to ‘Totally disagree’ on a 5-point scale with the opportunity to choose ‘Don’t know’, which is coded as three. In order to create
- 15 -
dummy variables for the regressions, these variables were recoded into a 0-1 binary scale, with 0 being the person felt his/her voice was not heard in general and 1 being their voice was heard.
Control variables
The variables controlled for in this thesis are gender, age, education, occupation as individual-level characteristics, in line with several earlier studies (Anderson & Reichert, 1996; Gabel, 1998b). The variables gender and occupation needed recoding into dummy variables. Moreover, in this thesis alternative explanations such as cognitive mobilisation, political values and class partisanship, and utilitarian appraisals are also considered and controlled for (Franklin et al., 1995; Gabel, 1998b; Inglehart, 1970a; 1970b; Inglehart et al., 1991; Ray, 2003; Rohrschneider, 2002). High and low cognitive mobilisation, consisting of two components, political interest and political knowledge, were coded as two dummies. The questions used for the construction of this variable are the political interest index and the scoring on three EU politics questions. In addition, political values and class partisanship, meaning individuals are shaped by their party affiliation and their level of nationalism is operationalised as a Left to Right index of peoples voting habits, were coded as dummies of Left- and Right-wing political partisanship. Utilitarian appraisals, which states that differences in economic welfare shape attitudes towards integration, is controlled for with the question whether an individual perceived benefit from the EU on different types of policies, following the definition of Gabel and Palmer (1995, as described in Gabel 1998b) (Rohrschneider, 2002). Finally, support for government was added to the list of control variables. Support for government is linked to support for European integration via party affiliation (Franklin et al., 1995).
Analysis
First, the data will be checked to have an image of the population. Secondly, the coherence between the explanatory variables, the different covariates and the outcome variables will be evaluated. Thirdly, the influence of the independent variables on the separate dependent variables will be tested in three different models. Fourthly, the influence of perceived voice on the three different components of support will be studied across all nations. Finally, logistic regression will be used to evaluate the predictive power of the model per country.
- 16 -
Results
Data
In this thesis, the data from Eurobarometer 89.1 (March 2018) was used. This dataset consists of 27,988 people across the European Union and 5,142 people from countries that can be associated with the EU. During this thesis, only the data from EU28 Member States was used to test the hypotheses. The exact distribution of the different groups over the countries can be found in Table 1 (see Table 1 in the appendix). Over the whole sample, 14,576 women and 13,412 men participated. 3,631 people aged from 15-24 years old, 6,460 from 25-39, 7,021 from 40-54, and 10,876 were older than 55 years old. They stopped full-time education at different ages. 4,591 stopped before 15 years old, 11,869 stopped between 16-19 years old, and 8,459 individuals stopped their education at an age higher than 20. 2,376 people were still studying. They can also be categorised into different job types, self-employed (2,133), managers (,3072), other white collars (3,167), manual workers (6,014), house persons (1,493), unemployed (2,054), retirees (7,680), and students (2,376). For a distribution per country, check Table 2 (see Table 2 in the appendix).
For this thesis a selection of multiple questions was made. Eleven questions were used to derive the underlying factors of the dependent variable. These questions loaded on three different factors and these factors were used to measure the effect of the perceived voice in policy in different constructs of the concept support for European integration. Eurobarometer data was used because of the open availability of the data and the standardization of the questions. This way, the thesis is replicable in the same and other periods and the same dependent variable can be used to research this field. To derive the underlying factors, proxies of questions in other studies were used (Boomgaarden et al., 2011). In Table 2 (see above) the factors are presented and the loading per question is presented. These three factors are all different components of the concept public opinion towards European integration.
The control variables utilitarian appraisal and support for government are constructed from a battery of several questions. Support for government consists of the questions: (1) “Are you satisfied in with the situation in your country in general?”, (2) “At the present time, would you say that, in general, thing are going the right way in your country?”, and (3) “On the whole, are you satisfied with the way democracy works in your country?”. These three questions are merged into the control variable on a 5-point scale. Utilitarian appraisal consists of the questions “Have you benefited from: (1) Less border controls, (2) improved consumer rights, (3) less mobile phone costs, (4) medical assistance abroad, (5) air transport passenger rights, (6) working abroad, (7) living abroad, and (8) studying abroad?”. Answers are given on a 3-point Likert scale and are recoded into a single control variable.
- 17 -
Analysis
Correlations
Before all regressions, correlations between all the explanatory, control, and outcome variables are studied. Table 3 (see appendix) shows the results of those correlations. One of the correlations that stands out is the moderate relationship between ‘Support for government’ and ‘General evaluations and performance’, r (27986) = -.464, p < .001. Next, between the first two factors, ‘General evaluations and performance’ and ‘Sense of EU identity’, there is a moderate link with the explanatory variable ‘Perceived voice in the EU’, r (27986) = .399, p < .001 and r (27986) = .359, p < .001. In addition, there is a medium link to be found between the factors ‘General evaluations and performance’ and ‘Sense of EU identity’, r (27986) = .379, p < .001. The rest of the correlations are below r = .300 and therefore too small to point out.
Linear regressions
A multiple linear regression was calculated to approach the predictions that will be done in the next part. The outcome of all linear regressions per model can be found in Table 4. Table 5, 6 and 7 give insight in the regression equations per country for all separate factors of public opinion towards European integration. Public opinion towards European integration can be split into three different factors, ‘General evaluation and performance’, ‘Sense of EU identity’, and ‘Stance on immigration’. These components will be referred to hereafter as factor 1, factor 2 and factor 3. From Table 4, the first thing that attracts the attention is the increase of the explanatory variance in all components of public opinion towards EU integration. Factor 1 has an initial explanatory power of 16,2% and this increases to a substantial height of 29,5%. Same goes for factor 2, from 17,9% to 28,2% and factor 3, from 9,5% to 19%. In model 1, both variables explained a significant proportion of variance in factor 1, R2 = .162, F(2, 27985) = 2701,94, p = < .001. In M2, a significant effect was found for perceived
European voice, b = .706, t(25856) = 50,83, p < .001, and for perceived domestic voice, b = .149, t(25856) = 10,45, p < .001. In this model 2, both variables explained a significant proportion of variance in factor 1, R2 = .176, F(11, 25856) = 500,85, p = < .001. In M3, a significant effect was found
for perceived European voice, b = .581, t(21480) = 41,24, p < .001, and for perceived domestic voice, b = -.071, t(21480) = -4,70, p < .001. In model 3, both variables explained a significant proportion of variance in factor 1, R2 = .295, F(17, 21480) = 529,03, p = < .001. Model 3 is the best fitting model in
all linear regression and in all factors.
In Table 5, 6 and 7 the linear regression equations are shown per country for model 3. What stands out is that on the first factor Italy and Romania have the smallest margin of significance. Furthermore, all regression equations for perceived voice in European policy is positive, while
- 18 -
significant perceived voice in national policy gives mixed results. Moreover, for different countries, different factors seem the most salient in explaining public policy towards European policy.
Table 4 Linear regressions per model over all countries
General evaluation and performance Sense of EU identity Stance on immigration
M1 M2 M3 M1 M2 M3 M1 M2 M3
My voice counts in the EU .722*** .706*** .581*** .814*** -.322*** .636*** .525*** .501*** .413*** My voice counts in <OUR
COUNTRY> .142*** .149*** -.071*** .135*** -.015** -.015 .159*** .150*** .012 Gender (Female) - .055*** .088*** - -.072*** -.009 - -.033*** -.016 Age - -.003*** -.004*** - -.005*** -.005*** - -.003*** -.003*** Education - .004*** -.002* - .022*** .012*** - .017*** .012*** Occupation Self employed - .027 -.016 - .082*** .040 - .057* .053 Managers - .141*** .016 - .206*** .035 - .207*** .140*** Other white collars - .080*** .024 - .065*** -.035 - .005 -.003 House persons - .008 -.021 - -.028 -.069* - .062 .076* Un-employed - -.142*** -,093*** - -.136*** -.144*** - -.029 .012 Manual workers - 0 -.025 - 0 -.058** - 0 -.011 Retired - .029 0 - .050* 0 - -.006 0 Students - 0 0 - 0 0 - 0 0 Cognitive mobilisation Low - - .055*** - - .202*** - - .044* High - - .125*** - - .401*** - - .115*** Class partisanship and political values Left - - .018 - - -.047*** - - .237*** Right - - .011 - - -.046** - - -.187*** Utilitarian appraisal - - .121*** - - .282*** - - .121*** Support for government - - -.304*** - - -.141*** - - -.181***
r .402 .419 .543 .423 .494 .531 .308 .346 .436 R2 .162 .176 .295 .179 .244 .282 .095 .120 .190 F value 2701,94*** 500,85*** 529,03*** 3049,43*** 757,09*** 496,69*** 1466,42*** 319,07*** 295,75*** B 2,734*** 2,738*** 3,742*** 2,855*** 4,188*** 2,770*** 2,610*** 2,404*** 2,869*** SE .009 .034 .053 .009 .038 .055 .009 .035 .057 N 27988 25868 21498 27988 25868 21498 27988 25868 21498 * p < 0.05, ** p < 0.01, *** p < 0.001
With the increase of the variance explained, there is a decrease in the influence of perceived voice on the regression equation. However, in all models, the standardised regression coefficient of the perceived voice in the EU remains the largest. The second explanatory variable, perceived voice in the country is far less influential. A significant regression equation was found in model 3 (M3) on factor 1 (F1) for perceived voice in policy in the EU, b = .581, t(21480) = 41,24, p < .001, and for perceived voice in policy domestically, b = -.071, t(21480) = -4,70, p < .001. On F2, perceived voice in policy in the EU, b = .636, t(21480) = 43,24, p < .001. Perceived voice in policy domestically is not significant in M2, t(21480) = -0,95, p = .342. Finally, on F3, perceived voice in policy in the EU, b =
- 19 -
.413, t(21480) = 27,06, p < .001, and for this model goes the same as for M2, perceived voice in policy domestically is not significant, t(21480) = 0,73, p = .463. This leads us to believe that of the explanatory variables, perceived voice in the EU has the biggest impact on the model and explains the most variance.
Besides the explanatory variables perceived EU voice and perceived domestic voice, more variables have a noticeable significant effect on F1 in M3. It seems that ‘Support for government’ is the most dominant covariate, b = -.304, t(21480) = -56,59, p < .001. Hereafter, high ‘cognitive mobilisation’, b = .125, t(21480) = 6,83, p < .001 and ‘Utilitarian appraisals’, b = .121, t(21480) = 12,97, p < .001, seem to have the biggest impact on ‘General evaluations and performance’. On ‘Sense of EU identity’, the same covariates have noteworthy impact, albeit in different magnitudes. High ‘cognitive mobilisation’ is after perceived European voice the second most influential regression coefficient, b = .401, t(21480) = 20,92, p < .001. Then, ‘Utilitarian appraisal’, b = .282, t(21480) = 28,87, p < .001, low ‘cognitive mobilisation’, b = .202, t(21480) = 10,51, p < .001, and ‘Support for government’, b = -.141, t(21480) = 25,12, p < .001. Additionally, being unemployed influences ‘Sense of EU identity’, b = -.144, t(21480) = -4,90, p < .001. On the factor ‘Stance on immigration’, left ‘Class partisanship and political values’ have the second largest coefficient, b = 237, t(21480) = 16,03, p < .001. Next, right ‘Class partisanship and political values’, b = -.187, t(21480) = -12,09, p < .001, ‘Support for government’, b = -.181, t(21480) = -31,22, p < .001, ‘Utilitarian appraisals’, b = 121, t(21480) = 11,95, p < .001, and high ‘cognitive mobilisation’, b = .115, t(21480) = 5,79, p < .001, influence ‘Stance on immigration’. Being a manager seems to have an impact as well, b = .140, (21480) = 5,51 , p < .001.
Logistic regressions
Finally, to predict what variables have an impact on what factor of public opinion towards EU integration, a logistic regression is done per country. Tables 9 to 35 (see appendix) show us the results for each individual country. To interpret the coefficients of the logistic regression, the exponential of the coefficient (Exp(B)) is used. For goodness-of-fit robustness checks, the Hosmer and Lemeshow Test for logistic regressions is used. If the test exceeds .05, this points towards a good model fit. To help interpretation of the logistic regressions, France is scrutinised, and the results are reported below. The interpretation is the same for the rest of the countries and noticeable results will be discussed in the discussion section.
France (EU6)
The odds of general evaluation of the EU and its performance are lower for an individual in France that supports the domestic government, Exp(B) = .298, p < .001. This is equivalent of saying that the probability of a more positive evaluation of the EU and its performance is lower for every French national that supports its domestic government. For every answer on the 5-Likert scale of the
- 20 -
questions “Are you satisfied in with the situation in your country in general?”, “At the present time, would you say that, in general, thing are going the right way in your country?”, and “On the whole, are you satisfied with the way democracy works in your country?”, that is more supportive towards the domestic government, the general evaluation of the EU and its performance declines by a factor .298 or 70,2%. The opposite counts for perceived voice in policy in the European Union, Exp(B) = 2.059, meaning that if an individual in France agrees that his or her voice counts in the EU, the likelihood of a more positive general evaluation of the European Union and its performance rises with 2.059 or 105,9%. All other variables in the model did not have a significant impact on the factor ‘General evaluation and performance’. With the variables used, 46,1% of the variability of the model can be explained. The Hosmer and Lemeshow goodness-of-fit Test for logistic regression exceed .05, HL = .735. In terms of sense of European identity, education raises the probability of having some sorts of EU identity by 1,3%. This is small compared to the effect of ‘Utilitarian appraisals’, 110,9% increase per perceived benefit from the EU, and perceived European voice, 289,3% increase of the probability of having some sense of EU identity. Again, support for government stands for a decrease of 40,4% on the sense of European identity. 34,6% of the variance can be explained by the model and the HL test points towards a good fitted model. Stance on immigration can also be predicted by education, 10% increase, perceived European voice, 75% increase, and support for domestic government, 32,3% decrease. Besides, age, a decrease of 2%, and Left partisanship, 172,5% increase, have a significant predicting effect. 30,8% of the variance in the model can be explained.
- 21 -
Discussion
In this section, the effect of perceived voice in European policy and perceived voice in national policy on public opinion towards European integration will be discussed. This will be done per group of countries. The countries are simply grouped by accession period. Next, the control variables that also added predictive power to the model will be discussed. Table 8 shows which predictors are significant per country and on what factor. The precise regression equations per country can be found in Table 9 – 36. Finally, a conclusion and corresponding implications of the research question will be drafted.
Table 8 Significant predictors for each country and for each factor (1, 2, 3)
Member States of the EU6
This group of countries belongs to the founding six countries of the European Union. From the European Coal and Steel Community onwards, these three countries pushed towards a more
FR BE NL DE IT LU DK IE UK GR ES PT FI SE 1, 2, 3 1, 2 1, 2, 3 1, 2, 3 2, 3 1 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 1, 2 1, 2 1, 2, 3 - - - 1 1, 2 - - 2, 3 2 - - - - -- - 1 - - - 1, 3 - 1 - - - 1, 3 -3 - - 1 - - - 1 3 2, 3 - 2, 3 2, 3 3 - - - 3 2, 3 - 2, 3 3 3 Sel f-empl oyed - - - - 1 - - - - 3 - 3 -
-Ma na gers - - - 3 -
-Whi te col l a rs - 1 3 1 - - - 3 - -Ma nua l workers - 1 - - 1, 2 - - - 3 - - - 3 -Hous e pers ons - - - - 1 - - 3 3 3 - - 3 -Unempl oyed - - - - 1 - - - 3 Left 3 - 2, 3 - - - 3 - 1, 2, 3 - - 3 3 3 Ri ght - - 3 2, 3 2 2 - - 1, 2, 3 3 3 3 - -Hi gh - 2 1, 2 2 - - 1 1, 2, 3 1, 2, 3 2 2, 3 2, 3 - 3 Low - - 1 1, 2 - - 1 3 1, 2, 3 2 2 - - -2 1 2 1, 2, 3 1, 2, 3 - 1, 2, 3 2 1, 2, 3 2 3 - 2, 3 1, 2 1, 2, 3 1, 2 1, 2, 3 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 1, 2, 3 1 1, 2 1, 2 1, 2, 3 1, 2, 3 1, 2, 3 AT CY CZ EE HU LV LT MT PL SK SI BU RO CR 1, 2, 3 1, 2 1, 2, 3 1, 2 1, 2, 3 1, 2 1 1, 2 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 2, 3 1, 2, 3 3 - 1 1 3 - - 1 - 1 - 2 1, 2 2, 3 1, 3 3 - - - -- 1, 3 2 1 - 3 - - 1 1, 3 1 - 2 -1, 3 - - - - 2 - - 1 2 1, 3 2 2 1, 2 Sel f-empl oyed - - - 1 1 - - - 2
-Ma na gers 1 - - 1 - - - 1 -
-Whi te col l a rs - - - 1, 2 1, 2 - - 2 1 Ma nua l workers 1 - 2 1 1 2 - - - 1 - - 2 -Hous e pers ons - 3 - - - 3 - - 2 2 Unempl oyed - - - 1 - - - 1 - - 2 1 Left 1, 3 - - - 1 1 1 - - - 3 - 1 2 Ri ght 1, 2, 3 - 1, 2 3 - - 1 - 1, 2 1 - 2 1 2 Hi gh 2 - - 2 1, 2 2 1, 2 - 2 2 - - 2, 3 2, 3 Low 1 - - - 2 2 1 - - - 2 -1, 2, 3 2, 3 1, 2 2 1, 2, 3 2 3 2 2, 3 2 2 2 3 1, 2 1, 2, 3 1 1, 2, 3 1, 2, 3 1, 3 1, 2 1, 3 1, 2 - 1, 2, 3 1, 2, 3 1, 2, 3 1, 2 1, 2, 3 Pa rtis a ns hi p Cogni tive mobi l i s a tion
Util i tari a n a ppra i s a l s Support for Government Occupa tion
Occupa tion
Percei ved voi ce i n <OUR COUNTRY> Percei ved voi ce i n the EU
Gender Age Educa tion Pa rtis a ns hi p Cogni tive mobi l i s a tion
Util i tari a n a ppra i s a l s Support for Government Percei ved voi ce i n the EU Percei ved voi ce i n <OUR COUNTRY>
Gender Age Educa tion
- 22 -
integrated Europe. In France, Belgium, and the Netherlands, perceived voice in European policy is a significant predictor. In these countries, a higher perceived voice leads to a more supportive attitude towards European integration and a lower perceived voice leads to a more Eurosceptical view on EU integration. In Germany, European perceived voice is significant in the factors ‘General evaluation and performance’ and ‘Sense of EU identity’, but not on the factor ‘Stance on immigration’. In Luxembourg, perceived European voice is significant for the factor ‘General evaluation and performance’, but not for the other two factors. Finally, in Italy, perceived voice in European policy is a significant predictor in the model that evaluates ‘Sense of EU identity’ and ‘Stance on immigration’. This counts not for the general evaluation of the European Union and its performance. Besides this, Italy is also the only country in the EU6 where perceived voice in national policy predicts variance explained by the model. A higher perception of domestic voice leads to a better general evaluation and performance of the EU and a better sense of identification with the European identity. In Germany, perceived national voice also accounts for some predicted probability in the model for general evaluation. In the other four countries, France, Belgium, the Netherlands and Luxembourg this is not the case.
Member States of the EU9
These three countries applied for accession together. After a failed attempt to join the European Communities in 1963, these countries joined the EC in 1973. Characteristic for these countries is that all have had referenda on whether they should ascend the EC and on what terms. In Denmark, Ireland and the United Kingdom, perceived voice in European integration can significantly predict some of the variance. In all these countries, higher perception of voice in EU policy leads to a more supportive attitude towards EU integration, whilst, a lower perception leads to a less supportive view on more European integration. Moreover, in Ireland, higher perception of domestic voice leads to higher levels of support for EU integration and vice versa for the factors ‘Sense of EU identity´ and ‘Stance on immigration’. In the United Kingdom the same counts for ‘Stance on immigration’. In Denmark, no such predictionary affect has been found.
Member States of the EU12
The next wave of countries that ascended the European Community can also be referred to as the Mediterranean enlargement. This enlargement happened in two steps. In 1981, Greece was admitted to the Community and was followed by Spain and Portugal in 1986. In Greece, Spain and Portugal, perceived European voice is a significant predictor for the components ‘General evaluation and performance’ and ‘Sense of EU identity’ of public opinion towards European integration. The predicted relationship between perceived European voice and the separate components of public opinion is positive. In Greece, the model does not fit the last factor ‘Stance on immigration’ good
- 23 -
enough, whilst in Spain and Portugal this is no significant predictor in the model. In all three countries, neither is perceived voice in domestic policy a significant predictor.
Member States of the EU15
In 1995, Finland, Sweden and Austria joined the European Union. All three countries used to be part of the European Free Trade Agreement and wanted to join the European Union to integrate the internal markets. All three countries held referenda resulting in accession to the EU. Is Sweden and Austria, perceived European voice can significantly predict variance explained by the model in all three factors. In Finland this is the case for ‘General evaluation and performance’ and ‘Sense of EU identity’, but not for ‘Stance on immigration’. High perception of voice in the European Union leads to support towards EU integration and low perception of EU voice leads to Euroscepticism. High perceived voice in national policy leads to less support for European integration in the model on the factor ‘Stance on immigration’ in Austria. All other models in these countries do not include perceived national voice as a significant predictor.
Member States of the EU25
In 2004, ten Central and Eastern European Countries joined the European Union. They are sometimes referred to as the A10. They consist of Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia, and Slovenia. After the Cold War, these countries searched for closer ties with the West of Europe and this eventually led to the accession to the EU. First, the model on ‘Sense of EU identity’ in Czech Republic and Estonia is rejected due to a lack of fit. Therefore, these models will not be further evaluated. Concerning the predictive ability of perceived voice in European policy, there is a positive relationship of the model with all components in Czech Republic, Hungary, Poland, Slovakia and Slovenia. In Cyprus, Malta, Estonia, and Latvia this is the case for the factors ‘General evaluation of the EU and its performance’ and ‘Sense of EU identity’. In Lithuania, perceived voice only has a significant predictive ability on the first factor. Regarding perceived voice in national policy, this gave mixed results. Three categories of countries can be discriminated. First, in Cyprus, Latvia, Lithuania, Poland and Slovenia, this variable is no significant predictor of any model. Secondly, in Estonia, Hungary, and Malta, high perceived voice in domestic policy leads to a less supportive attitude towards European integration on different factors. In Czech Republic and Slovenia however, high perceived voice in national policy matters does lead to a more supportive attitude on the factor of general evaluation.
Member States of the EU27
Both Bulgaria and Romania applied for accession together with the A10, but due to backlog on some reforms, the accession date was three years later, in 2007. In Bulgaria, perceived European voice is a significant predictor in the models on general evaluation and European identity. Like in all other
- 24 -
cases, this relationship is a positive one. Perceived national voice is only significant in the model on the factor ‘Sense in EU identity’. The model does not fit ‘General evaluation and performance in the EU’ in Romania. Therefore, this factor will be disregarded. On the other two factors, perceived European voice has predictive power. Higher perceived voice in Europe will probably lead to more sense of European identity and a more positive stance on immigration. Again, a lower perceived voice leads to more Eurosceptical views on these factors of public opinion. A higher perceived voice in national policy leads to more identification with the EU and a lower perceived voice is more probable to lead to a lower sense of European identity.
Member States of the EU28
Finally, Croatia currently is the last accessor of the European Union. They entered in 2013 after applying in 2003. Perceived voice in European policy is a significant predictor in all models on all factors, with a higher perception of voice leading to a more supportive opinion regarding European integration. Perceived national voice is also an important predictor in the model on the factors ‘Sense of EU identity’ and ‘Stance on immigration’. In both cases a higher perceived voice in Croatia is more likely to lead to a less supportive view on European integration.
- 25 -
Other variables in the model that play a role
In the Netherlands, Denmark, the United Kingdom, Finland, and Austria, women were more supportive towards European integration. This opposes the view of Gabel (1998b) who claimed that men were more likely to benefit from the EU and therefore more supportive. In Cyprus on the other hand, this was the case. Older people in Slovenia, Poland and Cyprus were more likely to evaluate the EU and its performance more positively than younger people. In Cyprus older people were also more supportive on immigration. However, in France, Germany, Finland, Sweden, Latvia, Estonia, Slovakia and Romania younger people were more supportive of European integration than the elderly. There was no country where education made a huge significant impact. In several countries more educated people were more likely to be more supportive towards one or more components of European integration. This was the case for France, the Netherlands, Germany, Italy, the United Kingdom, Greece Spain, Portugal, Finland, Sweden, Austria, Latvia, Slovakia, Slovenia, Bulgaria, Romania, Croatia. This is in line with the hypothesis as posed in the study by Gabel (1998b). In contrast to what we might expect from this research, in Poland, people with higher education were less likely to evaluate the European Union and its performance more positively, than people with lower education. In the tables 8 – 35 in the appendix all logistic regression models are shown. We can see that there is no occupation that has a clear significant predictive effect in all cases. In different countries several occupations made a significant difference in the models.
The most influential control variable in the model is support for government. This variable was build up from three different questions: (1) “Are you satisfied in with the situation in your country in general?”, (2) “At the present time, would you say that, in general, thing are going the right way in your country?”, and (3) “On the whole, are you satisfied with the way democracy works in your country?”. In all models, except the model on ‘Sense of EU identity’ in Malta, there is an inverted relationship between support for the domestic government and support towards European integration. Hence, if a European citizen is generally satisfied with the situation in his or her country, believes things are heading the right way, and is generally satisfied with the way democracy works in his or her country, then there is probably less support for European integration. According to several different studies (Rohrschneider, 2002; Sánchez-Cuenca, 2000) this would be the case. All countries, besides Luxembourg and Estonia, have utilitarian considerations as a significant predictor of the model on at least one of the factors. Utilitarian considerations are the appraisal of costs and benefits from integrative policy (Gabel, 1998b). This was tested through eight different self-reporting benefit questions. Gabel (1998b) stated that people with larger (human) capital are more likely to benefit from the integration of the markets and they are therefore more likely to support more European integration. The results indeed point in that direction. Partisanship plays a role in many models. Left
- 26 -
class partisanship should lead to less support for European integration than Right class partisanship (Inglehart et al., 1991). However, this effect may vary with differences in party affiliation, disagreements within parties, issue salience and party attachment (Ray, 2003). In general, we found that citizens who deem themselves politically left orientated are more supportive towards European integration. In Portugal, Latvia, Lithuania, and Croatia though, these people are less supportive towards EU integration. People who regard themselves as politically right are generally less supportive of European integration, however in the Czech Republic, Estonia, Lithuania and Slovakia, these people are more supportive in their opinion towards European integration. All self-reported answers are relative to the perceived centre of the political index. And the centre is not directly comparable with other countries. Spain and Portugal for instance have a Social-Democratic prime minister, whereas Poland has a right winged government. Therefore, all these results should be interpreted in the light of each national government. Cognitive mobilisation is the needed to grasp the full idea of the complex political arena. Therefore, Inglehart (1970a) expects that this is a condition for mass support for the EU. From this model we can say that this is the case for several countries, such as Belgium, Germany, Ireland, United Kingdom Greece, Spain, Portugal, Sweden, Austria, Hungary, Latvia, Lithuania, Poland, Slovakia, Romania, and Croatia. However, some of the citizens in these countries, who have scored low on objective political knowledge and perceived political interest, are more supportive towards European integration as well. Examples are Ireland, United Kingdom, Greece, Spain, Hungary, Latvia, Lithuania, and Romania. Germany and the Netherlands report an ambivalent image, where one group is supportive on one factor of opinion towards EU integration and the other group is not. Hence, cognitive mobilisation is an important predictive term in our model, and high cognitive mobilisation is indeed related to a more supportive view, but low cognitive mobilisation is not necessarily unsupportive towards European integration. It could be possible that low cognitive mobilised citizens are successfully cued by the political elites, but this is more a subject to future research.
- 27 -
Conclusion
After reviewing the hypothesised relationships in all European countries, a few conclusions can be drawn. The first hypothesis “A high perceived voice in European policy would predict more support towards European integration” can be confirmed in almost all countries on at least the most important first factor of public opinion for European integration. In Italy, this was not the case, but perceived voice in European policy did predict the other two factors. In Romania, the model was rejected on the first factor, but again this variable has predictive power on the other two factors. This can also be expected from Table 5 (see Appendix). Italy and Romania are the two countries with the smallest margin of significance. The second hypothesis “A high perceived voice in European policy would predict less support towards European integration” can be confirmed in some countries, but needs to be rejected in other countries. This ambivalent outcome across nations, makes that no clear overarching conclusion can be drawn concerning the effect of perceived voice in national policy. This is also anticipated in the theory section. Various theories have attempted to clarify the relationship between national institutions and European institutions, but not unambiguously (Franklin et al., 1995, Rohrschneider, 2002; Sánchez-Cuenca, 2000). A possible schema is proposed earlier, but this relationship should be further investigated in future research. The most important contribution of this thesis is the predictive power perceived voice in policy yields. In future research on public opinion towards European integration, these results should be taken into consideration. Though it is not possible to explain all the variance in public opinion, attention should go to explaining more variance and substantiate the underlying mechanisms in one overarching theory.
Finally, some limitations and implications for future research. In this thesis, the Eurobarometer was used to standardise research in this area, so that thorough research on public opinion towards European integration is possible. In addition, this use of a single database, could also help to overcome future dependent variable problems. A limitation to this method is that all future research should be in line with the questionnaire of the Eurobarometer. Also, the Eurobarometer has sample limitations. In this version, the elderly and retirees are overrepresented for instance. This could be solved in the future by weighing the population outcomes or attaining a more balanced dataset.
- 28 -
References
Anderson, C. J. (1998). When in doubt, use proxies: Attitudes toward domestic politics and support for European integration. Comparative political studies, 31(5), 569-601. Retrieved from
https://doi.org/10.1177/0010414098031005002
Anderson, C. J., & Reichert, M. S. (1996). Economic benefits and support for membership in the EU: A cross-national analysis. Journal of Public Policy, 15(3), 231-249. Retrieved from
https://www.cambridge.org/core/journals/journal-of-public-policy/article/economic-
benefits-and-support-for-membership-in-the-eu-a-crossnational-analysis/700139A9A0C6438CAA0DCD15075E9B7B
Baldwin, R., & Wyplosz, C. (2015). The economics of European integration (5th ed.). London, England:
McGraw-Hill Education.
Boomgaarden, H. G., Schuck, A. R., Elenbaas, M., & De Vreese, C. H. (2011). Mapping EU attitudes: Conceptual and empirical dimensions of Euroscepticism and EU support. European Union Politics, 12(2), 241-266. doi:10.1177/1465116510395411
Brinegar, A. P., & Jolly, S. K. (2005). Location, location, location: National contextual factors and public support for European integration. European Union Politics, 6(2), 155-180. doi:10.1177/1465116505051981
Bruter, M. (2003). Winning hearts and minds for Europe: The impact of news and symbols on civic and cultural European identity. Comparative political studies, 36(10), 1148-1179.
doi:10.1177/0010414003257609
Carey, S. (2002). Undivided loyalties: is national identity an obstacle to European
integration?. European union politics, 3(4), 387-413. doi:10.1177/1465116502003004001
Christin, T., & Trechsel, A. H. (2002). Joining the EU? Explaining public opinion in
Switzerland. European Union Politics, 3(4), 415-443. doi:10.1177/1465116502003004002
CVCE (July 8, 2016). Historical events in the European integration process (1945-2014) [Web article]. Retrieved on April 23, 2019 from
https://www.cvce.eu/en/collections/unit-content/-/unit/02bb76df-d066-4c08-a58a-d4686a3e68ff
De Vreese, C. H., & Boomgaarden, H. G. (2005). Projecting EU referendums: Fear of immigration and support for European integration. European Union Politics, 6(1), 59-82.
