Who leads whom? : eurosceptic cueing of the Party for Freedom examined

Who Leads Whom? Eurosceptic Cueing of the

Party for Freedom Examined

Master thesis Political Science: International Relations

Student: Leon Punt (10411321)

First reader: Gijs Schumacher, PhD Second reader: Eelco Harteveld, PhD

2 Abstract

Euroscepticism has been extensively studied over the last two decades. Scholars have discussed the causes of increasing Euroscepticism among electorates and political parties. Furthermore, attention has been paid to the nature and intensity of Eurosceptic attitudes. More recently, studies have tried to determine the direction of the causal effect between elite and public attitudes. In other words, who follows whom regarding Eurosceptic opinion-making? This study tests the argument of elite cueing. LISS panel-data are used in order to longitudinally examine changes of individual attitudes before and after voting for the PVV. Paired t-tests show that first-time PVV voters in the 2010 and 2012 parliamentary elections clearly become more opposed towards European unification after switching to the party. Moreover, fixed-effects regression analyses demonstrate that this effect is caused by voting for the PVV. The results also show that post-election attitude changes regarding European unification are moderated by the degree of sympathy towards Wilders and level of political interest. Cueing effects are particularly strong for first-time PVV voters with a high degree of sympathy towards Wilders or a low level of political interest. Future research should focus on the stability of these attitude changes and the underlying psychological mechanisms of cueing.

3 Table of contents

1. Introduction………..… 4

2. Far-right Euroscepticism………..… 6

2.1 Defining Euroscepticism……….………..….. 6

2.2 From ‘permissive consensus’ to ‘constraining dissensus’….……… 8

2.3 Left-wing versus right-wing Euroscepticism……….………. 10

3. Do politicians follow voters?... 12

3.1 Party identification and cue taking………. 13

3.2 Euroscepticism: who follows whom?... 16

3.3 The psychology of cue taking………. 17

4. Hypotheses………. 19 5. Methodology………. 21 5.1 Data selection……… 21 5.2 Case selection……… 22 5.3 Measures……….... 23 5.4 Regression model………..……… 24 5.5 Descriptive statistics………..……. 25 6. Results……… 27 6.1 Paired t-tests………. 27 6.2 Regression analyses………. 29 7. Conclusion……….. 31 8. Discussion……… 33 9. Bibliography……….. 35

4 1. Introduction

Euroscepticism has gone from the margins to the mainstream and is arguably more salient than ever (Brack and Startin, 2015). Large segments of European populations are not satisfied with the current workings of the EU and oppose further unification of European states (Van Elsas, 2017). Eurosceptic parties were able to articulate these feelings of disconnect with the EU and European unification at the 2014 European parliament elections. One year prior these elections the Eurobarometer showed that 66 percent of European voters thought that their opinion was unheard in the EU. Moreover, nearly 50 percent declared to be dissatisfied with the functioning of European democracy (Eurobarometer 79.5 and 80, 2013). Recent survey research underlines this sense of Euroscepticism among European electorates. Pew research, a public opinion research center studying global attitudes and trends, found that across the publics of ten EU member-states a majority of just 51 percent of respondents was favourable towards the EU in 2016. On the other hand, 47 percent of these electorates had an unfavourable view regarding the EU. Furthermore, merely 19 percent of respondents were favourable towards an increase of power transfers from national governments to the European level, whereas 27 percent argued that the division of powers should remain the same. Across this broad spectrum of EU member-states, including countries economically and politically as diverse as Germany and Greece, 42 percent of respondents stated that some powers should be returned to national governments (website Pewresearch, 2016). These findings are in line with scientific studies that show increasing sceptical attitudes among European publics and political parties concerning the EU and European unification (Brack and Startin, 2015; McLaren, 2007; Usherwood and Startin, 2013).

There is debate on the interplay between parties and publics regarding attitude formation. Traditional models deriving from the perspective of issue voting state that voters elect politicians based on their policy stances. In this view voter attitudes determine policy positions of parties (Pennock, 1979). More recent research finds that voters elect politicians based on character traits and past performances (Lenz, 2012). From this perspective voters align to the policy stances of a party after voting. There is a great variety of contemporary studies linking attitudinal changes to voting behaviour (Gilens and Murawaka, 2002; Steenbergen et al., 2007). In the literature this process is referred to as elite-cueing (Brader

5 and Tucker, 2010). Research has mainly focused on cueing effects in the US, although more recently scholarly interest has shifted to European countries (Merolla et al., 2008).

The question whether voters take elite cues after voting for a Eurosceptic party has received increasing attention in recent years (Brader and Tucker, 2012). Empirical evidence in this field has been inconclusive. Some research finds evidence for a bottom-up relationship. From this perspective elite positions on European unification are shaped by European electorates (Carrubba, 2001). Other studies, however, find a top-down relationship between party positioning and public attitudes. In this view voters align to stances of political parties on European unification through a process of information and persuasion (Gabel, 1998; Ray, 2003; Steenbergen et al., 2007).

Studies have consistently found correlations between Eurosceptic attitudes and positioning on the extremes of the political spectrum (De Vries and Edwards, 2009; Van Elsas et al., 2016; Van Elsas; 2017). The Dutch Party for Freedom (PVV) is an example of a far-right Eurosceptic party. Their prime concerns on Dutch EU-membership in particular and further European unification in general are a loss of national sovereignty and immigration as a result of fading border controls (PVV manifesto, 2010; 2012). The nature and firmness concerning the Eurosceptic rhetoric of the PVV are characteristic for right-wing Euroscepticism. The Euroscepticism of far-right parties is primarily driven by cultural issues. On the other hand, economic issues such as welfare policy are characteristic for far-left Euroscepticism. Eurosceptic parties on the far-left express their concerns with the current functioning of European institutions but do not advocate a break-up of the EU (Van Elsas et al., 2016; Van Elsas, 2017).

Scholars have argued that despite increased scholarly interest the understanding of the causal relationship between party positioning and public attitudes remains incomplete (Steenbergen et al., 2007). The aim of this research is to test the argument of elite cueing in the context of the Netherlands. Does voting for the far-right Eurosceptic PVV causes changes in attitudes on European unification? It is the contention of this thesis that the lack of consensus regarding the causal direction between party positioning and voter attitudes makes research worthwhile. Survey data from the Longitudinal Internet Studies for Social Science (LISS) panel will be used to test the development and persistence of Eurosceptic attitudes. In order to do so, samples of first-time PVV voters during the 2010 and 2012 parliamentary

6 elections will be created. The analysis covers the period 2007-2013 in order to compare pre- and post-stimulus data.

The structure of this thesis is as follows. The next section starts with conceptualising Euroscepticism. Moreover, the history of Euroscepticism is addressed in order to demonstrate the development of Eurosceptic attitudes over time. Next, specific attention is paid to the far-right characteristics of Euroscepticism. The following section will discuss the debate on elite-mass cueing and its empirical findings in the context of European integration. Moreover, attention is paid to the psychology of cue-taking in order to identify causal mechanisms. Third, hypotheses will be formulated based on theoretical expectations. Next, the operationalisation of concepts, methodology of research and collected data will be discussed. The results of the t-tests and regression analyses are demonstrated in the following section. Finally, conclusions will be drawn and discussed in the last two sections.

2. Far-right Euroscepticism

This section starts with the conceptualisation of Euroscepticism. Next, the development of Eurosceptic attitudes is demonstrated from a historical perspective. Moreover, explanations concerning the rise of Eurosceptic attitudes among parties and publics are outlined. The third part of the section identifies the intensity and nature of far-right Euroscepticism.

2.1 Defining Euroscepticism

Eurosceptics reject institutionalised co-operation between European states, although to varying degrees (Fitzgibbon et al., 2017). Therefore, the terminology of Euroscepticism begs for clarification. Taggart (1998) provided the first widespread academic definition of Euroscepticism. This definition will be followed in this research as it concisely captures the main features of the concept. Taggart defines Euroscepticism as the opposition and doubt to the process of European integration. He argues that increasing levels of Euroscepticism among European parties and electorates go hand in hand with increasing unification of European states. European unification is defined as the process 'whereby political actors in several, distinct national settings are persuaded to shift their loyalties, expectations and political activities toward a new centre, whose institutions possess or demand jurisdiction over the

7 pre-existing national states' (Haas 1958: 16). The definition of Taggart has been refined over the years. In a much cited article on party politics and Euroscepticism Taggart and Szczerbiak (2002) differentiate soft and hard Euroscepticism. The former refers to principled opposition towards European unification. The latter refers to concerns on specific European policy areas, without opposing the process of European unification per se (Taggart and Szczerbiak, 2008). Since the PVV opposes EU-membership of the Netherlands and European unification in a broader sense the party can be grouped among hard Eurosceptics (PVV manifesto, 2010; 2012).

Numerous alternative definitions of Euroscepticism have been proposed. Some scholars have criticised the binary conceptualisation of Taggart and Szczerbiak for being too loose (Mudde, 2012). These critics argue that reality is more complicated than Taggart and Szczerbiak propose. One point of critique is that this definition fails to account for party-level gradations in Eurosceptic attitudes (Flood, 2009). Kopecky and Mudde (2002) argue that the definition of Taggart and Szczerbiak overlooks the ideological dimension of party positioning on European unification. These authors present a two-dimensional typology that differentiates EU support and support for European unification. Kopecky and Mudde distinguish specific and diffuse support. The former relates to support for the practise of European unification. The latter means support for the principles of European unification. This distinction derives from Easton’s (1965) model on distinctive forms of political support. This model is widely used in studies regarding Euroscepticism (Krouwel and Abts, 2007). The typology of Kopecky and Mudde identifies four stances on party positioning. Euroenthusiasts support the underlying principles of unification. Moreover, these parties advocate a strong institutionalisation of these principles by the EU. In this view, Eurosceptic parties support the principles of unification. However, these parties are pessimistic concerning the capability of the EU to embody these principles. Eurorejects disagree with the principles of European unification and the EU in general, whereas Europragmatists do not have a clear ideological stance on European unification. This latter category of parties tends to support the EU as long as the European project is beneficial to their national economy.

Flood and Usherwood (2007) have directly criticised the categorisation of Kopecky and Mudde as oversimplifying. Flood and Usherwood identify a six-point attitude scale ranging from EU-maximalist to EU-reject. EU-maximalist parties favour fast integration of EU member-states. EU-reformists prefer advancing integration but stress the need to address deficiencies

8 of the EU. A widespread example of these deficiencies is the perceived lack of electoral influence on European policy making. Furthermore, EU-gradualists advocate slow advance of European unification, whereas minimalists propose to maintain the status quo. EU-revisionists want to return to a prior situation such as European co-operation before the Treaty of Maastricht. Lastly, Euro-rejectionists fundamentally oppose supranational policy and European unification.

There are similar debates about Euroscepticism at the public level. In line with work on party positioning studies on public attitudes usually distinguish categories of Euroscepticism, even though it has been argued that categorisations add to a semantic confusion about the concept (Fitzgibbon et al., 2017; Leruth and Lord, 2015). Krouwel and Abts (2007) propose a two-dimensional approach to public Euroscepticism. This approach focuses on both the degree and targets of public dissatisfaction towards European unification. The targets of public dissatisfaction are divided in scepticism towards European political actors and authorities, the European regime including its institutions and the existence of a political European community. The most positive degree of attitudes citizens have towards European unification is Euroconfidence. Euroconfident voters trust EU politicians and believe in the principles of European unification. The authors distinguish Euroscepticism, Eurodistrust, Eurocynicism and Euroalienation as intensifying degrees of negative attitudes on the EU.

In sum, a variety of conceptualisations on Euroscepticism at both the level of parties and publics has been proposed. However, the classic definition of Taggart still prevails in scholarly literature (Fitzgibbon et al., 2017).

2.2 From ‘permissive consensus’ to ‘constraining dissensus’

Euroscepticism is traditionally a rare phenomenon. The post-Second World War modernisation process was characterised by widespread positive attitudes towards the European project at both the mass and elite level (Taylor, 2008). Therefore, scholars have argued that this period was characterised by a permissive consensus on European unification (Down and Wilson, 2008). The general perception among European elites was that economic integration of European markets would increase interdependence between former enemies and therefore decrease the possibility of war. European electorates accepted this argument silently (Usherwood and Startin, 2013). The insignificant opposition towards the European

9 project failed to have a substantial impact. Efforts of Eurosceptic actors such as the Danish People’s movement (1972) to change the direction of the debate on European unification were temporary and unsuccesful (Usherwood and Startin, 2013).

The Single European Act of 1986 and subsequent plans for a European internal market led to increased discussion among European elites about the feasibility of European authority (Hooghe and Marks, 1997; Usherwood, 2007). However, the elite consensus on economic integration remained strong. Governments of countries that were traditionally critical regarding further European unification, such as the UK and Greece, ratified the Single European Act. There was little political opposition towards the Single European Act. Even the renowned Eurosceptic French Front National did not vote against this effort to strengthen the bonds within the European Economic Community. Moreover, the populations concerned did not actively oppose the ratification of the Single European Act (Usherwood and Startin, 2013). Margaret Thatcher was the first European leader to challenge the direction of the European project. In her Bruges speech (1988) Thatcher critizised efforts of Jacques Delors, President of the European Commission, to increase political and economic bonds between the member-states of the European Economic Community (Thatcher, 1988). Thatcher’s Eurosceptic speech paved the way for a more generally accepted Eurosceptic discourse among European elites and electorates (Flood, 2002).

Scholars have argued that the era of a permissive consensus faded after the Maastricht treaty of 1992, in which the euro was created and the foundations of the EU were agreed upon. Instead, a time of so called constraining dissensus began in which European unification became highly debated and politicised (Down and Wilson, 2008). The Eurocrisis has expanded negative attitudes regarding European unification among both elites and voters (Braun and Tausendpfund, 2014). Particularly far-right Eurosceptic populist parties such as the Austrian Freedom Party, Swiss People’s Party, Dutch Party for Freedom and Front National have become successful in mobilising electoral support against European integration (Golder, 2016). In short, Eurosceptic discourse is nowadays widespread among European elites and electorates.

Since Euroscepticism has become a persistent feature of contemporary European politics, research regarding this concept has become an established part of EU studies (Usherwood and Startin, 2013). In the words of Mudde (2011: 3) a ‘true cottage industry of Euroscepticism studies has emerged’. Previous studies have focused primarily on two

10 subfields: explanatory factors and understanding the intensity and nature of Euroscepticism (Flood, 2009; Mudde, 2011). Evidence on the causes of the strong decline of a permissive consensus in the post-Maastricht era is conflicting. Some scholars have argued that subjective indicators at the level of individuals account for increased Euroscepticism (Sørensen, 2008). Research finds strong support for the utilitarian argument, meaning that welfare gains have a positive relationship with pro-integrational attitudes (Gabel, 1998). If the EU fails to align with the economic expectations among publics, European voters will become more sceptical towards European unification. Another line of reasoning explains Eurosceptic attitudes through a revival of national identity. The argument goes that nationalist Europeans perceive increasing authority of the EU as threat to the identities of European countries. These studies stress that the transfer of power towards European institutions decline the national sovereignty of European states. Therefore, nationalist Europeans in particular oppose the supranational character of European integration (Sørensen, 2008). Studies consistently find that feelings of national identity lead to low levels of support on European integration (Carey, 2002; McLaren, 2007). These studies are in line with the buffer hypothesis, meaning that voters with a weak European identity tend to be more discontent with EU politics (Wessels, 2007). Evidence from a comparative perspective, however, shows the emergence of a European identity since the 1970s. Bruter (2005) shows that political identities reinforce one another in France, Britain and the Netherlands. These findings counter the assumption that identities are exclusive. Recent research has focused on the effect of political ideologies and the sphere of influence between publics and parties (Hooghe, 2007; Lubbers and Scheepers, 2010; Steenbergen et al., 2007). There is debate on the impact of party strategy and ideology in explaining Eurosceptic attitudes (Mudde, 2011). The Sussex school perceives Eurosceptic policy preferences at the party level as part of a political strategy in order to maximise electoral gains. The school of North Carolina, on the other hand, states that party attitudes on European unification are programmatic and ideological (Mudde, 2011).

In sum, scholars are divided in explaining the post-Maastricht surge in Euroscepticism among European electorates and parties. There is, however, broad consensus that Euroscepticism is not a temporary phenomenon (Fitzgibbon et al., 2017).

11 2.3 Left-wing versus right-wing Euroscepticism

The left-right distinction in politics is perceived as the core dimension structuring political conflict in Western-Europe (Van der Eijk et al., 2005). The political left-right terminology, however, begs for clarification as it has increasingly been used to describe a variety of policy areas and political issues (Van Elsas, 2017). The left-right dimension originally refers to social-economic issues in the context of the state-market debate (Downs, 1957). In current studies, however, issues such as immigration, the environment and law and order are also studied within the left-right framework (Van der Brug and Van Spanje, 2009). In the context of attitudes towards European unification cultural and economic interpretations of the left-right dimension are most salient (De Vries et al., 2013). In this regard, there is consensus that right-wing parties advocate monoculturalist and pro-market policies. Left-wing parties on the other hand advocate multiculturalism and pro-state policies (Van der Brug and Van Spanje, 2009).

Eurosceptic attitudes are not limited to one side of the political spectrum (Taggart and Szczerbiak, 2002). In Denmark’s referendum on the Maastricht-treaty the far-right Progress Party sided with the left Socialist People’s Party (Siune, 1993). Moreover, in France the far-right Front National and a coalition of left-wing actors including communists and socialists mutually campaigned against the EU constitution of 2005 (Hobolt and Brouald, 2011). In the Netherlands the far-left Socialist Party and the far-right PVV co-operated against the Association Agreement between the EU and Ukraine in 2016 (Website NRC). These examples are in line with the horseshoe model. This model theorises that parties at the far end of the political spectrum tend to be Eurosceptic. Empirical research has showed that far-left or right placement strongly correlates with party-level Euroscepticism (Arnolds et al., 2012). In contrast, parties in the middle of the political spectrum generally have strong attitudes in favour of the EU and European unification (Koopmans and Statham, 2009; van Elsas, 2017).

Scholars agree that far-right and far-left parties tend to be more Eurosceptic than mainstream parties (Van Elsas, 2017). In addition, recent research has showed that the intensity and nature of far-right and far-left Euroscepticism varies (Hoeglinger, 2016). Therefore, it has been argued that the horseshoe model needs refinement (Van Elsas, 2017). Van Elsas, in line with Taggart (1998), identifies two dimensions of Euroscepticism: opposing European unification and dissatisfaction with the functioning of the EU (Van Elsas, 2017). Van

12 Elsas argues that far-right voters more firmly oppose European unification than far-left voters, who are in general more dissatisfied with the functioning of the EU. These attitudes can be explained through far-right monoculturalism and nationalism. In contrast, far-left voters tend to have a universal and multicultural worldview. Van Elsas states that cultural issues such as migration are particularly important to far-right voters. On the other hand, economic issues such as welfare distribution characterise far-left Euroscepticism (Van Elsas, et al., 2016; Van Elsas, 2017). The two-dimensional political and economic approach to the nature of Euroscepticism is widespread in the literature (Abts et al., 2009; Lubbers and Scheepers, 2007). The argument that left-wing Euroscepticism essentially is a critique on the neoliberal character of the EU whereas right-wing Euroscepticism is driven by sovereignty is in line with previous research (De Vries and Edwards, 2009). Moreover, the argument of Van Elsas is in line with scholars who argue that the nature of Euroscepticism among far-right voters is based on anti-outgroup and nationalist attitudes (Abts et al., 2009).

The PVV can be categorised as a right-wing Eurosceptic party. This categorisation, however, is not unquestionable. Right-wing stances of the PVV are for instance strict laws on immigration and protecting Dutch culture against growing Islamic influence (Ivarsflaten, 2008). The economically left-wing focus of the PVV regarding welfare issues such as state supplied healthcare and generous benefits for the unemployed is not in line with the far-right image of the party. However, these policy stances are rooted in welfare chauvinism. Outsiders such as asylum seekers are excluded from the social system and therefore only advocate a strong welfare state for Dutch natives (Schumacher and Van Kersbergen, 2016). This exclusionary focus gives these policy positions a right-wing connotation.

In conclusion, the horseshoe model theorises that Euroscepticism is likely to occur at the far-end of the political spectrum. Far-right Eurosceptic parties combine opposition towards European unification with an emphasis on cultural issues such as migration and national sovereignty. The PVV can be classified as a far-right Eurosceptical party, despite left-wing welfare stances.

13 3. Do politicians follow voters?

First, this section will address the contradicting arguments of elite cueing and issue voting. Next, empirical evidence of these approaches regarding voter attitudes on European unification will be examined. The third part of this section presents psychological mechanisms of cue-taking. Lastly, issues and individual features most susceptible to cueing effects are demonstrated.

3.1 Party identification and cue taking

Democratic party systems in Western-Europe have been shaped by issues. Conservative, socialist and liberal parties were founded on issues such as the monarchy, the influence of unions and the protection of individual rights (Borre, 2001). Therefore, political scientists have traditionally argued that voters elect politicians based on issues (Pennock, 1979). Issue voting or policy voting models have a positive image of voter capacities. Research deriving from this perspective finds that the consideration of policy arguments prior an election determines voting. Voters form individual policy preferences and elect politicians that align to their wishes. Therefore, in order to get elected politicians adapt to the policy stances of voters (Frohlich et al., 1978). Borre (2001) distinguishes several issue voting models. Linear position models assume that voters compare their policy stances with parties and simply choose the party that aligns most to their policy preferences. The more voters agree with the policy stances of a party, the stronger the possibility of voting for that party. Spatial models build upon this assumption and place voters on a spatial field with policy stances of the politicians involved. These models firmly underline that voters under any circumstances elect the party most closely resembling their own position on issues that individuals perceive as important. Salience models, on the other hand, theorise that voters elect parties based on issue salience in the media and public discourse. If the salience of an issue increases, voters will favour those parties that provide the most appealing stance on that specific issue.

Research as early as in the 1940s has challenged the argument of issue voting (Borre, 2001). The People’s Choice study (1944) showed that party identification was the determining factor in the American presidential elections of 1940. More precisely, this study found that party identification strongly influenced policy positions of individuals. These findings were

14 underlined by the American Voter study of 1960, a research primarily focused on voting behaviour in the presidential elections of 1956. Party identification is nowadays broadly perceived as a key determinant of mass political behaviour in democracies (Green et al., 2002). Research finds that party identification influences political involvement, perception bias and electoral choices (Miller and Shanks, 1996). Research also finds that party identification functions as an informational shortcut for voters (Shively, 1979). More recently it has been shown that voters tend to align their policy preferences to party positions after voting. Scholars refer to this effect as elite or party cueing (Brader and Tucker, 2010; Hooghe and Marks, 2005; Steenbergen et al., 2007). The argument of elite cueing is built on the assumption that voters have substantial cognitive limitations, such as lacking knowledge about the issues at stake (Chong, 2000). Therefore, it is perceived as rational for voters to take elite cues through a process of persuasion and information (Lupia and McCubbins, 1998).

In scholarly literature elites are generally perceived as a heterogeneous group of organisations and individuals (Gilens and Murakawa, 2002). Depending on the perspective of research elites can among others refer to journalists, interest groups, policy experts, religious leaders or political officials. This study focuses on the effects of parties on mass public attitudes. Therefore, elites will be perceived as parties and aligned politicians. Furthermore, several notes about cue taking must be made in order to clarify the concept. Scholars do not argue that policy stances are completely determined by party preference. Nor is it assumed that partisanship influences all voters identifying with a party (Brader and Tucker, 2010). It must also be noted that non-partisan voters potentially take cues from parties. Information articulated by parties through for example the media can influence the process of opinion formation (Lupia and McCubbins, 1998). Finally, individuals in a person’s private environment can also provide cues provided that their opinion is perceived as credible and trustworthy by the cue taker (Gilens and Murawaka, 2002). In sum, cue taking is not limited to the relationship between parties and voters.

An extensive body of research concerning elite cueing effects has focused on the level of countries (Huber et al., 2005). Researchers vary in their expectations of cueing effects at the crossnational level. Brader and Tucker (2010) identify a universalist and exceptionalist perspective. From the universalist perspective scholars argue that the influence of party identification between countries is similar since people have an intrinsic motivation to identify with groups (Brader and Tucker, 2010). Exceptionalist scholars reject this view and state that

15 party identification is foremost present in the context of American politics. The increase of opinion research has expanded the field of partisanship studies beside the context of the United States, finding evidence for the universalist argument. Recent research shows that party cues influence policy preferences in the UK, Poland, Hungary and Canada (Brader and Tucker, 2010; Merolla et al., 2008). The effect of elite cueing has even been found in pseudo-democratic Russia (Brader and Tucker, 2009).

Based on experimental data, panel surveys and case studies Lenz (2012) shows that voters align to the policy stances of the politician they elect. Cueing effects are shown in a variety of different elections and countries including the Netherlands, the US, Canada and the UK. Lenz finds that past performances and character traits of politicians determine voting behaviour. Voters judge politicians on previous political achievements and personality traits such as trustworthiness and competency, but also on more superficial traits such as appearance and likeability. The more voters perceive a politician as likeable and capable, the more they tend to follow his or her stances on policy issues. These results are in line with previous research and underline that voting based on policy stances is rare in contemporary democracies (Fowler and Margolis, 2011; Riker, 1982). Lenz’ findings strengthen the widespread assumption that voters use leadership evaluations in order to decide which politician to vote for (Garzia, 2011). The personality of a political leader is perceived as a predicting factor regarding behaviour in office. Moreover, evaluations of leadership are less time-consuming than actively following political news or studying election programs. (Bittner, 2008). In fact, Lenz’ study finds that much of the public is not even aware of the policy positions parties and candidates take. In addition to previous work (Fowler and Margolis, 2011), Lenz concludes that as a consequence of ignorance voters often vote for politicians they disagree with without realising it. For example, his study shows that merely 60 percent of American voters correctly knew that the Democrat Obama favoured a large government in comparison to Republican candidate McCain in the 2008 presidential elections, even though this was a core issue in Obama’s campaign and underlined a fundamental and long-standing ideological conflict between these parties on the role of government.

In conclusion, scholars stress the methodological problem of observational equivalence regarding attitude changes. Research has demonstrated a correlation between attitudes and voting, but the causal direction of this relationship remains widely debated. Models of elite cueing and issue voting identify an opposite causal arrow of the linkage

16 between policy stances of mass publics and elites. Although initial research in this field derived from the perspective of issue voting, recent studies have increasingly found sound empirical evidence for elite cueing effects.

3.2 Euroscepticism: who follows whom?

The debate on elite-mass attitudinal linkages continues in the context of European politics (De Vries and Edwards, 2009). The central question in this field is whether attitudes of European voters concerning the process of European unification are shaped by political elites or vice versa. Research has showed mixed evidence on the causal direction of elite-mass attitudes in the context of European politics (Steenbergen et al., 2007). Steenbergen et al. distinguish a top-down and bottom-up perspective. This distinction is in line with other research, although the terminology might vary – e.g. electoral connection instead of bottom-up (Carrubba, 2001). Studies have predominantly found top-down evidence. This strand of research finds that attitudes of European publics regarding European unification are cued by political elites (Wessels, 1995; Steenbergen and Jones, 2002; Ray, 2003). For example, cross-country analysis by Anderson (1998) demonstrates the strong influence of political elites on public support towards European unification in most notably France and Denmark. These findings are underlined by Van der Eijk and Franklin (1996), who based their research on the European Election Study.

Contrary to top-down studies bottom-up research finds that electoral policy preferences regarding European politics determine the behaviour of political elites (Arnolds et al., 2012). The work of Carrubba (2001) is considered as the most prominent bottom-up study in the field (Steenbergen et al., 2007). Carrubba theorises that parties follow a rational strategy of vote-maximisation. It is assumed that European publics have systematic preferences concerning European unification. Parties have significant incentives to adopt these policy preferences. Therefore, rational parties do not take unpopular policy stances regarding European unification. The study of Carrubba analyses pre-Maastricht Europe. There is consensus that European unification during this time was substantially less salient than nowadays (Brack and Startin, 2015; Down and Wilson, 2008). However, deriving from data on party manifestos Carrubba concludes that in this era national parties become more in favour of European unification when their voters have similar attitudes. These pre-Maastricht

17 findings counter the widespread assumption of a silent consensus among voters during the permissive consensus of this era (Fitzgibbon et al., 2017).

In spite of increased scholarly attention the main strands of research remain focused on either top-down or bottom-up linkages (Ray, 2003). Dual-process models, however, find a complex interplay between party elites and voters (Steenbergen et al., 2007; Wessels, 1995). These models state that elite-mass linkages have a reciprocal character, meaning that European voters both shape and adopt policy stances of party elites simultaneously. For example, Sanders and Toka (2002) find that publics are influenced by elites in opinion forming on European politics. However, elites also respond to the stances of the electorates they represent. Therefore, based on mass and elite surveys in a variety of sixteen EU member-states both publics and political elites are found to be cueing one another.

In conclusion, empirical evidence shows support for both the bottom-up and top-down perspective. However, most studies point to a top-down relationship between European elites and publics on attitudes regarding European unification. Dual-process models advocate a mixed-perspective as it is argued that party elites and electorates influence one another’s attitudes on European politics.

3.3 The psychology of cue taking

Studies have identified numerous cognitive heuristics in order to simplify individual decision making regarding policy issues. Widespread examples of these mental shortcuts are likeability, availability and representativeness (see Kahneman and Tversky, 1972; Sniderman et al., 1991). Although it is assumed that no single heuristic determines policy stances of voters, recent studies have identified elite cues as main shortcut on political decision-making (Gilens and Murawaka, 2002). It has been argued that taking elite cues is logical from a psychological perspective as studies consistently show that voters in general are ignorant on policy matters (Gilens and Murakawa, 2002). Therefore, voters seek assistance of policy-experts that are perceived to have sound judgement and similar political goals (Downs, 1957).

Previous research has identified several mechanisms in order to explain why voters use elite cues as informational shortcuts. Cognitive dissonance theory provides insights on the psychological processes that underlie political behaviour. This theory is based on the assumption that individuals strive towards consistency between their opinions and actions in

18 order to feel secure and have a positive self-image (Festinger, 1957). Numerous studies in the context of attitude changes and opinion research have found evidence for the claim that behaviour shapes preferences (Dinas, 2013, McCann, 1997; Mullainathan and Washington; 2007).

There have been efforts to combine conflicting empirical findings on persuasion processes in one model (Petty and Cacioppo 1981, 1986). This elaboration likelihood model identifies a central and peripheral pathway towards persuasion. Petty and Cacioppo argue that individuals following the central pathway consider substantive content such as the logic and strength of an argument and available evidence. They argue that individuals who are strongly involved with an issue tend to follow the central pathway. This pathway requires more cognitive effort than peripheral processing (Petty and Cacioppo, 1986). Individuals following the peripheral pathway consider elements beside the actual content of information. From this perspective affective associations such as attractiveness and reliability of the message source are crucial in an individual’s decision to adopt information. Therefore, the peripheral pathway is most relevant in explaining elite cueing (Petty and Cacioppo, 1981). Moreover, research using similar models such as Chaiken’s (1980) heuristic-systematic persuasion model underlines the finding that attitude changes through the central pathway are relatively stable compared to the peripheral pathway.

Scholars have found that cueing effects vary between issues and individuals. Carmines and Stimson (1980) distinguish easy and hard issues. The former refers to emotional issues that are familiar to voters. The latter refers to technical issues that demand specialised judgment. There is broad empirical evidence that voters primarily take cues on hard issues (Ratneshwar and Chaiken, 1991). Therefore, persuasion models generally assume that voters are most likely to follow elite cues on issues that are complex. Moreover, elite cues tend to be particularly strong on issues in which individuals are not interested. Lack of political interest results in a low level of individual motivation and ability to gather information (Gilens and Murawaka, 2002). In the case of individual involvement people prioritise the strength of arguments (Petty and Cacioppo, 1986). Social-psychological experiments have shown that people become more resistant towards attitude persuasion when becoming more interested in an issue (Hample, 1985; Miller, 1965; Sherif and Hovland, 1961).

In conclusion, substantial evidence has been found for cognitive dissonance theory in explaining political behaviour. Moreover, authors have distinguished several pathways

19 towards persuasion. Issues with a high demand of specific knowledge are most likely to have cueing effects on individuals. In particular those with low levels of political interest are found to be susceptible to elite cueing.

4. Hypotheses

As previously discussed, the argument of elite cueing theorises that voters align to policy stances of parties after electing a politician (Brader and Tucker, 2010; Lenz, 2012; Lupia and McCubbins, 1998). Processes of elite cueing are disputed by issue voting models, in which politicians follow policy stances of voters in order to get elected (Borre, 2001; Frohlich et al., 1978; Pennock, 1979). In the context of European politics some studies have found that politicians adapt to stances of electorates (Arnolds et al., 2012; Carrubba, 2001). However, most evidence has pointed to elite cueing effects. Numerous studies have shown that voters align to the Eurosceptic attitudes of the political elites they elect (Anderson, 1998; Hooghe and Marks, 2005; Ray, 2003; Steenbergen and Jones, 2002; Steenbergen et al., 2007; Van der Eijk and Franklin, 1996; Wessels, 1995).

The policy position of the PVV on further European unification is straightforward. European unification is perceived as a threat for Dutch culture and national independence (PVV Manifesto 2010; 2012). Therefore, the party argues that the Netherlands should leave the EU. These hard-Eurosceptic policy stances (Taggart and Szczerbiak, 2008) theoretically align with models showing that far-right parties tend to oppose European unification based on cultural and political arguments (De Vries and Edwards, 2009; Van Elsas, 2016; 2017). The argument of elite cueing in combination with top-down evidence in the context of European politics and the far-right characteristics of PVV Euroscepticism lead to the following hypothesis:

H1: voters become more opposed towards European unification after voting for the PVV

Based on the discussed studies on the relationship between characteristics of politicians and voting behaviour it is expected that a high degree of sympathy towards a political leader leads to stronger cueing effects (Fowler and Margolis, 2011; Lenz, 2012; Riker, 1982). Therefore, the degree of sympathy towards Wilders is expected to be an interaction effect between

first-20 time PVV voting and attitude changes regarding European unification. Voters tend to be more likely to follow politicians they perceive as capable and trustworthy. A high sympathy score reflects the likeability of a politician and positive judgment of his or her character traits and capacity to lead. This leads to the following hypothesis:

H2: the effect of first-time PVV voting on increasing anti-European unification attitudes is moderated by sympathy for Wilders

Research has shown that features of individuals to a large extent determine the degree to which voters are susceptible to cueing. As discussed, studies find that cueing effects tend to be stronger on voters who have a low degree of political interest (Gilens and Murawaka, 2002; Hample, 1985; Miller, 1965; Petty and Cacciopo, 1986; Sherif and Hovland, 1961). Therefore, political interest is expected to be an interaction effect between first-time PVV voting and attitude changes regarding European unification. A low degree of political interest results in a low level of motivation to extent one’s political knowledge and involvement. As a consequence, politically uninterested voters are more likely to follow the stances of politicians. Politicians are perceived as experts on particularly hard-issues that require specialised knowledge. These findings are summarized by the following hypothesis:

H3: the effect of first-time PVV voting on increasing anti-European unification attitudes is moderated by political interest

Additionally, a schematic overview of the proposed hypotheses is presented in figure 1.

Figure 1. Schematic overview of the hypotheses

Sympathy for PVV leader Wilders

↓

First-time PVV vote in 2010 or 2012  Post-election increase in anti-European unification attitudes ↑

21 5. Methodology

This section will outline the methodology of this research. First, the data will be examined. Next, the case selection is elaborated. The section continues with operationalisations in order to test the hypotheses. Subsequently, the regression model is clarified. The section concludes with descriptive statistics of the variables.

5.1 Data selection

The data in this research require a range of features in order to answer the research question. First, it must be possible to identify a sample of first-time PVV voters for at least one election. The N of this sample must be sufficient in order to draw conclusions about the Dutch population. Furthermore, it must be possible to measure far-right Eurosceptic attitudes. These requirements are met by LISS panel data. The LISS panel consists of a probability sample including 4500 Dutch households. These households are represented by roughly 7000 individuals. The LISS panel is supervised by CentERdata, a research institute aligned to Tilburg University. These representative households are selected through the Dutch population register. The individuals representing households get compensated financially to participate in surveys. Every survey addresses different topics varying from politics and values to income, housing and personality. The panel answers a yearly longitudinal survey in December. Based on the politics and value study it is possible to examine changes of attitudes regarding European unification over time. There are currently nine longitudinal survey waves of the politics and values panel. These waves range from 2007 to 2016. In conclusion, the large number of respondents in combination with the possibility to longitudinally examine voter attitudes on European unification makes these LISS-data very suitable for this research. Table 1 shows the panel structure of this research.

22 Table 1. Panel structure

Construct Measured LISS panel module Vote choice 2006 December 2007 Politics and Values Attitudes 2007 December 2007 Politics and values Attitudes 2008 December 2008 Politics and values Attitudes 2009 December 2009 Politics and values Vote choice 2010 December 2010 Politics and values Attitudes 2010 December 2010 Politics and values Attitudes 2011 December 2011 Politics and values Vote choice 2012 December 2012 Politics and values Attitudes 2012 December 2012 Politics and values

5.2 Case selection

The PVV is studied as case for several reasons. The PVV is an established far-right party in the Netherlands with an outspoken Eurosceptic stance. Moreover, the PVV has received much support of the Dutch electorate since the party first participated in the 2006 parliamentary elections. However, electoral support for the PVV has been inconsistent. The PVV achieved its most significant electoral success in the 2010 parliamentary elections. Therefore, these elections will provide the best sample of citizens who voted PVV for the first time. The party lost substantially in the 2012 elections. However, given the large N of the panel it is possible to examine a sufficient sample of first-time PVV voters for the 2012 parliamentary elections.

The analysis includes the years 2007 until 2013. Both samples, first-time PVV voters in 2010 and 2012, will be analysed over a pre- and post-stimulus period. The attitudes of the 2010 sample after the 2010 parliamentary elections will be compared with 2007, 2008 and 2009. The attitudes of the 2012 sample after the 2012 parliamentary elections will be compared with 2010 and 2011. The starting point of the pre-stimulus period for the 2012 sample is after the parliamentary elections of 2010 in order to avoid overlap with this influential event. Some remarks must be made in order to clarify why the parliamentary elections of 2006 and 2017 are not examined. First, there is no longitudinal LISS data available prior 2007. Therefore, it is not possible to longitudinally examine the years before individuals

23 decided to vote PVV in the parliamentary elections of 2006. Furthermore, the PVV won substantially in the most recent parliamentary elections of March 2017. However, relevant post-election data is not available at the time of analysis.

5.3 Measures

The answers of respondents regarding questions in the LISS data were used to operationalise the hypotheses of this research. The problem of multicollinearity was not salient since all concepts could be measured directly from the LISS data. Respondents were directly asked which party they voted for in the parliamentary elections through the question ‘’for which party did you vote in the parliamentary elections of 22 November 2006?’’. This question was also answered regarding the parliamentary elections of 9 June 2010 and 12 September 2012. First-time PVV voters and all other voters in the parliamentary elections of 2010 and 2012 were divided in a dummy variable (0=all other voters, 1=first-time PVV voters). In accordance with the discussed literature on nature and intensity of far-right Euroscepticism, Eurosceptic attitudes were examined through opinion changes on European unification. Increasing opposition towards European unification was measured through the variable ‘’where would you place yourself on a scale from 1 to 5, where 1 means that European unification should go further and 5 means that it has already gone too far’’. Sympathy towards Geert Wilders was measured on a 0-10 scale ranging from 0=very unsympathetic to 10=very sympathetic. Furthermore, political interest was measured through the variable ‘’are you very interested in political topics, fairly interested or not interested?’’ (1=very interested, 2=fairly interested, 3=not interested). These Wilders sympathy and political interest scales test whether the attitude changes of first-time PVV voters with a high leader sympathy or low political interest score are stronger than those with low leader sympathy and high political interest scores.

5.4 Regression model

The goal of this research is to examine the causal effect between first-time PVV voting and attitude change regarding European unification. Studies have shown difficulties in identifying causal effects with observational data such as surveys (Winship and Morgan, 1999; Wooldridge, 2010). The effect between first-time PVV voting and changes in Eurosceptic

24 attitudes is measured through a fixed-effects regression model. This model controls for unobserved stable confounders (Allison, 2009). Fixed-effects are constant over a time period and across individuals. The model ensures that time-invariant differences between individuals are subsumed in the fixed-effects (Angrist and Pischke 2009; Allison, 2009). Random and mixed-effects regression are less suitable for this research as these methods allow for individual effects (Allison, 2009). Contrary to the random variation that goes with random-effects regression, this study clearly distinguishes two groups: first-time PVV voters in a given parliamentary election and all other voters. Furthermore, fixed-effects regression has proved to be a widespread method to estimate causal effects in academic research. Scholars have used similar models in a variety of studies and contexts to identify causal relationships (Brown et al., 2008; Freeman, 1984). The availability of pre-test and post-test data for a control and treatment group are required in order to use fixed-effects regression analysis (Winship and Morgan, 1999). In this study the pre-test for the 2010 sample of first-time PVV voters concerns the years 2007, 2008 and 2009. The post-test is done in December 2010. The pre-test for the 2012 sample of first-time PVV voters concerns the years 2010 and 2011. The post-test is done in December 2012.

In sum, the longitudinal character of the LISS-data makes fixed-effects regression an excellent method to examine patterns of individual political behaviour and attitude changes. Lastly, robust standard errors are demonstrated to account for heteroscedasticity (Angrist and Pischke 2009; Arellano 1987;Hayes and Cai, 2007).

5.5 Descriptive statistics

Descriptive statistics for each wave are presented in table 2. Over a period of six years a mean increase of 0,28 on a five-point scale is noticed for the dependent variable European unification. Therefore, respondents in general became more Eurosceptic over this period. It should be noticed, however, that these mean changes fluctuate over the years. Furthermore, Wilders receives an extremely low mean score on personal sympathy. This score ranges from 2,71 to 3,61 across all respondents over the examined period. The Wilders sympathy score has a high standard deviation, meaning that there are substantial differences of appreciation for Wilders between respondents. Lastly, according to the mean score on political interest the average LISS respondent is fairly interested in politics.

25 Table 2. Descriptive statistics

Wave 1 2007/2008 Variable M SD Min Max N % European Unification 3,38 1,11 1 5 4725

Sympathy for Wilders 2,78 2,69 0 10 4769 Political Interest 2,02 0,56 1 3 5134 Vote in 2006 election

PVV 4979 3 Other voters 155 97 Wave 2 2008/2009 Variable M SD Min Max N % European Unification 3,30 1,08 1 5 5482

Sympathy for Wilders 2,76 2,64 0 10 5528 Political Interest 2,07 0,58 1 3 5134 Vote in 2006 election

PVV 158 2,6 Other voters 5878 97,4 Wave 3 2009/2010 Variable M SD Min Max N % European Unification 3,30 1,09 1 5 4309

Sympathy for Wilders 2,90 2,83 0 10 4416 Political Interest 2,09 0,59 1 3 5134 Vote in 2006 election

PVV 207 3,4 Other voters 5830 96,6 Wave 4 2010/2011 Variable M SD Min Max N % European Unification 3,62 1,12 1 5 3611

Sympathy for Wilders 3,61 2,88 0 10 3750 Political Interest 2,06 0,60 1 3 5134 Vote in 2010 election

PVV 403 6,7 Other voters 5634 93,3

26 Wave 5 2011/2012 Variable M SD Min Max N % European Unification 3,57 1,21 1 5 3412

Sympathy for Wilders 3,16 2,73 0 10 3547 Political Interest 2,08 0,60 1 3 5134 Vote in 2010 election

PVV 310 5,1 Other voters 5727 94,9 Wave 6 2012/2013 Variable M SD Min Max N % European Unification 3,66 1,15 1 5 3213

Sympathy for Wilders 2,71 2,6 0 10 3358 Political Interest 2,09 0,60 1 3 5134 Vote in 2012 election

PVV 5841 96,8 Other voters 196 3,2

6. Results

This section starts with the results of paired t-tests for both first-time PVV voters in 2010 and 2012. The t-tests demonstrate whether these groups change their attitudes on European unification after the parliamentary elections. Next, the fixed-effect regression results are shown in order to examine whether these attitude changes are the result of switching to the PVV.

6.1 Paired t-tests

Paired t-tests provide a first tool in order to examine differences between variable means over time. The paired t-tests in this study demonstrate the means of two samples on attitudes regarding European unification. Figure 2 shows whether individuals who voted PVV for the first time in 2010 (N=230) changed their attitudes on European unification after these elections compared to three time points (2007, 2008 and 2009). Figure 3 shows whether individuals who voted PVV for the first time in 2012 (N=42) changed their attitudes on

27 European unification after these elections compared to two time points (2010 and 2011). The starting point of the 2012 sample are attitudes measured at the end of 2010 in order to avoid overlap with the parliamentary elections of June 2010. The sample of first-time PVV voters in 2012 is small compared to 2010. This was expected since the PVV declined from 24 to 15 seats in the parliamentary elections of 2012. Furthermore, the higher the number of elections individuals participate in after the first appearance of the PVV in 2006, the more likely they are to have already voted for the PVV at least once. It should be noted that every LISS politics and values study was carried out in December. For example, the results of 2007 demonstrate changes between 2007 and 2008. Therefore, the results of 2010 and 2012 are measured after the parliamentary elections of that year.

First-time PVV voters in 2010 clearly become more Eurosceptic after voting for the party. The Eurosceptic attitudes of first-time PVV voters increase with a mere 0,03 points on a 5-point European unification scale between 2007 and 2008. In 2009 a slightly stronger increase is noticed (0,11 points). The difference between 2009 and 2010 is 0,36 points. This is substantial on a 5-point European unification scale. Moreover, the change of means in 2010 is significant compared with 2009.

First-time PVV voters in 2012 also become increasingly more opposed to European unification after voting for the party. Eurosceptic attitudes among this group in 2011 decrease slightly compared to 2010 (-0,09 points). Subsequently, a substantial increase of anti-European unification sentiment is shown after the parliamentary election of 2012 (0,30 points). The difference between the mean of 2012 is, however, not significant compared to 2011.

In conclusion, substantial attitude changes on European unification are noticed after voting for the PVV. The mean change of first-time PVV voters in the year 2010 is relatively large compared to the expected increase based on the mean differences in 2008 and 2009. Therefore, the results of the paired t-tests provide a first suggestion that voting for the PVV increases Eurosceptic attitudes. The strong although not significant mean increase of first-time PVV voters in 2012, following a decrease between 2010 and 2011, underlines this suggestion.

28 Figure 2. Means and 95% confidence intervals of first-time PVV voters in 2010

* Difference between mean and mean of previous year is significant with p<0.05

Figure 3. Means and 95% confidence intervals of first-time PVV voters in 2012

3,6 3,8 4 4,2 4,4 4,6 4,8 5 2006 2007 2008 2009 2010 2011

European Unification

European Unification

29 6.2 Regression analyses

General developments in the population are excluded in simple t-tests. Therefore, the causal effects between first-time PVV voting on attitude changes regarding European unification are isolated through fixed-effects regression analysis. Table 3 demonstrates the regression results for first-time PVV voters in 2010. Table 4 demonstrates the regression results for first-time PVV voters in 2012. Both tables are split into three models. Models 1 and 4 address the direct effect between switching to the PVV and attitude changes on European unification. Models 2 and 5 test whether a high sympathy score of Wilders results in stronger attitude changes. Lastly, models 3 and 6 examine whether a low political interest score results in stronger attitude changes.

Analysis reveals that first-time PVV voters become significantly more opposed to European unification after voting for the party. The results show that voters switching to the PVV score 0,67 points higher on the 5-point European unification scale compared to all other voters in 2010. The anti-European unification score of voters switching to the PVV is even higher in 2012 (0,78). Furthermore, the results provide evidence for moderating effects of sympathy for Wilders and political interest. Analysis shows a positive and significant effect regarding anti-European unification attitudes for those first-time PVV voters with a higher sympathy score regarding PVV leader Wilders. An increase of sympathy for Wilders results in similar findings for both first-time PVV voters in 2010 (0,10 points) and 2012 (0,11 points). Moreover, it is demonstrated that first-time PVV voters with a low level of political interest score higher on the post-election European unification scale. The results show a significant and positive effect with only minor differences between first-time PVV voters in 2010 (0,08) and 2012 (0,13) to be noted.

In conclusion, fixed-effect regression confirms that Eurosceptic attitude changes are caused by voting for the PVV. These effects are demonstrated to be stronger for those first-time PVV voters with a high sympathy score of Wilders. Moreover, having a low level of political interest shows to increase cueing effects.

30 Table 3. Fixed-effects regression models 2010

Model 1 Model 2 Model 3 First-time PVV vote (2010) 0,67* 0,31* 0,67* (0,07) (0,07) (0,07) Sympathy for Wilders - 0,10* - (0,01) Political interest - - 0,08* (0,03) Intercept 3,63* 3,31* 3,46* (0,02) (0,03) (0,07) N 3575 3469 3575 R² 0,02 0,10 0,03

Fixed-effects regression estimates. Robust standard errors are reported in parentheses to account for heteroscedasticity. *p<0.05

Table 4. Fixed-effects regression models 2012

Model 4 Model 5 Model 6 First-time PVV vote (2012) 0,78* 0,34 0,77* (0,11) (0,13) (0,12) Sympathy for Wilders - 0,11* - (0,01) Political interest - - 0,13* (0,04) Intercept 3,77* 3,45* 3,50* (0,02) (0,03) (0,08) N 3204 2844 3204 R² 0,01 0,12 0,03

Fixed effects regression estimates. Robust standard errors are reported in parentheses to account for heteroscedasticity. *p<0.05

31 6. Conclusion

Euroscepticism has become increasingly mainstream across European elites and electorates. Particularly parties at the far-left or far-right side of the political spectrum have been electorally succesful in spreading a critical message on European unification and the functioning of European institutions. Far-right Eurosceptic parties such as the PVV tend to oppose European unification. Far-left Eurosceptic parties are generally more dissatisfied with the functioning of the EU, without dismissing further European unification per se.

A number of perspectives and empirical studies on the attitudinal relationship between elites and mass publics have been put forward. Scholars have traditionally argued that political elites follow the policy preferences of voters. In the literature this perspective is referred to as issue voting. More recently, scholars have found overwhelming evidence that electorates follow elites instead of vice versa. In the literature this perspective is referred to as elite-cueing. Studies in the context of European politics have shown mixed evidence. Most evidence, however, points in the direction of a top-down causal effect. Studies predominantly find that European publics adjust their policy positions to elite attitudes regarding European unification.

The research question of this thesis was ‘’Does voting for the far-right Eurosceptic PVV causes changes in attitudes on European unification?’’. Therefore, the aim of this study was to test the argument of elite cueing on Eurosceptic attitudes. This argument was tested through an analysis of first-time PVV voters in 2010 and 2012. The general assumption was that in accordance with the party stance first-time PVV voters would become more opposed towards European unification. Moreover, it was examined whether cueing effects were moderated by sympathy for Wilders and political interest. These hypotheses were build upon studies that found particularly strong cueing effects among voters who strongly identified with the politician they elected and those that were uninterested in politics.

Longitudinal LISS data ranging from 2007 to 2013 were used to test the hypotheses. Paired t-tests showed attitude changes on European unification for first-time PVV voters. Voters became more opposed towards European unification after voting for the party. Fixed-effects regression analyses showed that these attitude changes were caused by voting for the PVV. Furthermore, it was found that this cueing effect was stronger for those with a higher sympathy score regarding Wilders. The results also showed that a low level of political interest

32 led to higher scores, and therefore a more negative attitude change, regarding European unification.

In conclusion, voting for the far-right PVV indeed causes changes in attitudes on European unification. After voting for the party, first-time PVV voters align their policy preferences with party positions. As hypothesised, voters switching to the PVV became significantly more opposed towards European unification. Moreover, as expected, sympathy towards Wilders and political interest were found to moderate this cueing effect.

7. Discussion

The results of this study provided evidence for elite cueing effects. In general, scholars should doubt whether voters are followed by politicians on policy issues. Researchers in the field of political behaviour should consider the finding that voting for a party influences voter attitudes, besides the widely accepted assumption that attitudes influence party choice. Research should not be narrowed down to the question whether support for Eurosceptic parties is the result of anti-European sentiment. Instead, the possibility of attitude changes after voting for a Eurosceptic party should get more attention. This study was, however, not a dual-process model. Based on the results it cannot be argued if, and to what extent, parties base their policy positions on public attitudes. Therefore, the possibility of a complex interplay between voter and elite-cues should not be overlooked.

Although this thesis demonstrated that voting for the PVV causes an increase in anti-European unification attitudes, it should be noted that prior voting these individuals already had a negative attitude towards European unification. Therefore, it is more precise to say that these attitudes were strengthened by voting for the PVV. This is particularly the case for first-time PVV voters in 2012, who already had a substantial mean score of 4,4 points on the 5 point European unification scale in 2010 before increasing to an even higher level after voting for the PVV.

How can we explain changes in Eurosceptic attitude after voting for a Eurosceptic party, in this case the PVV? This study does not provide a satisfying answer to this question. Based on the methodology and features of the data it was not possible to examine the psychology of individuals. However, given the discussed psychological processes regarding elite cueing it appears sound to assume that cognitive dissonance plays a major role in