'Framing' the European Union

‘FRAMING’ THE EUROPEAN UNION:

EXPLAINING THE 2005 CONSTITUTIONAL REFERENDA RESULTS

Ece Ozlem Atikcan

Department of Political Science McGill University, Montreal June 2010

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Ph.D. in Political Science

ABSTRACT The 2005 European Union (EU) constitutional referenda results reflected growing Euroscepticism. While the referenda in Spain and Luxembourg approved the European Constitution (TCE), the ones in France and the Netherlands rejected it. Polls show that public opinion in all four countries originally favored the TCE several months before the referenda. Why could this initial positive public opinion not be sustained in the French and Dutch cases? I argue that the stronger a state’s No campaign relative to its Yes campaign – that is, the better the No campaigners linked existing contentious issues to the European Constitution – the greater the increase in the magnitude of the No vote. Based on 96 in-depth interviews with campaigners, media content analyses and public opinion data from all cases, I show that the initial favorable public opinion in the French and Dutch cases fell dramatically due to strong No campaigns because the French and Dutch No campaigners framed the issue effectively. The framing literature argues that politicians encourage voters to think along particular lines, by using frames that emphasize certain features of the subject. Vivid, concrete, image-provoking, emotionally compelling frames that contain negative information are more successful in influencing individuals’ opinions. In the French and Dutch cases, the No frames argued that the TCE would increase immigration, lead to market- friendly reforms, and cause rising unemployment. In contrast, the Yes campaign frames sounded overly technical, presenting the TCE as an institutional step towards a better Europe. Negative, immediate linkages to existing problems won over abstract, non-immediate benefits. Strong No campaigns successfully countered the initial favorable public opinion in these referenda. Nevertheless, I also argue that the temporal sequencing of the referenda could affect the relative strength of the No campaign in second-mover states; all campaigns were therefore not created equal. The later a state held its referendum, the more the previous referenda campaigns could influence campaign dynamics. However, this diffusion effect was not automatic and depended on channels such as shared language/culture, common media sources, and collaborative networks/transnational linkages. Where these channels were present and the campaign of the first-mover state was intense, diffusion amplified the strength of referenda campaigns in second-mover states.

RÉSUMÉ

Les résultats des référendums constitutionnels de l'Union européenne de 2005 ont reflété un euroscepticisme croissant au sein de l’Union. Bien que les référendums en Espagne et au Luxembourg aient approuvé la Constitution européenne (TCE), ceux de la France et des Pays-Bas ont rejeté la proposition de réforme. Les sondages de l’époque montrent que l'opinion publique dans ces quatre pays favorisait le TCE quelques mois avant les référendums. Pourquoi cette opinion publique favorable ne s’est pas maintenue dans les cas français et néerlandais? Je soutiens que plus forte était la campagne du Non par rapport à celle du Oui – c’est à dire plus il existait des liens entre le référendum et des enjeux locaux controversés – plus élevé s’en trouvait le support pour le Non le jour du vote. Cette analyse se base sur une série de 96 entrevues en profondeur menées auprès de militants, des analyses de contenu média et des données d’opinion publique. Dans les cas français et néerlandais, l’opinion favorable du public a considérablement diminué suite à l’usage d’un meilleur cadre cognitif durant la campagne par le camp du Non. La littérature sur le cadrage cognitif suggère que les politiciens encouragent les électeurs à garder en tête un ordre d’idées particulier, grâce à un argumentaire qui met l'accent sur certaines caractéristiques de l’enjeu. Des images vives et concrètes, en plus de messages négatifs sont plus efficaces pour agir sur l’opinion des individus. Dans les cas français et néerlandais, le cadrage cognitif du Non a soutenu que le TCE augmenterait l'immigration, conduirait à des réformes favorables au marché, et provoquerait la montée du chômage. Par contre, le cadrage cognitif de la campagne du Oui s’est montré trop technique en présentant le TCE comme une étape institutionnelle vers une Europe meilleure. Des liens négatifs et immédiats avec les problèmes locaux existants ont donc remporté la bataille contre les avantages abstraits et non immédiats présentés par le camp du Oui. Les fortes campagnes du Non ont neutralisé la première opinion favorable du public dans ces référendums. De plus, je soutiens que l’ordre chronologique des référendums peut influencer la force relative des campagnes nationales. Toutes les campagnes ne sont donc pas créées égales. Cependant, cet effet de diffusion n'est pas automatique et dépend de facteurs tels que la langue et/ou culture communes, des sources médiatiques partagées et des réseaux de collaboration transnationaux. Lorsque ces facteurs sont présents et les campagnes précédentes intenses, la diffusion amplifie la force des campagnes au sein des référendums ultérieurs.

TABLE OF CONTENTS

Acknowledgments vi List of figures and tables ix List of abbreviations xiv Chapter 1: The Puzzle of EU Referenda 1 Understanding Attitudes towards the EU 3 Understanding the EU Referenda Results 5 Understanding the 2005 Constitutional Referenda Results 10 The Comparative Framework: Ruling out Existing Explanations 14 The Argument 19 Dependent Variable: The Magnitude of the No Vote 20 Independent Variable: Relative No Campaign Strength 20 Conditioning Variable: Diffusion 21 Analytical Framework 23 Cases 24 Methodology 30 Interviews and campaign materials 31 Media content analysis 33 Public opinion data 34 Structure of the Dissertation 37 Chapter 2: Analyzing EU Referendum Campaigns 38 Relative No Campaign Strength 38 Campaigns in the Public Opinion Literature 39 Agenda-setting and Priming Theories 40 Framing 42 Framing in the public opinion literature 43

Framing in the social movements literature 50 Strength of Political Campaigns 54 Referendum Campaigns 55 2005 EU Constitutional Referendum Campaigns: The Analytical Framework 63 Diffusion 67 Shared Language/Culture 72 Common Media Channels 73 Collaborative Networks/Transnational Linkages 73 Institutional networks 73 Personal connections 76 Preview of the Case Findings 78 Chapter 3: Spain: The Quiet Case 82 Evolution of the Public Opinion 85 Discontent with Domestic and EU Governance 86 Contention concerning domestic politics 86 Contention concerning the EU 92 Political Party Attitudes towards Europe 96 The Long Campaign 98 Agenda-setting 100 Priming and Framing 101 Content of the campaign frames 101 Relative strength of the campaign frames 106 Credibility of the Speakers 114 Mobilizational Structures 115 Analysis of the Vote 119 Shifts over Time 119 Media Content Analyses 119

Public Opinion Data 122 Diffusion 128 The Quiet Campaign 132 List of Personal Interviews 134 Chapter 4: France: The Momentum Case 135 Evolution of the Public Opinion 137 Discontent with Domestic and EU Governance 138 Contention concerning domestic politics 138 Contention concerning the EU 144 Political Party Attitudes towards Europe 146 The Long Campaign 151 Agenda-setting 153 Priming and Framing 155 Content of the campaign frames 155 Relative strength of the campaign frames 159 Credibility of the Speakers 165 Mobilizational Structures 168 Analysis of the Vote 172 Shifts over Time 173 Media Content Analyses 175 Public Opinion Data 178 Diffusion 186 The Momentum Campaign 190 List of Personal Interviews 191 Chapter 5: The Netherlands: The Turmoil Case 192 Evolution of the Public Opinion 194 Discontent with Domestic and EU Governance 195

iii

Contention concerning domestic politics 196 Contention concerning the EU 203 Political Party Attitudes towards Europe 206 The Long Campaign 209 Agenda-setting 211 Priming and Framing 212 Content of the campaign frames 212 Relative strength of the campaign frames 215 Credibility of the Speakers 227 Mobilizational Structures 229 Analysis of the Vote 232 Shifts over Time 232 Media Content Analyses 233 Public Opinion Data 235 Diffusion 247 The Turmoil Campaign 252 List of Personal Interviews 255 Chapter 6: Luxembourg: The Diffusion Case 256 Evolution of the Public Opinion 259 Discontent with Domestic and EU Governance 259 Contention concerning domestic politics 260 Contention concerning the EU 268 Political Party Attitudes towards Europe 271 The Long Campaign 274 Agenda-setting 275 Priming and Framing 276 Content of the campaign frames 276

Relative strength of the campaign frames 279 Credibility of the Speakers 288 Mobilizational Structures 290 Analysis of the Vote 293 Shifts over Time 293 Media Content Analyses 295 Public Opinion Data 296 Diffusion 303 The Diffusion Campaign 319 List of Personal Interviews 322 Chapter 7: Conclusion 323 Contribution to the EU Literature 330 The Double EU Referenda 332 Broader Theoretical Implications 337 Policy Implications for Europe 339 Bibliography 342 Appendix: Research Ethics Approval 360

ACKNOWLEDGMENTS

It is a great pleasure to express my gratitude to those who made this dissertation possible. I owe my deepest gratitude to my supervisor Dr. Juliet Johnson. Her guidance has been extraordinary throughout this challenging journey. Thanks to her sharp intellectual input, critical feedback on every word of my work, patience in answering my never-ending questions, and constant encouragement, I never felt lost. I can only hope to become as great an academic as her some day.

Dr. Hudson Meadwell, Dr. Brian Rathbun, Dr. Éric Bélanger, and Dr. Stuart Soroka gave me helpful comments and contributed significantly to my understanding of the subject. I am very thankful for their accessibility and support throughout the process. I am particularly indebted to Dr. Rathbun for continuing to guide me even after leaving Montreal. Special thanks go to my colleague Tania Jenkins, who has not only been a very close friend but also kindly accepted to revise my dissertation, and did so with incredible diligence. I do not know how I would have survived this final stretch without her support. I would also like to thank the staff of the Political Science Department for helping me navigate through the bureaucratic channels of McGill with such efficiency.

To conduct my field work, I spent four months in Spain, France, the Netherlands and Luxembourg. Without the necessary funding, this research would have been impossible. I gratefully acknowledge the contribution of the McGill University-Université de Montréal Institute for European Studies, and the McGill Faculty of Arts for their specific research trip awards; and McGill University and the Political Science Department for the Nathan Steinberg Fellowship in Political Science, McGill Majors Scholarship, and McGill Faculty of Arts Award. I would also specifically like to thank Dr. Johnson and Dr. Rathbun for their financial support for my field research.

During my field research, so many people helped me in so many ways. I would like to extend my warmest thanks to Dr. Kees Aarts, Dr. Rudy Andeweg, Hans Anker, Dr. Rosa Berganza, Dr. Arantxa Capdevila, Dr. Bruno Cautrès, Dr. Carlos Closa, Dr. Ben Crum, Dr. Yves Déloye, Dr. Jos de Beus, Dr. Claes de Vreese, Dr. Catherine de Vries, Dr. Jan Erk, Fernand Fehlen, Dr. Joan Font, Dr. Jacques Gerstlé, Dr. Lorena Gómez, Dr. Otto Holman, Philippe Hubert, Dr. Jan Kleinnijenhuis, Dr. André Krouwel, Charles Margue, Dr. Peter Neijens, Dr. Henk Overbeek, Dr. Philippe Poirier, Dr. Andreas Schuck, Jürgen Stoldt, Janet Takens, Dr. José Ignacio Torreblanca, Dr. Wouter van der Brug, Dr. Henk van der Kolk, Dr. Joop van Holsteyn, Dr. Jordi Vaquer, and Dr. Tània Verge for generously sharing their data and/or analyses with me. I am very grateful to Claudine Lefort, Honorary Consul of Luxembourg in Montreal, for her kind help translating Lëtzebuergesch; and to Murat Yurdakul for his assistance with French. I am especially indebted to all of my interviewees who spared time in their busy schedules to answer my questions. Last but not least, I would like to thank my sister and brother-in-law Begum and Gokce, and my friends Yasemin, Pedro, and Tada for opening their homes to me, making sure that I would not get lost, and for all of their logistical and emotional support. Without them, my field work would have been so much harder and much less fun.

The support of my friends and family has been indispensable during these intense years. Major thanks go to my colleagues and close friends at McGill; Marc André, Bahar A., Zeynep, Françoise, Ora, Adam, Taryn, Betul, Mert, Andy, Julie, Theo, Dan D., Melanee, Nina, Sarah-Myriam, Imad, Jennifer, and Lamis. It was a pleasure to have such great company while going through the program, sharing so many stressful and joyful moments over these years. They brought all the meaning and fun to my McGill life. Samia, Bahar K., and Pinar deserve special thanks for bearing with me throughout and reminding me that I had a life outside of my dissertation. Thanks also to Rosie and Sammy for providing me such a joyful second home in Boston, where I wrote quite a few of my chapters.

vii

Despite being halfway around the world, there was not one day that I did not feel the love and emotional support of my parents Şükran and Sadık, and my sister Cansu. Just as everything else in my life, I simply would not be able to achieve this without their encouragement. My final and most special thanks go to my husband Kerem, who has been there every single step of the way. His presence, right beside me writing his own dissertation, turned this experience into a pleasure. It is to the four of you that this dissertation is dedicated.

viii

LIST OF FIGURES AND TABLES Figures 1.1 Analytical Framework 23 1.2 Diffusion effects among the 2005 TCE referenda 23 2.1 Diffusion effects among the 2005 TCE referenda 69 3.1 Evolution of voting intentions in Spain 85 3.2 Spanish support for EU membership (1984-2004) 94 3.3 Spanish public’s perceived benefit from EU membership (1984-2004) 95 3.4 The PSOE’s poster 109 3.5 The PP’s poster 109 3.6 The information campaign’s poster 109 3.7 The IU’s poster 109 3.8 The CiU’s poster 109 3.9 The ERC’s poster 110 3.10 The ICV’s poster 110 4.1 Evolution of vote intentions in France (October 2003-May 2005) 137 4.2 Trust levels in Chriac and Raffarin`s ability to address problems (2003-2005) 139 4.3 French support for EU membership (1984-2004) 144 4.4 French public`s perceived benefit from EU membership (1984-2004) 145 4.5 French support for EU membership (1973-2005) 145 4.6 The UMP’s poster 161 4.7 The MPF’s poster 161 4.8 The PS’ poster 161 4.9 The FN’s poster 161 4.10 Posters of the UDF and PCF 161

4.11 The FN’s poster 162 4.12 The ATTAC’s poster 162 4.13 Cartoon on the French 2005 TCE campaign 172 4.14 Cartoon on the internal split of the PS in the French 2005 TCE campaign 172 4.15 Evolution of vote intentions in France (September 2004-May 2005) 173 4.16 Evolution of vote intentions in France and campaign events (September 2004-March 2005) 174 4.17 Evolution of vote intentions in France and campaign events (September 2004-May 2005) 174 4.18 Party share of the French No vote 184 5.1 Evolution of No vote intentions in the Netherlands 195 5.2 Satisfaction levels with the Dutch government (1995-2006) 197 5.3 Dutch support for EU membership (1984-2004) 203 5.4 Dutch public’s perceived benefit from EU membership (1984-2004) 204 5.5 Dutch support for EU membership (1973-2005) 204 5.6 Documents used by the government 220 5.7 The GL’s poster 220 5.8 Poster of the Foundation for a Better Europe 220 5.9 The CDA’s poster 220 5.10 The PvdA’s poster 220 5.11 The SP’s poster 221 5.12 The ATTAC’s poster 221 5.13 The ConstitutionNo’s poster 221 5.14 Posters of the SP and the PvdA 221 5.15 Cartoon depicting the Yes campaign material in the Dutch 2005 TCE referendum 231

5.16 Cartoon depicting the Yes campaign style in the Dutch 2005 TCE referendum 232 6.1 Evolution of vote intentions in Luxembourg 259 6.2 Breakdown of Luxembourg workforce according to nationality 263 6.3 Percentage share of foreigners in the EU member states 264 6.4 Luxembourgish support for EU membership (1984-2004) 268 6.5 Luxembourgish public’s perceived benefit from EU membership (1984-2004) 269 6.6 Luxembourgish support for EU membership (1973-2005) 269 6.7 The CSV’s poster 282 6.8 The LSAP’s poster 282 6.9 The DP’s poster 282 6.10 The Greens’ poster 282 6.11 The CSV’s final poster 282 6.12 Posters of the UNEL-No Committee 283 6.13 Posters of the No Committee 283 6.14 Cartoon on PM Juncker`s role in the Luxembourgish 2005 TCE campaign 287 6.15 Evolution of vote intentions in Luxembourg and campaign events 294 6.16 Number of temporary employees in Luxembourg over time 311 6.17 Breakdown of Luxembourg temporary workforce according to nationality 312 6.18 Geographical (communal) distribution of the Luxembourgish no vote 315 6.19 Geographical (percentile) distribution of the Luxembourgish no vote 317 7.1 Analytical Framework 324 7.2 Identical posters – against the TCE – in different European countries 325-6

Tables 1.1 The interaction between campaign strength and diffusion 24 1.2 The active campaigners in the Spanish case 26 1.3 The active campaigners in the French case 27 1.4 The active campaigners in the Dutch case 28 1.5 The active campaigners in the Luxembourgish case 29 2.1 The analytical framework of the independent variable – Relative No campaign strength 64 2.2 The analytical framework of the conditioning variable – Diffusion 68 3.1 Spain’s benefit and contribution in the EU budget (2000-2005) 92 3.2 Valence towards the TCE in the Spanish media coverage 120 3.3 Topical distribution of the Spanish media coverage 120 3.4 Reasons to vote Yes in Spain 123 3.5 Reasons to vote No in Spain 124 3.6 Spain's association of EU membership with modernization and democracy 125 3.7 Spaniards’ evaluation of the TCE concerning various campaign topics 125 3.8 Spaniards’ evaluation of the TCE concerning various campaign topics – the Yes/No breakdown 126 4.1 France’s benefit and contribution in the EU budget (2000-2005) 147 4.2 Topical coverage of the French Yes and No campaigns 176 4.3 Evolution of referendum debate in French people’s conversations 177 4.4 Issues taken into consideration by the French voters 179 4.5 Reasons to vote Yes in France 180 4.6 Reasons to vote No in France 181 4.7 French public’s evaluation of various campaign topics – the Yes/No breakdown 182

xii

4.8 The breakdown of the French Yes/No vote by party proximity 183 5.1 The Netherlands’ benefit and contribution in the EU budget (2000-2005) 206 5.2 Topical distribution of the Dutch news coverage 234 5.3 Dutch public’s opinions on the consequences of European integration 236 5.4 Dutch public’s evaluation of various steps taken by the EU 238 5.5 Dutch public’s evaluation of the Euro – the Yes/No breakdown 239 5.6 Dutch public’s opinions on the consequences of the TCE 240 5.7 Reasons to vote Yes in the Netherlands 241 5.8 Reasons to vote No in the Netherlands 243 6.1 Luxembourgers’ attitude towards foreigners in various domains 265 6.2 Luxembourgers’ attitude towards different groups of foreigners 266 6.3 Luxembourg’s benefit and contribution in the EU budget (2000-2005) 270 6.4 Breakdown of the Luxembourgish Yes/No vote by party proximity 296 6.5 Reasons to vote Yes in Luxembourg 297 6.6 Reasons to vote No in Luxembourg 298 6.7 Number of incoming cross-border employees 310 6.8 Number of outgoing cross-border employees 310 6.9 Breakdown of Luxembourg temporary workforce according to nationality 312 6.10 Timing of vote decision in Luxembourg – the Yes/No breakdown 315 7.1 The interaction between campaign strength and diffusion 326 7.2 The relative campaign strength in 2005 EU Constitutional Referenda 327 7.3 Diffusion effects in 2005 EU Constitutional Referenda 327

xiii

LIST OF ABBREVIATIONS

ABBL Association of Banks and Bankers of Luxembourg

ADR Alternative Democratic Reform Party (Luxembourg) ATTAC Association for the Taxation of Financial Transactions for the Aid of Citizens BNG Galician Nationalist Block (Spain) CC Canarian Coalition (Spain) CCOO Workers Commissions (Spain) CDA Christian Democratic Appeal (The Netherlands) CEOE Confederation of Employers and Industries of Spain CEVIPOF Centre for Political Research at Sciences Po (France) CFDT French Democratic Confederation of Labour CFTC French Confederation of Christian Workers CGC General Confederation of Executives (France) CGT General Workers’ Confederation (France) CHA Aragonese Council (Spain) CIDOB Barcelona Centre for International Affairs (Spain) CIS Center of Sociological Investigations (Spain) CiU Catalan Nationalist Convergence and Unity (Spain) CLC Chamber of Commerce (Luxembourg) CPB Netherlands Bureau for Economic Policy Analysis CSA Advice on Survey Analysis (France) CSV Christian Social People's Party (Luxembourg) CU Christian Union (The Netherlands) D66 Democrats 66 (The Netherlands) DP Democratic Party (Luxembourg) EA Basque Solidarity (Spain)

xiv

EMU Economic and Monetary Union ENC European No Campaign EP European Parliament EPERN European Parties Elections and Referendums Network ERC Republican Left of Catalonia (Spain) ESF European Social Forum ETA Basque Homeland and Freedom (Spain) EU European Union EURES European Employment Services DLR Arise the Republic (France) FEDIL Federation of Luxembourg Industrialists FN National Front (France) FNV Federation Dutch Labor Movement FO Workers’ Force (France) GDP Gross Domestic Product GL GreenLeft (The Netherlands) ICV Initiative for Catalonia Greens (Spain) IFOP French Institute of Public Opinion IU United Left (Spain) KPL Communist Party (Luxembourg) LCGB Luxembourg Confederation of Christian Trade Unions LCR Revolutionary Communist League (France) LO Workers’ Struggle (France) LPF List Pim Fortuyn (The Netherlands) LSAP Luxembourg Socialist Workers' Party MEP Member of European Parliament MFA Ministry of Foreign Affairs

MNR National Republican Movement (France) MP Member of Parliament MPF Movement for France MRC Citizen and Republican Movement (France) NATO North Atlantic Treaty Organization NGO Non-governmental Organization OGBL Confederation of Independent Trade Unions of Luxembourg PCE Communist Party of Spain PCF French Communist Party PNV Basque Nationalist Party (Spain) PP Popular Party (Spain) PS Socialist Party (France) PSOE Spanish Socialist Workers’ Party PvdA Dutch Labor Party PVV Party of Freedom (The Netherlands) RPF Movement for France RTL Luxembourg Television and Radio SCP Netherlands Institute for Social Research SGP Reformed Political Party (The Netherlands) SOFRES French Survey Institute SP Socialist Party (The Netherlands) SUD Solidarity Unity Democracy (France) TCE Treaty establishing a Constitution for Europe TNS-ILRES Luxembourg Institute for Social Research UDF Union for French Democracy UGT General Workers Union (Spain) UMP Union for a Popular Movement (France)

xvi

UNEL National Union of Luxembourgish Students VVD People's Party for Freedom and Democracy (The Netherlands) WRR Scientific Council for Government Policy (The Netherlands)

xvii

CHAPTER 1 THE PUZZLE OF EU REFERENDA

The European Union (EU) increasingly resembles a state, and it has taken important steps to build a political union in recent years. Despite these developments, there are serious concerns about the ‘democratic deficit’ in the Union.1 Popular legitimacy only weakly supports significant decisions taken at the supranational level, as low voter turnout at the European Parliament (EP) elections and rising Euroscepticism throughout Europe indicate. Therefore, public attitudes towards the Union form a fresh and dynamic area of study, one that has not attracted sufficient attention to date in the EU literature. In October 2004, the 25 EU member states and three candidate countries signed the Treaty establishing a Constitution for Europe (TCE). Each member state then sought to ratify this treaty through either parliamentary vote or public referendum. Only four member states employed the referendum method: Spain (20 February 2005), France (29 May 2005), the Netherlands (1 June 2005), and Luxembourg (10 July 2005). While the referenda in Spain and Luxembourg approved the TCE, the ones in France and the Netherlands rejected it. Polls show that public opinion in all four countries favored the European Constitution in fall 2004, several months before the referenda. In all cases the mainstream parties – social democrats, liberals and Christian democrats – supported the TCE regardless of whether they were in or out of government. All four referenda were non-compulsory and binding. In all four cases both the Yes and No voters shared a well-defined sociodemographic profile. Yet the final referenda results varied strikingly, which leads to an interesting puzzle. Given the similarities among the cases, why did the public in France and the Netherlands reject the TCE while those in Spain and Luxembourg did not? Put differently, why was the public's initial positive attitude towards the TCE sustained in Spain and Luxembourg but not in France and the Netherlands?

1 For a recent discussion on the ‘democratic deficit’ in the European Union, see Andreas Follesdal and Simon Hix, "Why There is a Democratic Deficit in the EU: A Response to Majone and Moravcsik," Journal of Common Market Studies 44, no. 3 (2006). 1

The dependent variable does not simply reflect the outcome of the 2005 TCE referenda (ratification vs. non-ratification). Instead, the dependent variable encompasses both the absolute magnitude of the No votes in the referenda and the relative movement of public opinion (comparing the percentage opposed to the TCE in fall 2004 with the eventual No vote totals). The No vote percentages in the referenda were Spain 17.24%, France 54.68%, the Netherlands 61.5%, and Luxembourg 43.48%. The differentials were, approximately, +13% in Spain, +23% in France, +50% in the Netherlands, and finally +26% in Luxembourg. The existing literature has tried to attribute the referenda outcomes to either domestic or European-level factors, leading to inconclusive results. Moving beyond this false dichotomy, I propose a dynamic explanation that captures the interaction between the two domains. I argue that the stronger a state's No campaign relative to its Yes campaign – that is, the better the No campaigners linked existing contentious issues to the European Constitution – the greater the increase in the magnitude of the No vote. While campaign strength is the main independent variable, I also argue that the temporal sequencing of the referenda could affect the relative strength of the No campaign in second-mover states. The later a state held its referendum, the more the previous referenda results and campaigns in other states could influence campaign dynamics. Nonetheless, this diffusion effect was not automatic and depended on channels such as shared language/culture, common media sources, and collaborative networks/transnational linkages. Where these channels were present and the campaign of the first-mover state was intense, diffusion amplified the strength of referenda campaigns in second-mover states. In this chapter I will first introduce the literature on public attitudes towards the EU and the literature on EU referenda. Second, I will summarize the findings of the studies that focus specifically on the 2005 EU Constitutional referenda. Third, I will criticize the existing literature based on my comparative framework. Bringing together the positive and negative cases and discussing the important similarities among them, I will rule out the existing explanations in the

2 literature. Finally I will stipulate my hypotheses, provide a brief introduction to the cases, and present my methodology.

Understanding Attitudes towards the EU The first step to discussing EU referenda results is the analysis of what forms the basis of people’s attitudes towards the Union. Scholarly interest in public attitudes towards the EU is recent and multidimensional. There are two different approaches in the literature: Political party-based studies of Euroscepticism and individual-level analyses of public support for the EU. Studies of Euroscepticism focus on the manifestation of negative public opinion towards the Union in party politics. The main contribution of this literature is that pervasive Euroscepticism is mainly limited to parties on the periphery of the party system.2 While far left and far right parties share Euroscepticism, parties in the middle, including most social democratic, Christian democratic, liberal, and conservative parties, are generally much more supportive of European integration.3 While this literature has been useful in understanding the Eurosceptic party structure in the EU member states, it has so far not considered other venues of negative attitudes towards the EU such as civil society organizations. The second important strand in the literature examines public support for the EU at the individual level of analysis. Some scholars have adopted an economic perspective and questioned whether citizens’ attitudes towards international economic policy reflect their economic interests. Employing a macroeconomic model Eichenberg and Dalton questioned whether support for the EU was consistent with national economic performance through indicators such as GDP, unemployment or inflation, and found that only inflation had a

2 Paul Taggart, "A Touchstone of Dissent: Euroscepticism in Contemporary Western European Party Systems " European Journal of Political Research 33, no. 3 (1998). See also Simon Hix, "Dimensions and Alignments in European Union Politics: Cognitive Constraints and Partisan Responses," European Journal of Political Research 35, no. 1 (1999). 3 Liesbet Hooghe, Gary Marks, and Carole Wilson, "Does Left/Right Structure Party Positions on European Integration?," Comparative Political Studies 35, no. 8 (2002). Also see Gary Marks et al., "Party Competition and European Integration in the East and West: Different Structure, Same Causality," Comparative Political Studies 39, no. 2 (2006). 3 significant effect.4 Gabel and Palmer on the other hand opted for a microeconomic model and showed that support for the EU varies with several socioeconomic characteristics such as income influenced by market liberalization.5 There is strong evidence that citizens' support for the Union varies with occupation-based economic interests. From a different perspective, McLaren proposed that people have negative attitudes towards the Union because of their perceptions of threats posed by other cultures.6 Carrying this idea further, de Vreese and Boomgaarden stressed the impact of people’s attitudes towards immigration on support for the EU.7 This literature is crucial to specifying two sets of important variables; the positioning of the political parties towards the EU, and the determinants of people’s general attitudes towards the EU. Yet these variables, by themselves, are too static to account for the referendum results. They need to be studied in relation to the EU referendum campaigns, to understand how these factors interact with specific referendum topics and campaign dynamics. Studies on referendum campaigns show that especially in referenda on complex matters – such as the TCE – initially voters do not hold well-informed opinions, and the campaign cues shape public opinion significantly.8 Rather than explaining the referendum outcomes, the findings of this literature are therefore more valuable in

4 Richard Eichenberg and Russell Dalton, "Europeans and the European Community: The Dynamics of Public Support for European Integration," International Organization 47, no. 4 (1993). 5 Matthew Gabel and Harvey Palmer, "Understanding Variation in Public Support for European Integration," European Journal of Political Research 27, no. 1 (1995). For an excellent review of this literature see Matthew Gabel, "Economic Integration and Mass Politics: Market Liberalization and Public Attitudes in the European Union," American Journal of Political Science 42, no. 3 (1998). 6 Lauren McLaren, "Public Support for the European Union: Cost/Benefit Analysis or Perceived Cultural Threat?," The Journal of Politics 64, no. 2 (2002). Similarly Christin and Trechsel suggested that perception of threat to national interests is significant in understanding the public opinion towards the Union. See Thomas Christin and Alexander Trechsel, "Joining the EU?: Explaining Public Opinion in Switzerland," European Union Politics 3, no. 4 (2002). 7 Claes de Vreese and Hajo Boomgaarden, "Projecting EU Referendums: Fear of Immigration and Support for European Integration," European Union Politics 6, no. 1 (2005). 8 Lawrence LeDuc, "Opinion Change and Voting Behaviour in Referendums," European Journal of Political Research 41, no. 6 (2002). See also John Zaller, The Nature and Origins of Mass Opinion (New York: Cambridge University Press, 1992).; Sara Hobolt, "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums," Journal of Elections, Public Opinion and Parties 15, no. 1 (2005). 4 understanding the initial reactions towards the TCE several months before the referenda. Early positive public opinion could be interpreted as reflecting general attitudes towards the EU, as opposed to the specific and informed opinions on the TCE revealed by the referendum results.

Understanding the EU Referenda Results Neither the Euroscepticism nor the public support literature attempted to explain how party or individual attitudes translated into EU referenda results. I introduce the general EU referenda literature in this section, after which I will discuss the studies on 2005 referenda results specifically. The literature on referenda results constitutes a separate field as it requires the analysis of voting behavior. Academic interest in analyzing EU referenda results has grown in recent decades, with its roots in the study of EP elections. The main division within this literature is between ‘second-order’ and ‘attitude/issue-voting’ interpretations, in other words between attributing the results to domestic or European factors. The first approach points to the ‘second-order’ characteristic of EU referenda, proposing that national issues dominate the campaign agenda as well as voters’ preoccupations. Local, presidential and EP elections are considered to be ‘second-order’ elections, as voters do not perceive them to be as important as national elections. The implication for the EU is that EP elections do not reflect voters’ opinion on integration issues, and Union issues are relegated to second place.9 When applied to EU referenda, the second order model argues that voting behavior is strongly influenced by national factors such as the level of satisfaction with the government and voter identification with parties holding office.10 The ‘attitude/issue-voting’ model on the other hand proposes the alternative first order hypothesis that citizens vote in line with their underlying

9 Karlheinz Reif and Hermann Schmitt, "Nine Second-Order National Elections: A Conceptual Framework for the Analysis of European Election Results," European Journal of Political Research 8, no. 1 (1980). 10 Mark Franklin, Cees van der Eijk, and Michael Marsh, "Referendum Outcomes and Trust in Government: Public Support for Europe in the Wake of Maastricht," West European Politics 18 (1995). 5 broad attitudes towards European integration.11 This model rejects the first model’s claim that individual voters are unable to form their own opinions on low salience issues such as the EU treaties. Instead it suggests that voters’ attitudes towards EU integration explain the referenda results; the fewer areas voters want included in European integration, the more they vote No. However, there is empirical evidence pointing to the simultaneous presence of EU issue-voting and second-order effects. Survey-based studies find that both dissatisfaction with the government and attitudes towards the European integration influence the vote choice. As a result of this mixed evidence, scholars have started to ask the new question of why and under what circumstances individual voters are more likely to rely on attitudes rather than second-order effects and vice versa.12 This new perspective broadened the range of causal factors studied with new variables such as institutional setting, party cues, issue salience and level of information, and finally campaign effectiveness. To start with the institutional setting, Hug proposed that government support among voters would increase in non-compulsory and binding referenda.13 Alternatively, research on party cues suggests that if voters have limited information on the subject, political parties can provide them with cues.14 Similarly, if the political parties treat the referendum as an opportunity to test the government’s standing, voters are expected to be more influenced by the domestic

11 Karen Siune, Palle Svensson, and Ole Tonsgaard, "The EU: The Danes Said “No” in 1992, But “Yes” in 1993: How and Why?," Electoral Studies 13, no. 2 (1994). Also see Palle Svensson, "Five Danish Referendums on the European Community and European Union: A Critical Assessment of the Franklin Thesis," European Journal of Political Research 41, no. 6 (2002). 12 Sara Hobolt, "Direct Democracy and European Integration," Journal of European Public Policy 13, no. 1 (2006): 155. 13 Simon Hug, Voices of Europe: Citizens, Referendums and European Integration (Lanham: Rowman & Littlefield Publishers, 2002). See also Simon Hug and Pascal Sciarini, "Referendums on European Integration: Do Institutions Matter in the Voter’s Decision?," Comparative Political Studies 33, no. 1 (2000). 14 For sophisticated but conflicting variations of this argument see Leonard Ray, "Reconsidering the Link between Incumbent Support and Pro-EU Opinion," European Union Politics 4, no. 3 (2003).; Ben Crum, "Party Stances in the Referendums on the EU Constitution: Causes and Consequences of Competition and Collusion," European Union Politics 8, no. 1 (2007).; LeDuc, "Opinion Change and Voting Behaviour in Referendums."; Claes de Vreese, "Political Parties in Dire Straits? Consequences of National Referendums for Political Parties," Party Politics 12, no. 5 (2006).; Matthew Gabel and Kenneth Scheve, "Mixed Messages: Party Dissent and Mass Opinion on European Integration," European Union Politics 8, no. 1 (2007). 6 arena. Insufficient knowledge or competition among political parties can thus lead to reliance on political parties’ positions on the subject. As such, arguments emphasizing the institutional setting and party cues both point to the second-order effects under the specified conditions. In a similar fashion, another important variable discussed in the literature is issue salience and level of information. When salience is high and the voters have a great interest in and knowledge of European affairs, they are more likely to rely on their attitudes towards the Union and not on domestic dissatisfaction with the government.15 In other words, domestic concerns will be a powerful determinant of voting only in situations where the referendum is perceived to be unimportant or where voters lack knowledge of EU matters. Another important variable is the intensity of the campaign, suggesting that issue salience and knowledge of the EU are closely related to the campaigning process.16 In the most comprehensive comparative study so far, based on a statistical analysis of 19 referenda, Hobolt argues that the information provided to voters during the campaigns determines which model will prevail.17 In intense campaigns with conflicting messages, voters assess whether elite endorsements are compatible with their feelings about European integration and thereby rely on their underlying attitudes towards the EU. As this preview demonstrates, the literature tends to connect the last three variables: party cues, issue salience/level of information, and campaign effectiveness. I discuss some of these new concepts and research – particularly those concerning campaign intensity – in greater detail in the next chapter. However, these various studies

15 See Mark Franklin, "Learning from the Danish Case: A Comment on Palle Svensson's Critique of the Franklin Thesis," European Journal of Political Research 41, no. 6 (2002). Also see Hobolt, "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums."; Sara Hobolt, "Campaign Information and Voting Behavior in EU Referendums," in The Dynamics of Referendum Campaigns: An International Perspective, ed. Claes De Vreese (Basingstoke: Palgrave MacMillan, 2007). 16 John Garry, Michael Marsh, and Richard Sinnott, "'Second-order' versus 'Issue-voting' Effects in EU Referendums: Evidence from the Irish Nice Treaty Referendums," European Union Politics 6, no. 2 (2005).;Sara Hobolt, "How Parties Affect Vote Choice in European Integration Referendums," Party Politics 12, no. 5 (2006). ; ———, Europe in Question: Referendums on European Integration (New York: Oxford University Press, 2009). 17 The second part of her argument deals with how individual’s attention to political information also affects the voting behavior. Those who pay more attention to politics are expected to be less susceptible to the recommendations of national politicians. 7 still employ an ‘either/or’ logic, as the main goal is to specify when the results reflect domestic or EU-related concerns. A few single-case studies move beyond this dichotomy to a certain extent, but eventually still identify attitudes towards Europe as the key mechanism in determining referendum outcomes. De Vreese and Boomgaarden expect that high levels of anti-immigrant sentiments, pessimistic economic outlooks, and unpopular governments lead to a No vote in a referendum.18 Nevertheless, they present these factors as ‘antecedents’ of voting behavior – essentially attitudes towards the EU – and acknowledge the need to study campaign dynamics. Studying the 2000 Danish referendum campaign on the Euro, de Vreese and Semetko predict that the following factors increase the likelihood of voting No in referenda: EU-scepticism, government disapproval, pessimistic personal (economic) expectations, left ideology, and frequent exposure to certain types of media channels.19 However, the issue-related contextual variable of Euroscepticism comes out as the most important predictor of the vote. Schuck and de Vreese present similar findings on the Dutch 2005 TCE referendum.20 They measure Euroscepticism as a multiple index scale covering general EU support, support for the Euro, Turkish EU membership and the speed of EU enlargement. While they confirm that government disapproval, national identity and fear of globalization also have effects on the No vote, they show that these are partially mediated by their index of Euroscepticism. In their analysis, this general Euroscepticism is the dominant factor explaining why people intended to vote against the TCE three weeks prior to the referendum. As for the

18 Claes de Vreese and Hajo Boomgaarden, "Immigration, Identity, Economy and the Government: Understanding Variation in Explanations for Outcomes of EU-related Referendums," in The Dynamics of Referendum Campaigns: An International Perspective, ed. Claes De Vreese (Basingstoke: Palgrave Macmillan, 2007). Also see Claes de Vreese, "Why European Citizens Will Reject the European Constitution," Harvard University Minda de Gunzburg Centre for European Studies Working Paper 116 (2004). 19 See Claes de Vreese and Holli Semetko, Political Campaigning in Referendums: Framing the Referendum Issue (Abingdon: Routledge, 2004), 164. This work analyzes the campaign frames (both elite framing and media frames) in detail and discusses their contribution to the result but not as a causal factor per se. It assesses campaign effects through media. 20 Andreas Schuck and Claes de Vreese, "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign," Journal of Elections, Public Opinion, and Parties 18, no. 1 (2008). 8 campaign, they find that it mattered only to a limited extent and that higher exposure to referendum news contributed to voters shifting over to the Yes side. However their data – a two-wave panel study – covers a short period, the last three weeks of the campaign. Focusing only on the last few weeks of the campaign is problematic because it ignores the longer opinion formation period. Strikingly, the public in all four countries held positive attitudes towards the TCE several months before the referenda, in fall 2004. In this regard, studying only the last weeks limits the analysis to the final phase where many voters have already made up their minds. More importantly, a closer look at the 2005 referendum campaigns reveals that when earlier polls indicated favorable opinions towards the TCE, both the French and Dutch publics were already dissatisfied with their governments and concerned about issues such as deterioration of the welfare system, increasing immigration, unemployment. Put simply, the contention was already there and these variables carried similar negative values at the time of the initial positive opinion polls. Thus it is crucial to look at the long campaign, to take a step back and assess how the political elite framed the debate at the starting point. I argue that prior to effective framing by No campaigners, these problems remained unlinked to the TCE in the public's mind several months before the referenda. Another factor in understanding the EU referenda results is the sequencing of cases. In 2005, the goal was arguably to build momentum by starting with Spain, a highly positive country.21 The EU referenda literature has a few references to the importance of sequencing, however there is no research on how it works.22 So far, only one study looked into the sequencing of the 1994 EU membership referenda in Austria, Finland, Sweden, and Norway. Jahn and Storsved argue that there was a domino strategy, where the most pro EU countries (Austria and Finland) started the referendum vote followed by the two unsure

21 Personal interview with José Ignacio Torreblanca, 6 October 2008, Spain. 22 Lawrence LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," Journal of Elections, Public Opinion & Parties 15, no. 2 (2005): 186. Concerning the 2005 TCE referenda, LeDuc recognized the potential for the continual flow of ‘new’ information among the cases and warned about a ‘domino effect’. 9 cases (Sweden and Norway).23 The idea was that the presumed positive vote of the ‘safe’ countries would bring the other countries into the EU. While the strategy worked in Sweden, it failed in Norway. Moreover in a survey, only 56% of the Finns showed awareness of the positive results in Austria, whereas the Scandinavian countries were much more interested in each others’ decisions. Some opinion polls showed that a No victory in Finland would have brought a 54% No vote in Sweden and 55% in Norway. While these findings are highly interesting, this article does not elaborate on the specific mechanisms that transmitted such influence among the cases. It suggests that Norway rejected the treaty mainly because the Norwegian Centre Party, which as an established political force, was able to organize the anti-EU movement. It successfully connected agrarian (including fishing) interests with the national character and independence. The authors also highlight the positive Norwegian economic situation and the long history of foreign occupation which made national independence more important. These findings not only point to the importance of interconnectedness and diffusion among Scandinavian countries, but also support my independent variable’s significance by emphasizing the strategic No campaign. Further research should try to disentangle these factors in understanding EU referendum campaigns. I seek to achieve this in my theoretical framework.

Understanding the 2005 Constitutional Referenda Results The literature that specifically seeks to explain the 2005 constitutional referenda results reflects the problematic patterns explained above. This literature is dominated by single in-depth case studies, mainly focusing on the reasons for rejection in the French or Dutch cases.24 These country-specific studies suggest

23 Detlef Jahn and Ann-Sofie Storsved, "Legitimacy through Referendum? The Nearly Successful Domino-Strategy of the EU-Referendums in Austria, Finland, Sweden and Norway," West European Politics 18, no. 4 (1995): 21. Austria held its referendum on 12 June (66.6% Yes), Finland on 16 October (56.9% Yes), Sweden on 13 November (52.3% Yes), and finally Norway was on 27 November (47.8% Yes). 24 See Gilles Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second-order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," West European Politics 29, no. 1 (2006). Kees Aarts and Henk van der Kolk, 10 various explanations. Most relate the negative votes to second-order effects due to the unpopular incumbent governments in France and the Netherlands, the Chirac and Balkenende administrations respectively. Also, these works present the recent 2004 enlargement of the EU and the potential membership of Turkey as perceived threats to Western culture and national identity. Most case studies of the French vote emphasized concerns over the loss of the social model and argued that French voters responded to deteriorating social welfare conditions. However, they also mentioned that the French No vote was for another – more social – Europe. The most thorough study of the French case ruled out the party cues hypothesis, and highlighted the significance of both domestic and European factors: general scepticism towards Europe, specific EU-scepticism (concerning primarily social fears and partly enlargement), and government popularity.25 Similarly, the most detailed case studies of the Dutch vote presented factors relating to both domestic and European domains: the gap between the mainstream elite and voters, the negative attitudes towards the EU enlargement process, and the perception that the pace of the integration process is too fast, highlighting the unpopularity of measures such as the Euro.26

"Understanding the Dutch 'No': The Euro, the East, and the Elite," Political Science and Politics 39, no. 2 (2006). Henry Milner, "YES to the Europe I Want; NO to This One? Some Reflections on France's Rejection of the EU Constitution," Political Science and Politics 39, no. 2 (2006). Sylvain Brouard and Vincent Tiberj, "The French Referendum: The Not So Simple Act of Saying Nay," Political Science and Politics 39, no. 2 (2006), Mabel Berezin, "Appropriating the 'No': The French National Front, the Vote on the Constitution, and the 'New' April 21," Political Science and Politics 39, no. 2 (2006). Raphael Franck, "Why Did a Majority of French Voters Reject the European Constitution?," European Journal of Political Economy 21, no. 4 (2005). Leonard Besselink, "Double Dutch: The Referendum on the European Constitution," European Public Law 12, no. 3 (2006). Bertrand Lemennicier, "Political Polarization and the French Rejection of the European Constitution," European Journal of Political Economy 21, no. 4 (2005). Paul Hainsworth, "France Says No: The 29 May 2005 Referendum on the European Constitution," Parliamentary Affairs 59, no. 1 (2006). Jacques Gerstlé, "The Impact of Television on French Referendum Campaign in 2005," Notre Europe (2006).; Marcel Lubbers, "Regarding the Dutch 'Nee' to the European Constitution: A Test of the Identity, Utilitarian and Political Approaches to Voting 'No'," European Union Politics 9, no. 1 (2008). 25 Brouard and Tiberj, "The French Referendum: The Not So Simple Act of Saying Nay." See also Sylvain Brouard and Nicolas Sauger, "Comprendre la Victoire du "Non" : Proximité Partisane, Conjoncture et Attitude à l’égard de l’Europe," in Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, ed. Annie Laurent and Nicolas Sauger (Paris: CEVIPOF, 2005). 26 See Aarts and van der Kolk, "Understanding the Dutch 'No': The Euro, the East, and the Elite."; Schuck and de Vreese, "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign." 11

A few studies focused on the positive cases. Font and Rodriguez’s analysis of the Spanish case argued that party cues were the strongest determinant of the vote.27 Strikingly, a thorough analysis of the Luxembourgish case by the University of Luxembourg showed that despite being a positive vote, this case carried similarities with the French and Dutch cases such as the elite-public gap, and social- and identity-related concerns.28 This finding confirms that these factors were not sufficient by themselves to cause the negative results. So far, only a few studies have very briefly compared the two cases of No votes, occasionally referring to the positive cases. Stefanova argued that in both cases, due to negative spillover, political discourse related the TCE to the Union’s enlargement policies.29 However, her work does not put forward a causal analysis of the No votes; instead she focuses on the implications of these negative votes for Eastern Europe. Dehousse, on the other hand, suggested that the negative votes were essentially the result of second-order voting because of the unpopularity of the incumbent governments.30 Nevertheless, his analysis is very brief and not based on rigorous research. Similarly, Qvortrup emphasized that the Constitution was largely negotiated by center-right governments, which led to concerns over the deterioration of welfare state.31 In her larger comparative study, Hobolt argued that the underlying attitudes towards the EU explained the French and (to a lesser extent) the Dutch votes but suggested that campaigns played an important role in shaping these attitudes.32 In the less intense Dutch referendum, party cues were stronger. Crum carried out a comparative analysis of all four cases, but focused on explaining the behavior of political parties in the referendum campaign and not

27 Joan Font and Elisa Rodriguez, "The Spanish Referendum on the EU Constitution: Issues, Party Cues and Second-Order Effects," in The Midwest Political Science Association Annual National Conference (Chicago, U.S.A. 2006). 28 Patrick Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés (Luxembourg: University of Luxembourg, 2007). 29 Boyka Stefanova, "The 'No' Vote in the French and Dutch Referenda on the EU Constitution: A Spillover of Consequences for the Wider Europe," Political Science and Politics 39, no. 2 (2006). 30 Renaud Dehousse, "The Unmaking of a Constitution: Lessons from the European Referenda," Constellations 13, no. 2 (2006). 31 Matt Qvortrup, "The Three Referendums on the European Constitution Treaty in 2005," The Political Quarterly 77, no. 1 (2006). 32 Hobolt, Europe in Question: Referendums on European Integration. 12 the referenda results.33 He argued that it was up to the governing parties to win the referendum as some voters of the opposition parties were bound to vote against. This is what he suggested Spanish and Luxembourgish governing parties succeeded in doing. While this is an important observation, he did not conduct a thorough analysis of the campaign dynamics to specify which factors contributed to their success, nor has he proposed how attitudes towards Europe might have interacted with this pattern. An unpublished manuscript by Glencross and Trechsel used Eurobarometer survey data to examine whether second-order or attitude models explain voting in these four cases, finding stronger support for the first-order character of the votes.34 Overall the explanations based on single-case studies are too disparate, while the comparative ones only ask whether second- order or attitude models explain the results better. I argue that the distinction between the second-order versus attitude models presents a false dichotomy. As the mixed evidence suggests, these theories are not mutually exclusive. In countries that have been members of the Union for decades, the distinction between domestic and European domains is considerably blurred. The Union has gained more and more competences over the years, and now legislates for its members in a vast number of policy areas. Among the policies that are under the competence of the EU, some enjoy significant visibility to the citizens such as monetary or immigration policies. Therefore, as the cases considered here are three founding members and a country that joined in the 1980s, the distinction between domestic vs. European factors cannot be maintained.35 What is needed is not to argue that one or the other model better explains the referenda outcomes, but instead to construct a strong theoretical framework that captures the interaction between the two domains.

33 Crum, "Party Stances in the Referendums on the EU Constitution: Causes and Consequences of Competition and Collusion." His work implies that the referendum outcomes were a factor of governing parties’ ability to mobilize the Yes vote in the face of protest voting. 34 Andrew Glencross and Alexander Trechsel, "First or Second Order Referendums? Understanding the Votes on the Constitutional Treaty in Four EU Member States," Manuscript, European University Institute (2007). 35 This dichotomy can be useful for accession countries, when they hold referenda on the question of joining the EU. It can be argued that these countries maintain a clear distinction between the domestic and European domains. 13

Moreover, the literature explaining the 2005 constitutional referenda results not only is inconclusive on the relative success of the two models but also suffers from a lack of in-depth comparative study of all four cases. These theoretical and methodological problems reflect those in the broader literature on EU referenda results. Few scholars proposed questions that could be evaluated theoretically and tested empirically in order to develop wider hypotheses about the general dynamics of referenda.36 Therefore to successfully explain the 2005 constitutional referenda results, the challenge is to develop a theoretical framework that encompasses all four cases, both the positive and the negative votes.

The Comparative Framework: Ruling out Existing Explanations Studying both positive and negative cases helps in assessing the relative value of variables used in the literature so far. In this section I show that as all four cases share important characteristics, some key arguments and variables in the literature can be ruled out. I consider the above-mentioned explanations in the same order as they were presented. First, political party-based studies of Euroscepticism and individual-level analyses of public support for the EU only focus on the broad public opinion towards the Union. As discussed, these approaches do not explain how these attitudes translate into EU referenda results. In other words, they do not take the dynamic campaigning process into account. Regarding the Euroscepticism school, Crum in his analysis on the positioning of political parties towards the TCE, finds that parties in the pro-integration ideological core – social democrats, liberals and Christian democrats – supported the TCE regardless of whether they were in or out of government.37 Thus, party opposition to the TCE was concentrated among the nationalist and far right, the far left and smaller protest parties. While this is in line with the findings of the Euroscepticism literature, this pattern exists in all

36 An example of this type of work on Eastern European countries can be seen in Aleks Szczerbiak and Paul Taggart, "The Politics of European Referendum Outcomes and Turnout: Two Models," West European Politics 27, no. 4 (2004). 37 Crum, "Party Stances in the Referendums on the EU Constitution: Causes and Consequences of Competition and Collusion," 19. 14 four cases. The only exception was the de facto division of the French center-left, as some members of the Socialist Party and the Greens have joined the No camp. While this division led to credibility problems for the Yes campaign and strengthened the No campaign, it does not reflect a shift to Euroscepticism on the part of these parties given their official pro-TCE campaign. Concerning the individual-level analysis of public support, comparative post-referendum Eurobarometer survey analyses highlight similarities among the four cases.38 Comparing the sociodemographic breakdown of the votes, the Luxembourg report shows that both the Yes and No voters share a well-defined profile in all four cases.39 While women and the elderly tended to vote Yes, most young people sided with the No faction. In terms of education, support for the TCE was particularly strong among those who have studied beyond the age of 21, while voters with less education rejected it. Furthermore, while the Yes vote was prevalent among the self employed, the No vote was mainly to be found among manual workers. The US Census Bureau International Database shows that all four cases have similar population pyramids and median ages, while Penn World Tables indicate that they have similar numbers on variables such as Purchasing Power Parity and GDP per capita.40 Thus the Eurobarometer analysis provides evidence for the socioeconomic determinants of the public support literature, however as this pattern is constant in all four cases, it does not explain the variation in the results. The other variables presented as determinants of individual-level support for the Union are national economic performance through indicators such as unemployment and inflation, or perception of threats posed by other cultures, and negative attitudes towards immigration. The four countries do indeed hold

38 Eurobarometer, "The European Constitution: Post-referendum Survey in Spain," European Commission: Brussels (2005). ———, "The European Constitution: Post-referendum Survey in France," European Commission: Brussels (2005). ———, "The European Constitution: Post- referendum Survey in the Netherlands," European Commission: Brussels (2005). ———, "The European Constitution: Post-referendum Survey in Luxembourg," European Commission: Brussels (2005). 39 Eurobarometer, "The European Constitution: Post-referendum Survey in Luxembourg," 9. 40 US Census Bureau International Database. available at: http://www.census.gov/ipc/www/idb/summaries.html; Penn World Tables. available at: http://pwt.econ.upenn.edu/php_site/pwt62/pwt62_form.php 15 different values on these variables. However, these values were at these same levels when the early public opinion polls showed positive attitudes towards the TCE in all four countries. This shows the importance of investigating the importance of these different values in relation to the referendum campaigns. Moving on to the explanations that explicitly consider referenda results, I presented two models: Second-order and issue voting. The literature provides evidence for the simultaneous presence of these factors, leading to inconclusive results. Thus scholars brought in new causal variables – institutional setting, party cues, issue salience and level of information, and campaign effectiveness – in order to answer the new question of under what circumstances are one of the two models more likely to explain the results. To start with the institutional setting argument, Hug argues that the legal character of referenda is an important determinant of the votes. In non- compulsory and binding referenda, government support among voters is expected to be higher. In 2005, all four cases held non-compulsory and binding referenda.41 Therefore the institutional design was not the source of variation. As for the party cues argument, the attitudes of the political parties towards the TCE were very similar in all four cases. The mainstream parties – social democrats, liberals and Christian democrats – supported the TCE regardless of whether they were in or out of government.42 Thus the opposition was confined to the extremes of the political spectrum in all four cases, which all had low vote shares in the parliament. Hence the existence of a similar pattern in all cases suggests that party identification or cues did not cause the variation in results on their own. Rather, party cues became important in relation to the campaign. Another variable used widely in the existing explanations is issue salience and the level of information. Voters are more likely to rely on their attitudes towards the EU if the salience is high and they have knowledge of European

41 While the referendum was legally binding in France, it was not binding constitutionally in the other three cases. In all three, however, prime ministers or the parliament promised to respect the results of the referenda. Arsène Richard and Ronald Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," Democracy International Monitoring Reports (2005). 42 Crum, "Party Stances in the Referendums on the EU Constitution: Causes and Consequences of Competition and Collusion," 19. 16 affairs. However the TCE had an all-encompassing character, as it was a draft ‘Constitution’ including all the legislation passed during the last 50 years of integration. This makes it very difficult to discuss the salience of the issue at hand. Previous referenda on the euro, accession, or even Treaty amendments presented a specific dimension of the integration project. The referenda on the TCE, on the other hand, covered everything concerning the EU. The pro and anti- EU arguments used during the campaign reflected this, as they covered almost every aspect of the Union. Therefore, I argue that issue salience, in such referenda, is not a relevant variable. Instead of the salience variable, one could look at the level of general interest in EU affairs because of the broad nature of the TCE. A possible proxy would be the turnout rate in the 2004 European Parliament elections, as this rate is an important indicator of institutional legitimacy and the public’s perception of the Union’s relevance. All cases except Luxembourg (due to compulsory voting) had similar turnout rates in the 2004 elections: Spain 45.1%, France 42.76%, and the Netherlands 39.3%. This implies that there was no major variation concerning the general level of interest in the EU. The level of information, on the other hand, is vague. In a previous report, the Eurobarometer conducted a test on the knowledge of the TCE, asking the respondents if six statements presented about the TCE were true or false. The results show that the Spanish and French were the least well informed among the four.43 Thus the level of information variable does not have clear implications, and does not account for the variation in the vote. Lastly, the literature has discussed the intensity of campaigns. While campaigns are crucial, no work in the literature has conducted a thorough and comparative analysis of campaign dynamics and strategies. A common indicator used in the literature is the timing of vote decision, showing the extent to which an opinion formation process takes place during the campaign. The Eurobarometer provides data on when the decision on how to vote was taken in the 2005 referenda. The percentage of the voters that decided ‘at the time the

43 Eurobarometer, "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution," European Commission: Brussels, November (2004): 6. 17 referendum was announced’ is fairly low and similar across all cases, ranging from 26% in the Netherlands to 35% in Spain. The rest of the voters answered that they decided either ‘fairly early during the campaign’, ‘in the final weeks of the campaign’ or ‘on the day of the referendum itself’. This shows that in all cases the campaigns played a role in voters’ decisions. In other words, there was no case where the campaign did not have any influence. However, the campaigns were different in terms of intensity, dynamics, strategies, and argumentation. These factors have not been considered carefully and comparatively in the literature. In the next chapter, I provide a detailed analysis of campaign strategies and dynamics to understand the full impact of the campaigns. The literature has also overemphasized the impact of unpopular governments on the No votes. While the governments in France and the Netherlands were indeed unpopular, they were equally unpopular when the early opinion polls showed positive attitudes towards the TCE. I argue that this variable needs to be operationalized carefully and more precisely in relation to the referendum campaigns. Moreover, unpopular government policies form only a subset of contentious issues in the society. An example can be seen in the Luxembourg vote. In Luxembourg political life is organized around the invincible Christian Democratic Party, which has been in power almost continuously since 1945, choosing alternatively the socialists or the liberals as coalition partners. Yet, despite the absence of an unpopular government, the No vote reached 43%. Therefore, a sound theoretical explanation of the results requires deconstruction of ‘dissatisfaction with incumbent governments’ to understand how these aspects played out in the campaign. This section has demonstrated the significance of a comparative framework in ruling out or deconstructing certain prominent variables in the existing literature. None of the current explanations can account for the results of the 2005 constitutional referenda in their entirety. Therefore, given that all four cases started with a favorable public opinion towards the TCE, and that the sociodemographics of the vote and political party preferences were similar in all, why did the public in France and the Netherlands reject the TCE while those in

Spain and Luxembourg did not? Why was the positive attitude sustained in Spain and Luxembourg but not in France and the Netherlands?

The Argument Abandoning the false dichotomy between domestic and European factors, I propose a dynamic explanation that captures the interaction between the two domains. What needs to be specified is not the circumstances under which attitudes towards domestic or European issues gain the upper hand but instead how these two fields come together in forming people’s attitudes towards the integration project. I argue that successful No campaigns built a connection between existing contentious issues (at both the domestic and EU levels) and the referendum proposal, claiming that these problems were caused or would be exacerbated by the proposed European treaty. The existing contention remained unlinked to the referendum proposal in the absence of effective No campaign framing. Furthermore, sequencing mattered as the cases were not independent of each other. Later campaigns were exposed to previous ones and thereby were not created equal. Diffusion effects conditioned campaign strength under certain circumstances.

Hypothesis 1: The stronger a state's No campaign relative to its Yes campaign, the greater the magnitude of the No vote.

Hypothesis 2: Under certain conditions (2a and 2b), referenda campaigns in first-mover states can amplify the strength of referenda campaigns in second- mover states. 2a: The stronger the diffusion channels between first-mover and second-mover states, the more the second-mover campaigns will be affected; and 2b: The higher the campaign intensity in the first-mover state, the greater its influence on second-mover campaigns.

Below I discuss the dependent, independent, and conditioning variables in turn, providing the definition and measurement of the key terms.

Dependent Variable: The Magnitude of the No Vote The dependent variable is not simply the results of the 2005 TCE referenda (ratification vs. non-ratification), because categorizing the cases as ratification vs. non-ratification conceals the significant differences among the results. Instead, my dependent variable encompasses both the differences in the magnitude of the No vote among the cases and the relative movement of public opinion in each case. In terms of magnitude, the percentages of the No votes were: Spain 17.24%, France 54.68%, the Netherlands 61.5%, and Luxembourg 43.48%. In terms of public opinion, I compare the percentage opposed in fall 2004 with the eventual No vote totals: in Spain +13%, in France +23%, in the Netherlands +50%, and in Luxembourg +26%.

Independent Variable: Relative No Campaign Strength What constitutes a strong campaign? As I will discuss in detail in the next chapter, a campaign is stronger if it engages in strategic communications (agenda- setting and priming/framing) and can back its political rhetoric with credible sources and powerful mobilizational structures. I argue that the stronger the No campaign relative to the Yes campaign, the more successfully the No campaigners could link contentious issues to the European Constitution and thus increase the magnitude of the No vote. Because the TCE was extremely complicated, public opinion was highly susceptible to campaign messages. While the literature tends to define the campaign effect based on the last intense period of a campaign, I instead focus on the long campaign starting from fall 2004 in order to understand what changed in public opinion so drastically in the Netherlands and France, less so in Luxembourg, and barely at all in Spain. Why is the independent variable defined based primarily on the No campaign rather than the Yes campaign? There are two good reasons to pay special attention to the strength and strategies of the No campaign. First, the

20 positive initial opinion polls meant that in the absence of a strong No campaign, the Yes campaigners had an easy task, as the Spanish case illustrates. Second, the literature suggests that the political advantage rests with the No side in a referendum campaign. No campaigners do not necessarily need to make a coherent and persuasive case against a proposal.44 Rather, it is sufficient for the No campaign to raise doubts in the minds of voters, play upon known fears, or link the proposal to other less popular issues or personalities.45 Therefore, in cases where No campaigns were especially strong – France and the Netherlands – this was an important obstacle for the Yes campaign. In essence, the Yes campaigners’ job in referenda is essentially defined by the No campaigners, so their strategies and campaign should be assessed in relation to that of the No campaign.

Conditioning Variable: Diffusion This variable aims to identify the impact of sequencing and networking on the referenda campaigns. Politicians do not engage in priming and framing in a vacuum. They tailor campaign strategies not only to the parameters of public opinion but also to the strategic opportunities offered by the political conditions of their time.46 I argue that the sequencing of the 2005 TCE referenda should be taken into account as the later cases were exposed to the previous campaigns and their results. Defined broadly, diffusion is the process by which an innovation is communicated through certain channels over time among the members of a social

44 LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," 179. LeDuc explores the debate in the 2000 Danish and 2003 Swedish referenda on the Euro, and argues that the no side holds the advantage. See also Jennifer Jerit, "Survival of the Fittest: Rhetoric during the Course of an Election Campaign," Political Psychology 25, no. 4 (2004). In her research on the framing of the Canada-US free trade agreement in the 1988 Canadian federal election, she emphasizes that the proponents of the policy had the burden of explaining why the agreement was worth supporting, while the opponents could appeals to anger or fear. While the proponents pointed out the dire consequences of failing to sign this treaty, they also had to convince the electorate that there were real economic gains to be had. 45 LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," 191. 46 James Druckman, Lawrence Jacobs, and Eric Ostermeier, "Candidate Strategies to Prime Issues and Image," Journal of Politics 66, no. 4 (2004). See also Wiliam Riker, The Strategy of Rhetoric: Campaigning for the American Constitution (New Haven: Yale University Press, 1996). I do not present their findings on candidate image priming here as in a referendum campaign this aspect is not applicable. 21 system.47 More specifically, it is a process wherein new ideas, institutions, policies, models or repertoires of behavior spread geographically from a core site to other sites.48 As such, all campaigns were not created equal. This diffusion effect did not have an independent influence on the referendum results; instead it could condition the magnitude of the No and Yes campaign strength in the second-mover states. I argue that the later a country held its referendum, the more likely it was to be influenced by the previous referenda results and campaigns. The campaign strategies in the second-mover countries are partly shaped by the networking and learning processes across cases. Nevertheless, diffusion is not automatic and needs specific channels to work. Where transmission belts existed between the cases, such as shared language/culture, common media sources, and collaborative networks/transnational linkages, the earlier cases influenced the later ones. Campaign intensity also matters. The campaign needs to be intense in order to instigate transmission of such influence through the existing diffusion channels. Campaign intensity is measured as a combination of partisan polarization, perceived closeness of the race, and news coverage.49 Where these conditions were met, diffusion amplified the campaign strength directly. Spain held the first referendum on 20 February, followed by France on 29 May, the Netherlands on 1 June, and Luxembourg on 10 July. Spain did not have an intense campaign or public debate as the No campaign was virtually non- existent. Therefore it sustained the initial positive public attitude towards the TCE. Having a rather low-profile and quiet vote, Spain did not have a significant impact on the following cases. France, on the other hand, had the most intense

47 Everett Rogers, Diffusion of Innovations, Fourth ed. (New York: New York: Free Press, 1995). 48 Valerie Bunce and Sharon Wolchik, "International Diffusion and Postcommunist Electoral Revolutions," Communist and Post-Communist Studies 39, no. 3 (2006): 286. 49 I adopt Sara Hobolt’s definition, see Hobolt, Europe in Question: Referendums on European Integration. In her formulation, partisan polarization is the percentage of parties who recommend a no vote in the referendum weighted by these parties’ share in the parliament, thus an estimation of the opposition to the ballot proposal in parliament. The perceived closeness of the race is the average difference between intended Yes and no campaigners in the opinion polls, in the last six weeks leading up to the referendum. Finally, the news coverage is the number of daily articles monitoring the referendum issue during the three months leading up to the referendum in the two mainstream national daily newspapers. 22 campaign and public debate among the four cases. The French campaign influenced the other campaigns, to a great extent in Luxembourg and to a minor extent in Spain and the Netherlands.

Analytical Framework

Relative No campaign strength Referendum result (Magnitude of the No vote)

Diffusion effects

Figure 1.1: Analytical Framework

The 2005 TCE referenda took different forms depending on the interaction between campaign strength and diffusion effects. Luxembourgish campaigners were heavily influenced by the French example in running their campaign. In contrast, Spain, France and the Netherlands were only minimally influenced by the other ones.50 Strikingly, because the intense French debate pre-dated the Spanish campaign, even Spain was exposed to the French argumentation on the TCE. Figure 1.2 illustrates the diffusion effects based on campaign intensity and strength of the diffusion channels.

Spain France The Netherlands Luxembourg

Figure 1.2: Diffusion effects among the 2005 TCE referenda

50 While the French campaigners were primarily the source of diffusion and not the receiver, I am assuming low diffusion rather than none, as their heavy interaction with the other campaigners might have had a minimal impact on them. 23

In sum, applied to the 2005 TCE referenda, this analytical framework leads to the following distribution of cases:

High Diffusion Low Diffusion

Strong No Campaign France The Netherlands Weak No Campaign Luxembourg Spain

Table 1.1: The interaction between campaign strength and diffusion

Cases My case selection reflects the universe of cases that held constitutional referenda in 2005: Spain, France, the Netherlands, and Luxembourg. So far, most work on the subject has only looked at the No vote cases, either individually or comparatively. Considering both positive and negative votes allows for variation on the dependent variable, rendering the causal inference more reliable.51 Furthermore these four cases are comparable, similar in a large number of important characteristics but dissimilar with regard to the study variables.52 The initial reaction of the public to the TCE was positive in each case several months before the referenda. Moreover, the No campaigns were supported only by the extremes of the political spectrum, and the sociodemographic profiles of the Yes/No voters were similar in all cases. Also the four cases shared very similar institutional structures, and held non-compulsory and binding referenda. Similarly, the call for referenda opened up the political space for political parties and civil society groups to campaign in the same way in all the cases. In other words, no political actor faced political constraints, which blocked them from campaigning. Nevertheless, the strength and the intensity of the campaigns varied across cases. Because the campaign intensity was very low in the Spanish case, this case is particularly important as a control case. Below I will briefly introduce

51 See Gary King, Robert Keohane, and Sidney Verba, Designing Social Inquiry: Scientific Inference in Qualitative Research (Princeton: Princeton University Press, 1994), 129. 52 Arend Lijphart, "The Comparable-Cases Strategy in Comparative Research," Comparative Political Studies 8, no. 2 (1975): 163. 24 each case, focusing on the initial poll results and the composition of the Yes and No campaigns. Spain held the first referendum on 20 February. Four months before the referendum, in October 2004, a Center of Sociological Institutions (CIS) survey showed that the public was in favor of the TCE.53 In response to the question ‘How would you vote if the referendum was tomorrow?’, 44.6 % were in favor while only 4% were against the TCE. Similarly in the same survey, when asked whether they would like the February referendum to approve the TCE or not, 56.7% said Yes and only 5.5% said No. As in all cases, the Yes side was formed by the political mainstream in Spain. The Yes camp parties consisted of the governing Spanish Socialist Workers’ Party (PSOE), the main opposition party Popular Party (PP), the Catalan Nationalist Convergence and Unity (CiU), the Basque Nationalist Party (PNV), and the Canary Islands liberal nationalist party Canarian Coalition (CC). The No camp was limited to the United Left (IU) at the national level, but at the regional level the Republican Left of Catalonia (ERC), the Initiative for Catalonia Greens (ICV), the Basque nationalist social democratic Basque Solidarity (EA), the Galician Nationalist Block (BNG), and the Aragonese Council (CHA) also campaigned for a No vote. Civil society support was minor on both sides of the campaign. Table 1.2 shows the actors who actively campaigned in the Spanish case.

53 Available at: http://www.cis.es/cis/opencms/-Archivos/Marginales/2560_2579/2577/Es2577.pdf 25

Spanish Yes Campaign Spanish No Campaign

Political Parties Civil Society Political Parties Civil Society

Spanish Socialist Employer United Left (IU) The left-wing ‘No to Workers’ Party organizations and the Constitution’ (PSOE) the trade unions platform: ATTAC, Social Forum, Ecologists in Action, Alternative Space, Red Current, and other left-wing groups Popular Party (PP) Republican Left of Catalonia (ERC) Convergence and Initiative for Unity (CiU) Catalonia Greens (ICV) Basque Nationalist Basque Solidarity Party (PNV) (EA)

Canarian Coalition Galician Nationalist (CC) Block (BNG) Aragonese Council (CHA) Table 1.2: The active campaigners in the Spanish case

The French referendum was the second one, and took place on 29 May. Eight months before the referendum, in September 2004, the Advice on Survey Analysis (CSA) found that 69% of the population was in favor of the TCE, while only 31% were against.54 Following the pattern in the other cases, the French Yes camp was formed by the political mainstream. The political parties on the Yes side were the Union for a Popular Movement (UMP), the Union for French Democracy (UDF), the Socialist Party (PS), and The Greens. The No campaigners were confined to the far right and far left, as in all other cases. The National Front (FN), the Movement for France (MPF), the French Communist Party (PCF), and the Revolutionary Communist League (LCR) campaigned for a No vote. Importantly, the French No campaign included key figures from three Yes camp

54 Available at: http://www.csa.eu/dataset/data2004/opi20040915c.htm 26

political parties, i.e. the Socialist Party, the Greens, and the UMP. This led to splits inside the Socialist Party and the Greens, which I will discuss specifically in the French case chapter. Regarding the contribution of civil society, France had the strongest mobilization particularly among the left-wing No camp. On the yes side, civil society support was considerably weaker. Table 1.3 lists the active campaigners in the French case.

French Yes Campaign French No Campaign

Political Parties Civil Society Political Parties Civil Society

Union for a Popular Employer National Front (FN) The left-wing No Movement (UMP) organizations and Platform: Aside from trade unions that local offices of the are not mentioned left-wing political in the No campaign parties, Copernic list Foundation, ATTAC, three trade unions – General Workers’ Confederation (CGT), the Workers’ Force (FO), Solidarity Unity Democracy (SUD), and other left-wing groups Union for French New Republic, Movement for France The Gaullist ‘Group Democracy (UDF) Europa Nova, the (MPF) for a Confederation of Robert Schuman the States of Europe to Foundation Say No’

Socialist Party (PS) Various initiatives French Communist by different Party (PCF) occupation groups such as business leaders or legal experts The Greens Revolutionary Communist League (LCR) Table 1.3: The active campaigners in the French case

The Netherlands held its referendum on 1 June. Seven months before the referendum, in November 2004 the special Eurobarometer showed that 63% of the Dutch supported the TCE, while those against were limited to 11%.55 Furthermore, Marketresponse found that 61% of the population was still in favor of the treaty in January 2005.56 As in the other cases, the political mainstream campaigned for a Yes; the Christian Democratic Appeal (CDA), the People's Party for Freedom and Democracy (VVD), the Democrats 66 (D66), the Dutch Labor Party (PvdA), and the GreenLeft (GL) were all in favor. The No campaigners were the far right and far left political parties and the small Orthodox protestant parties; the List Pim Fortuyn (LPF), the Socialist Party (SP), the Christian Union (CU), and the Reformed Political Party (SGP). Here, both the Yes and No campaign had limited civil society support, and these groups did not run substantial campaigns. The full list of active campaigners can be seen in Table 1.4. Dutch Yes Campaign Dutch No Campaign

Political Parties Civil Society Political Parties Civil Society

Christian Employer List Pim Fortuyn The left-wing Democratic Appeal organizations and (LPF) ‘ConstitutionNo’ (CDA) trade unions group: ATTAC and other left-wing groups People's Party for Foundation for a Socialist Party (SP) Young Fortuynists, Freedom and Better Europe youth organization of Democracy (VVD) the LPF

Democrats 66 (D66) Christian Union (CU)

Labor Party (PvdA) Reformed Political Party (SGP) GreenLeft

Table 1.4: The active campaigners in the Dutch case

55 Eurobarometer, "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution." 56 Paul Dekker and Charlotte Wennekers, "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen," Internationale Spectator 59, no. 12 (2005). 28

The last referendum was held in Luxembourg on 10 July. The Luxembourg Institute for Social Research (TNS-ILRES) surveys show that in November 2004, eight months before the referendum, 64% of the public was in favor of the TCE, while only 17% was against.57 Paralleling the pattern in the previous cases, the Yes campaigners in the Luxembourgish case were the mainstream political parties; the Christian Social People's Party (CSV), the Democratic Party (DP), the Luxembourg Socialist Workers' Party (LSAP), and The Greens. Once again, the No campaign was limited to the extremes of the political spectrum. The Alternative Democratic Reform Party (ADR), The Left and the Communist Party (KPL) campaigned against the Treaty. In this case both the Yes and No campaigns had civil society contribution but the No campaign civil society actors were far more active. Table 1.5 lists the active campaigners in Luxembourg.

Luxembourgish Yes Campaign Luxembourgish No Campaign

Political Parties Civil Society Political Parties Civil Society

Christian Social Employer Alternative The left-wing ‘No People's Party (CSV) organizations and Democratic Reform Committee’: ATTAC, trade unions Party (ADR) National Union of Luxembourgish Students (UNEL), and other left-wing groups Democratic Party Various initiatives The Left (DP) by different occupation groups Luxembourg The Communist Party Socialist Workers' (KPL) Party (LSAP)

The Greens

Table 1.5: The active campaigners in the Luxembourgish case

57 Available at: http://www.tns-ilres.com/cms/Home/Nos-secteurs/Politique---Opinion 29

The cases share important similarities concerning the early public opinion polls, the sociodemographic breakdown of the Yes/No voters, and the political party preferences. As seen in the tables, the extent to which the civil society participated in the campaigns differed across cases, but the type of the organizations involved was similar in all. The cases were different in terms of campaign intensity, dynamics, strategies, and argumentation. I will discuss the campaigners and their strategies in detail in the individual case chapters.

Methodology To carry out the proposed research, I systematically collected data on the same variables across all four cases in order to answer theoretically guided questions.58 While case studies have been criticized for problems of generalizability, I sought to put qualitative flesh on quantitative bones by using quantitative data as a starting point for framing my qualitative study.59 I used the Eurobarometer survey data on background variables such as the sociodemographics of the Yes/No vote, the level of public information, and when the decision on how to vote was taken. Moreover, my research design combines qualitative (interview data) and quantitative (public opinion and media content data) methods. The first step in my research was to show the similarities in the background conditions across the four cases using the Eurobarometer reports. I used the literature on positioning of the political parties and the vote shares of the No camp parties in the national parliaments to assess the political opportunity structure in each case.60 Furthermore, I examined the legal framework for election/referendum campaigns in each case to understand the resources available to Yes and No camps, such as the funds granted and the broadcasting time allowed to both camps.

58 Alexander George, "Case Studies and Theory Development: The Method of Structured, Focused Comparison," in Diplomacy: New Approaches in History, Theory, and Policy, ed. Paul Gordon Lauren (New York: Free Press, 1979). 59 Sidney Tarrow, "Bridging the Quantitative-Qualitative Divide in Political Science," The American Political Science Review 89, no. 2 (1995): 473. 60 Increasing access, shifting alignments, divided elites, influential allies, repression of dissent are all important dimensions of this structure. 30

I used both the Eurobarometer and national public opinion polls to show that the public in all four countries held positive attitudes towards the TCE between September and November 2004. Next, for each individual case, I mapped the contentious themes approximately between 2002 and 2004, leading up to the polls in fall 2004. These themes can be political, economic or social. I paid special attention to recent welfare state reforms, inflation, unemployment, share of foreigners in each country, and changes in the perceived threat from ethnic minorities. To discern these issues, I used national public opinion polls (monitoring the most important problems facing the country), real-world indicators (of important issues such as official rates of crime or unemployment), government reports, and secondary sources. Once I completed this background analysis, I moved on to the analysis of campaigns. I used three sequential methodological steps: 1. Interviews with political elite and campaign managers to discern campaign strategies and frames. I also explored campaign materials systematically. 2. Media content analysis to observe the extent to which the campaign messages were picked up by the news media. 3. Public opinion data analysis to examine the extent to which the public echoed the campaign messages. These are the three essential steps to observe the impact of a political campaign. The literature also identifies a feedback loop where public opinion expressed through the outcome influences the shaping of party communication strategies.61 Below, I discuss the significance of each step in detail.

Interviews and campaign materials I conducted my field research in 2008, first in the Netherlands and Luxembourg in April 2008, and then in all four countries between September and December 2008. In each country, I interviewed campaigners from both political parties and the civil society groups that were active in the campaign. Overall I conducted 96

61 Pippa Norris et al., On Message: Communicating the Campaign (London: Sage, 1999). 31 in-depth interviews, using both opportunity and snowball sampling. These interviews were face-to-face and semi-structured as this method provides more room for flexibility to probe for details.62 Interviews are a very important starting point, to understand the campaign strategies fully. The existing literature on campaign effects mostly starts from the second step, the media analysis, as conducting interviews can be time-consuming and difficult, especially if the campaign took place several years ago.63 However, the modern campaign has a complex and increasingly fragmented news environment, which may not be captured fully by a regular media content analysis. Personal conversations, direct contact with parliamentary candidates, election coverage in local and regional press, party propaganda such as leaflets/posters, party election broadcasts, and the use of the internet as a new venue are all important in modern campaigning.64 The local campaign involving local activists canvassing and leafleting started to be seen as significant for voter mobilization.65 Because of the existence of all these various venues, it is critical to take a step back from the media analysis and interview campaigners. This step provides information on local level campaign activities and organization. Furthermore, it helps the researcher capture campaign frames more accurately, as most media content analyses are not sufficiently sophisticated to spot frames in their full extent. Most studies track the tone and topic of the news rather than the complete framing used. Interviews contributed to the evaluation of campaign effectiveness as well, by posing questions on party strategies in agenda-setting and campaign communications. Moreover, I explored diffusion effects, asking the campaigners

62 Only a few of my interviews were telephone interviews, due to scheduling problems. I conducted some of these interviews after I returned to Canada, in January 2009, as some of my interviewees were still referring me to other important contacts on the subject. 63 There are exceptions. For examples of studies that include interview data see de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue.; Druckman, Jacobs, and Ostermeier, "Candidate Strategies to Prime Issues and Image."; Hobolt, "How Parties Affect Vote Choice in European Integration Referendums."; Richard Nadeau, Francois Petry, and Eric Belanger, "Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election," Political Communication (2010). 64 Norris et al., On Message: Communicating the Campaign, 11. 65 Ibid., 12. 32 questions on the connections between them and their counterparts in the other cases, and on whether they were influenced by other countries in the formulation of their campaign frames and strategies. I also collected campaign materials from every campaigner. This helped me complement my interviews with an analysis of campaign materials to explore the presentation of the campaign frames to the public. It is important to analyze whether this presentation reflects the messages in a clear, coherent, concrete, vivid, image-provoking, and emotionally compelling manner. In each case, I identified the campaign frames primarily based on written campaign materials. Where available, one to two page leaflets listing the campaign arguments provided the best source. If not, I used interview data and secondary sources to complement the available materials.

Media content analysis Content analysis provides systematic maps of the contents of campaigns to understand the messages received by the public. Media content data can provide insights on many different dimensions. There are aggregate level studies focusing on the composition of the news agenda, as well as other studies looking into the influence of media exposure at the individual level.66 In my research design, I focused on four aspects of the aggregate level: The topical distribution of media attention, the time devoted to different political actors, the tone of the news concerning them, and finally, if available, the attention spent on the campaigns in other countries.67 In addition, where the analyses allow, I also tracked how these factors changed over time during the campaign. During my field work, I collected media content analyses for each case except Luxembourg. For Spain, I use the analyses conducted by Berganza, Capdevila, and Gómez Puertas, as well as a Royal Elcano Institute study by

66 See de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue, 110. 67 Norris et al., On Message: Communicating the Campaign. They warn that patterns of media use are related to other factors such as education and class, which are associated with political attitudes. Moreover, it is also difficult to disentangle media sources as there may be little variation among mainstream sources or overlapping uses of different media. 33

Sampedro Blanco, Carriço Reis, and Ruiz Jiménez.68 For France, I use Gerstlé's analysis.69 For the Netherlands, I use the analyses of Schuck, de Vreese, Kleinnijenhuis, Takens and van Atteveldt.70 There is no specific media content analysis of the Luxembourgish case but because Luxembourg shares many of its media sources with France and Germany, the French media content analysis accounts for the Luxembourgish case to a certain extent.71 To make up for this loss, I interviewed both the director of the national television station Luxembourg Television and Radio (RTL), and the director of the national radio station in Luxembourg.

Public opinion data Similarly, during my field work, I collected public opinion data from national survey institutions as well as data and analyses from scholars working on the subject in all four countries. For the Spanish case, I used regular surveys from the Center of Sociological Investigations (CIS) and the Real Instituto Elcano, and analyses by Font and Rodriguez. For France, I employed data from survey institutions such as the Advice on Survey Analysis (CSA), the IPSOS, the French Institute of Public Opinion (SOFRES), and the French Institute of Public Opinion (IFOP) and analyses by the Centre for Political Research at Sciences Po (CEVIPOF) led by Brouard and Sauger. For the Dutch case, I used survey and focus group data from the Marketresponse, the Netherlands Bureau for Economic Policy Analysis (CPB), the Netherlands Institute for Social Research (SCP), and

68 Rosa Berganza, "Media Framing of the Spanish Referendum on EU Constitution," in International Communication Association (Dresden, Germany2006).; Víctor Sampedro Blanco, Antonia María Ruiz Jiménez, and Bruno Carriço Reis, "El Referendo del Tratado de la UE en la Prensa Española de Referencia: A Favor de la Constitución o en Contra de la Unión Europea," Elcano Royal Institute Working Paper 20 (2005). 69 Gerstlé, "The Impact of Television on French Referendum Campaign in 2005." 70 Schuck and de Vreese, "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign."; Andreas Schuck, "Referendum Campaign Dynamics: News Media, Campaign Effects and Direct Democracy" (University of Amsterdam, 2005), 14.; Jan Kleinnijenhuis, Janet Takens, and Wouter van Atteveldt, "Toen Europa de Dagbladen Ging Vullen," in Nederlanders en Europa: Het Referendum over de Europese Grondwet, ed. Kees Aarts and Henk van der Kolk (Amsterdam: Utigeverij Bert Bakker, 2005).; Janet Takens, "A Negative Campaign? A Study of the News Coverage of the European Constitution and its Effects" (Free University of Amsterdam, 2006). 71 Luxembourg has one national TV channel – RTL, which airs daily from 6pm to 8pm. Residents receive the rest of their TV channels from France and Germany directly. 34 data and analyses by Aarts and van der Kolk. For Luxembourg, I employed Luxembourg Institute for Social Research (TNS-ILRES) survey and focus group data, and analyses by the University of Luxembourg led by Dumont, Fehlen, Kies, and Poirier. In addition, I used Eurobarometer surveys and analyses in all four cases. Here, I focused on three aggregate-level components: Shifts in voters' intentions over time, the reasons behind voting decisions, and knowledge of the results of previous referendum campaigns.72 Public opinion polls showing aggregate level shifts in vote intentions over time are critical to assess the course of the campaign, and to detect the difference in attitudes on the same topic before and after the campaign. Chong and Druckman argue that the best method to measure framing effects is to compare each frame condition to a control condition in which participants express their preferences without having been exposed to a persuasive communication.73 These opinions expressed in the neutral control condition provide a baseline reference against which to judge the impact of framed conditions. For my research design, the public opinion data is helpful in this regard. Public opinion data is also useful to capture the match between the campaign frames and voters’ self-provided answers in explaining their vote choice. Eurobarometer surveys provide data on motivations behind the vote choice. Nevertheless, I prefer to rely more heavily on national public opinion polls to explore these motivations for several reasons. First, the national opinion polls are more detailed and use more than one type of question to investigate the reasons for voting Yes/No. Second, the Eurobarometer data on the motivations

72 The aggregate level data I use for all three components have limits as aggregate-level analysis makes strong assumptions about what is going on at the individual level. The literature warns that aggregate-level correlations may be misleading as the parallels over time may be independent or the association may be spurious. The literature has also been using experiments to test assumptions on causal mechanisms, whereas the panel studies show whether these causal mechanisms existed in the actual campaign. While my research is not based on individual-level – regression – analysis, such research exists and has produced valuable insights but with mixed results. I am providing a different perspective to highlight some key aspects, to which the literature has not paid attention so far. 73 Dennis Chong and James Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," Journal of Communication 57, no. 1 (2007): 106. However they discuss this issue for experimental designs, and suggest using a control group in the experiments. 35 behind voting behavior for the Spanish case is extremely brief. Third, the literature has pointed to diverging results provided by the Eurobarometer and national polls.74 While the Eurobarometer reported that only 3% of the Dutch No- voters said No due to the possible accession of Turkey to the EU, a national poll taken on the day of the referendum found that 22% of the No-voters said No because of this issue. Focusing on three referenda cases, the SCP concludes that the available Eurobarometer survey data does not allow an explanation for why the vote became a Yes in Spain but a No in the Netherlands and France.75 It states that general preferences, knowledge and involvement do not provide an explanation, and suggest that data on these are limited in explaining the formation of public opinion and electoral behavior regarding specific European issues that attract public attention and become politicized.76 Finally, public opinion data sheds light on diffusion processes. Some polls include questions on the knowledge of the results of the previous referendum campaigns, which is helpful to compare diffusion effects across the cases. My research framework, which can be applied to all EU referendum campaigns, shows the importance of analyzing the long campaign to understand the opinion formation process fully. Individual-level analyses have highlighted the importance of anti-immigrant sentiments, pessimistic economic outlooks, unpopular governments, EU-scepticism (measured as support for the Euro, Turkish EU membership and the speed of EU enlargement), national identity, and fear of globalization. Nonetheless they do not pay attention to the values these variables carried before the long campaign started. Moreover, they usually focus only on the last intense phase of two-three weeks leading up to the referendum. My research shows that the initial public reaction to the TCE was positive several months before the referenda in all four countries. Yet at this point in time, the relevant, soon-to-be salient contentious issues already existed, as I show in detail with various national public opinion sources. While in Spain and Luxembourg

74 Besselink, "Double Dutch: The Referendum on the European Constitution," 346. 75 Paul Dekker and Sjef Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained (The Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis 2005), 7. 76 Ibid., 20. 36 these positive attitudes were sustained, the public in France and the Netherlands faced a sudden and radical opinion change over the course of the campaign. As the next chapter will argue more fully, in the absence of active agenda- setting, priming and framing by the No campaigns, contentious issues remained unlinked to the TCE. Therefore a careful analysis of contentious issues in relation to the campaigns is necessary. These concerns remained dormant until strong No campaigns tapped into them. A detailed campaign analysis also demonstrates the false nature of the dichotomy in the literature; the referendum results are not caused by either domestic or European factors. Campaign frames did not strictly distinguish the European domain from the domestic one. On the contrary, successful no campaigns linked the existing contentious issues (at both the domestic and the EU levels) to the referendum proposal, stressing that these problems were caused by the proposed European measure.

Structure of the Dissertation The rest of the dissertation will be structured as follows. In the second chapter I will present my theoretical framework, bringing together the public opinion, social movements, and diffusion literatures. In the following chapters I will discuss the four referenda cases, Spain, France, the Netherlands, and Luxembourg, in chronological order. Each case chapter will provide in-depth empirical evidence on the referendum campaign dynamics and the evolution of public opinion. Finally, the concluding chapter will summarize my findings, discuss the contribution of this research to the EU referenda and broader political science literatures, and specify the policy implications for the future of European integration.

CHAPTER 2 ANALYZING EU REFERENDUM CAMPAIGNS

Discussing the challenge of the French and Dutch negative votes to European studies, Taggart argues that we need to integrate the study of domestic politics with the study of the European integration process more systematically, with sensitivity to the growing interdependence between the two domains.77 The challenge is to not treat the EU as an exogenous factor in domestic politics and to not consider domestic politics as exogenous in the process of European integration.78 I respond to this call by integrating EU and domestic factors in a single model. In this chapter I build my theoretical framework using the public opinion, social movements and diffusion literatures. In the first section, I discuss the independent variable, relative No campaign strength, by focusing on strategic communications in both public opinion and social movements literatures. Using the findings of these literatures, I define campaign strength and then highlight the peculiarities of referendum campaigns to show how political campaigns have a greater impact in referenda than in regular elections. Based on this literature review, I develop the analytical framework of the independent variable. In the second section, I present the conditioning variable, diffusion. Outlining the mechanisms of diffusion, I demonstrate that under certain circumstances diffusion effects could condition campaign strength. At the end of the chapter I present a brief summary of the case findings.

RELATIVE NO CAMPAIGN STRENGTH The literature I presented in the previous chapter specifies important variables such as the determinants of people’s general attitudes towards the EU, positioning of the political parties, or the popularity of the incumbent government. But this literature does not study these variables in relation to political campaigns,

77 Paul Taggart, "Keynote Article: Questions of Europe - The Domestic Politics of the 2005 French and Dutch Referendums and their Challenge for the Study of European Integration," Journal of Common Market Studies 44, no. 1 (2006): 8. 78 Ibid.: 21. 38 neglecting how they interact with the campaign process. Below, I first explore the study of political campaigns in the public opinion literature, focusing on strategic communications: agenda-setting, priming, framing effects. I then present the study of framing in the social movements literature. When combined, these two literatures help us understand campaign strength and dynamics, and show precisely how variables such as party cues or dissatisfaction with the government may shape the campaign. In the following sections I apply the findings of these literatures to the EU referendum campaigns specifically.

Campaigns in the Public Opinion Literature Three schools of thought have studied the influence of political communications on public opinion; pre-war theories of mass propaganda, post-war theories of partisan reinforcement, and recent theories of cognitive, agenda-setting, and persuasion effects.79 The first phase focused on mass communications in the 1920s and 1930s, explaining that the public was easily swayed by propaganda on the radio and in the newspapers.80 In the second wave, after the Second World War, more systematic studies were conducted with the development of modern techniques of surveys. These studies found that the main impact of campaigns was reinforcement and not change, since partisans were strengthened in their voting choice.81 Overall, the literature in this wave stressed the minimal consequences of campaigns. It found that citizens’ opinions on political issues were largely unstable, superficial, uninformed, inconsistent, and unconnected from abstract principles and values.82 To explain fluctuations in public opinion, scholars first looked at the role of party attachment and class, and later, after the partisan and social dealignment in the 1970s, focused on the impact of aggregate-

79 Norris et al., On Message: Communicating the Campaign. For a detailed overview of the study of campaign effects see Henry Brady, Richard Johnston, and John Sides, "The Study of Political Campaigns," in Capturing Campaign Effects, ed. Henry Brady and Richard Johnston (Ann Arbor: University of Michigan Press, 2006). 80 See Walter Lippmann, Public Opinion (New York: The Free Press, 1997 (1922)). 81 See Paul Lazarsfeld, Bernard Berelson, and Hazel Gaudet, The People's Choice: How the Voter Makes Up His Mind in a Presidential Campaign (New York: Columbia University Press, 1944). 82 See Angus Campbell et al., The American Voter (New York: Wiley, 1960)., Philip Converse, "The Nature of Belief Systems in Mass Publics," in Ideology and Discontent, ed. David Apter (New York: The Free Press, 1964). Zaller, The Nature and Origins of Mass Opinion. 39 level macroeconomic variables and key political events.83 This approach did not succeed either in fully explaining election outcomes. Therefore, attention has turned to the role of strategic communications, leading to the third school of thought.84 This school studies the cognitive process, and focuses on three elements of strategic communications; agenda-setting, priming, and framing. I discuss each in turn below, paying special attention to framing in particular, as among the three it provides the most comprehensive and detailed framework to analyze strategic campaign communications.

Agenda-setting and Priming Theories Agenda-setting theories focus on the process in which competing political elites define their most relevant issues and present them to the public’s attention. The more successful they are, the better they place their issues on the agenda. Priming takes this one step further, and studies how the particular issues that are placed on the agenda affect people’s opinion. The more an issue gets attention, the more it becomes the basis of people’s judgments regarding the subject. Agenda-setting and priming studies share their independent variable, which is the priority given to an issue or problem. But they diverge in their dependent variables. The dependent variable of agenda-setting is the assessment of what problems or issues are deemed important, whereas the dependent variable of priming is the criteria that underlies the evaluation, e.g. of a political leader, or a policy.85 Agenda-setting is crucial for political elites, as agenda foreshadows outcomes and determines the path to final voting choice.86 Studies on the impact of agenda have shown that voters’ attention span is limited to the small subset of

83 Norris et al., On Message: Communicating the Campaign, 7. See Michael Lewis-Beck, Economics and Elections: The Major Western Democracies (Ann Arbor: University of Michigan Press, 1988). 84 For a detailed review of the literature see Richard Johnston, Michael Hagen, and Kathleen Hall Jamieson, The 2000 Presidential Election and the Foundations of Party Politics (Cambridge: Cambridge University Press, 2004). 85 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 112. 86 Wiliam Riker, ed. Agenda Formation (Ann Arbor: University of Michigan Press,1993), 1-2. For a detailed analysis of media, public, and policy agenda-setting theories and the multidirectional dynamics among them see Stuart Soroka, Agenda-Setting Dynamics in Canada (Vancouver: UBC Press, 2002). 40 relevant political issues that are highlighted by the news media.87 The issues placed on the agenda become the principal dimensions for evaluating and judging candidates or parties, which is called the priming effect. An individual’s choice is likely based on an analysis that provides greater weight to preferences on issues that receive heavy news coverage.88 Thus, it is the change in the salience of particular issues, which is likely to shift the distribution of preferences and thereby alter political outcomes. By calling attention to certain matters while ignoring others, news media influences the standards by which governments, presidents, policies, and candidates for public office are judged.89 It tells people not ‘what to think’, but ‘what to think about’.90 Therefore, it is a crucial battle for parties to set the issues in the media and thereby influence public priorities about the most important problems facing the country. It is important to note that neither of the two theories pays attention to the content of campaign messages. Priming theory posits that political parties try to prime the issues that are advantageous for them, which is known as the dominance principle, leading further to the ‘party ownership’ thesis that candidates advertise issues that are historically associated with their party.91 Nevertheless, Sides found that this was not the case. Using candidate advertisements from the 1998 United States House and Senate campaigns, he showed that party ownership is not consistently important and that candidates trespass and talk about each other’s issues.92 His finding is closer to the alternative ‘riding the wave’ hypothesis in the literature that both parties advertise

87 Shanto Iyengar, "Agenda Setting and Beyond: Television News and the Strength of Political Issues," in Agenda Formation, ed. Wiliam Riker (Ann Arbor: University of Michigan Press, 1993), 211. 88 Ibid., 216. 89 Shanto Iyengar and Donald Kinder, News That Matters: Television and American Opinion (Chicago: University of Chicago Press, 1987). 90 Norris et al., On Message: Communicating the Campaign, 69. Also see M McCombs and D Shaw, "The Agenda-Setting Function of Mass Media," Public Opinion Quarterly 36, no. 2 (1972).; Richard Johnston et al., Letting the People Decide: Dynamics of a Canadian Election (Montreal: McGill-Queens University Press, 1992).; Johnston, Hagen, and Jamieson, The 2000 Presidential Election and the Foundations of Party Politics.; Lawrence Jacobs and Robert Shapiro, "Issues, Candidate Image, and Priming: The Use of Private Polls in Kennedy's 1960 Presidential Campaign," American Political Science Review 88, no. 3 (1994). 91 Riker, ed. Agenda Formation. 92 John Sides, "The Origins of Campaign Agendas," British Journal of Political Science 36, no. 3 (2006). 41 on the same issues which are salient in the public’s mind.93 Sides explains that parties trespass by framing issues in particular ways that highlight dimensions consonant with their party’s traditional philosophy. Therefore, depending on the context at the time of the election, parties face a flexible agenda and have to determine their particular positions on these issues.94 Accordingly, parties not only strategically prime issues, but also need to strategically define and discuss the subject. Nadeau et al. suggest that while issue priming strategy is important, the research should be carried further with a focus on issue framing strategy.95 It is framing theory which looks into the particular description of issues.

Framing Framing is ‘a process by which people develop a particular conceptualization of an issue or reorient their thinking about an issue’.96 A framing effect occurs when, in describing an issue or event, a speaker’s emphasis on a subset of potentially relevant considerations causes individuals to focus on these considerations when forming their opinions.97 Both public opinion and social movements literatures consider framing analytically. I explore these literatures in turn, to specify which factors lead to an effective framing and thereby a strong campaign. In the next sections, I use these findings to analyze the role of campaign framing in EU referenda.

93 Stephen Ansolabehere and Shanto Iyengar, "Riding the Wave Claiming Ownership over Issues: The Joint Effect of Advertising and News Coverage in Campaigns," Public Opinion Quarterly 58 (1994). 94 See Nadeau, Petry, and Belanger, "Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election."; Adam Simon, The Winning Message: Candidate Behavior, Campaign Discourse, and Democracy (Cambridge: Cambridge University Press, 2002). 95 Nadeau, Petry, and Belanger, "Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election." 96 Dennis Chong and James Druckman, "Framing Theory," Annual Review of Political Science 10 (2007): 104. 97 James Druckman, "On the Limits of Framing Effects: Who Can Frame?," Journal of Politics 63, no. 4 (2001): 1042. 42

Framing in the public opinion literature The public opinion literature discusses two different types of framing effects. The first one is called the equivalency or valence framing effect, that occurs when different but logically equivalent phrases cause individuals to alter their preferences.98 This involves casting the same information in either a positive or negative light, by presenting alternative descriptions such as 90% employment versus 10% unemployment.99 These frames employ materially identical descriptions. The second one is the emphasis or issue framing effect, which refers to situations where, a speaker’s emphasis on a particular set of potentially relevant concerns causes individuals to focus on these specific concerns when building their opinions. These frames present qualitatively different but potentially relevant considerations. The distinction is very important as the two types of framing effects have different implications, take place through different psychological processes, and have different moderators.100 It is the second type, the emphasis or issue framing effect that is more important and relevant in political campaigns, and I refer to this particular type as the ‘framing effect’ in the following sections. The public opinion literature considers an attitude toward an issue as the weighted sum of a series of evaluative beliefs. In other words, attitude is the combination of the evaluation of that issue’s attributes and the salience weight associated with each of these attributes.101 For instance, a person’s attitude towards a new housing project might consist of both positive and negative evaluations, based on different dimensions such as economic benefits (positive) and environmental harms (negative). Therefore, the overall attitude would depend on the relative weights assigned to these dimensions. These dimensions are known as an individual’s ‘frame in thought’.102 In turn, politicians attempt to

98 Amos Tversky and Daniel Kahneman, "Rational Choice and the Framing of Decisions," in The Limits of Rationality, ed. Karen Cook and Margaret Levi (Chicago: University of Chicago Press, 1990). 99 Chong and Druckman, "Framing Theory," 114. For a detailed discussion see James Druckman, "Political Preference Formation: Competition, Deliberation, and the (Ir) Relevance of Framing Effects," American Political Science Review 98, no. 4 (2004). 100 Druckman, "Political Preference Formation: Competition, Deliberation, and the (Ir) Relevance of Framing Effects," 672. 101 Chong and Druckman, "Framing Theory," 105. 102 Ibid. 43 mobilize voters behind their policies by encouraging them to think along particular lines, emphasizing certain features of these policies. They thus invoke ‘frame in communication’, which organizes everyday reality, by providing meaning to events, and promoting particular definitions and interpretations of political issues.103 The influence these frames have on the voter is the framing effect. Essentially, the way the issue is presented produces change in public opinion. For example, when answering whether they would favor or oppose allowing a hate group to hold a political rally, 85% of respondents answer in favor if the question highlighted the importance of free speech, whereas only 45% were in favor when the question emphasized public safety and the risk of violence.104 Also, approximately 20% of the American public believes that too little is spent on ‘welfare’, while 65% thinks that too little is spent on ‘assistance to the poor’.105 There is wide consensus that framing has significant effects on public opinion formation. Framing is different from priming. Priming concerns only the transmission of salience, and not the determination of opinions on the issue. Whereas framing deals with how changes in the content of stories on a single issue affect attitudes toward the relevant public policy, priming focuses on how changes in the number of stories about an issue affect the evaluation of the broader subject, e.g. presidential performance or a policy proposal.106 In short, framing studies the qualitative content of campaign messages, while priming studies the quantity of campaign messages on a particular issue. Also, framing is not persuasion. Persuasion only occurs if the communication alters an individual’s evaluation of

103 Ibid. 104 Ibid. See Paul Sniderman and Sean Theriault, "The Structure of Political Argument and the Logic of Issue Framing," in Studies in Public Opinion: Attitudes, Nonattitudes, Measurement Error, and Change, ed. Willem Saris and Paul Sniderman (Princeton: Princeton University Press, 2004). 105 Chong and Druckman, "Framing Theory." See Kenneth Rasinski, "The Effect of Question Wording on Public Support for Government Spending," Public Opinion Quarterly 53, no. 3 (1989). 106 Druckman, "On the Limits of Framing Effects: Who Can Frame?," 1044. See also Chong and Druckman, "Framing Theory," 115. However, Chong and Druckman argue that the public opinion definition of framing can be generalized to priming. As an issue that is placed on the agenda receives considerable attention, this changes its accessibility and applicability, in turn leading to changes in the relative weight assigned to it. 44 the issue. Hence, persuasion refers to a change in the evaluation component, whereas framing is a change in the weighting component. In other words, the distinction is between a change in belief content versus belief importance.107 Effective framing changes only the importance attached to a particular dimension of the matter. Following the example above, a strong frame in favor of the housing project (emphasizing the economic benefits) would generate support for the project but would not persuade the individual that there are no environmental harms. As the process does not involve persuasion, the individual’s negative evaluation of the environmental harms would remain intact. Instead, such framing would successfully emphasize the economic benefits of the project and deemphasize the environmental dimension, and thereby make the economic aspect more salient in the opinion formation. As such, framing does not go as far as persuasion and may have a temporary impact. How does the framing effect work? The literature has mostly focused on the psychological processes behind framing effects.108 Framing effects, like other media effects, work through memory-based processes.109 It is essentially the availability, accessibility, and applicability of frames that matter.110 People draw their opinions from the set of available beliefs stored in memory, in other words they should be able to comprehend the meaning of the dimension emphasized by the frame. Availability increases with general knowledge of politics and frequent

107 Druckman, "On the Limits of Framing Effects: Who Can Frame?," 1044. See also Chong and Druckman, "Framing Theory," 116. However, in cases where the individual does not hold a clear attitude, frames can introduce new considerations as well as emphasizing existing beliefs. Therefore, some framing messages can also be persuasive. 108 Some scholars argued that framing effects worked by passively altering the accessibility of different considerations. However, there are important studies showing that framing effects work through a psychological process where individuals consciously and deliberately think about the relative importance of different considerations suggested by a frame, in other words, by changing belief importance. See Druckman, "On the Limits of Framing Effects: Who Can Frame?."; Thomas Nelson, Rosalee Clawson, and Zoe Oxley, "Media Framing of a Civil Liberties Conflict and its Effect on Tolerance," American Political Science Review 91, no. 3 (1997).; Paul Brewer, "Value Words and Lizard Brains: Do Citizens Deliberate about Appeals to their Core Values?," Political Psychology 22, no. 1 (2001). 109 James Druckman and Kjersten Nelson, "Framing and Deliberation: How Citizens' Conversations Limit Elite Influence," American Journal of Political Science 47, no. 4 (2003): 731. 110 Chong and Druckman, "Framing Theory." 45 exposure.111 However, within these only some are accessible at any given moment. Beliefs need to be activated, and regular and recent exposure to a communication frame emphasizing these considerations increases their accessibility. Whereas availability and accessibility are not conscious processes, individuals actively and consciously evaluate the applicability of accessible considerations. For instance, a demonstration may affect traffic negatively, and this may be an available and accessible concern. Nevertheless, this may not be considered applicable when thinking about the rally. Overall, framing can occur on all of these dimensions, by making new beliefs available about an issue, making certain available beliefs accessible¸ or making beliefs applicable or strong in people’s evaluations.112 What is a successful frame? To answer this question, scholars have looked into many factors by using sophisticated experimental settings. In a competitive context, the strength (persuasiveness) and the relative quantity (loudness) of a frame are the two key dimensions that determine its success.113 Chong and Druckman define strength as the extent to which a frame emphasizes available and applicable considerations.114 Weak frames on the other hand either rely on unavailable considerations, or are judged inapplicable. Similarly, frequent exposure to a frame is expected to increase the accessibility and availability of considerations highlighted by the frame.115 Nevertheless, experiments show that framing effects depend more on the strength than frequency of frames.116

111 ———, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 108. By referring to frequent exposure as a factor affecting framing, the literatures on priming and framing get entangled to a certain extent. But experiments show that framing effect depends less on the frequency of frames and more on the strength of a frame. I discuss this in detail below. Moreover, recent or frequent exposure does not necessarily imply recent campaign exposure. Instead, some issues can be recurring and thereby familiar for the public in general. 112 ———, "Framing Theory," 111. 113 ———, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 102-03. 114 Dennis Chong and James Druckman, "Framing Public Opinion in Competitive Democracies," American Political Science Review 101, no. 4 (2007): 640. 115 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 111. 116 See ———, "Framing Public Opinion in Competitive Democracies." In this article, the authors present their finding that repetition had a limited impact on the opinions of less knowledgeable individuals. However, they stress that repetition of a weak frame never allowed it to prevail over a 46

There are a variety of factors that affect the strength of a frame. Frames coming from credible sources and those that appeal to cultural values enjoy higher applicability and strength.117 Moreover, certain types of information stand out; vivid information is more easily recalled than dull or pallid stimuli.118 Elites can either use literal descriptions or adopt a thematic approach or invoke images to deliver their message. Emotionally interesting, concrete or image-provoking news stories are better remembered than those that are not. Furthermore, people pay more attention to and are better able to recall negative information.119 The literature has studied the role of emotions evoked by negative information. Jerit argues that politicians have strong incentives to use emotions such as fear, anxiety and anger.120 She presents several reasons why these emotions are important. First, the impact of emotional memories can last longer even when citizens are not conscious of them, therefore political elites who speak the language of emotion have a better chance of connecting with the electorate than those who do not. They are more likely to capture the attention of the public. Second, emotional appeals project images that are universally valued since they appeal to everyone regardless of their socioeconomic status or political views. Thus, they can be used strong opposition frame. This finding suggests that strength is more important than repetition, as they do not transform weak frames into strong frames. 117 Druckman, "On the Limits of Framing Effects: Who Can Frame?."; Chong and Druckman, "Framing Theory." 118 Jennifer Jerit and Jason Barabas, "Bankrupt Rhetoric: How Misleading Information Affects Knowledge about Social Security," Public Opinion Quarterly 70, no. 3 (2006): 281. See also Richard Nadeau and Richard Niemi, "Educated Guesses: The Process of Answering Factual Knowledge Questions in Surveys," Public Opinion Quarterly 59, no. 3 (1995).; Susan Fiske and Shelley Taylor, Social Cognition (New York: McGraw Hill, 1984).; Doris Graber, Processing Politics: Learning from Television in the Internet Age (Chicago: University of Chicago Press, 2001).; Felicia Pratto and Oliver John, "Automatic Vigilance: The Attention-Grabbing Power of Negative Social Information," Journal of Personality and Social Psychology 61, no. 3 (1991). 119 See also Riker, The Strategy of Rhetoric: Campaigning for the American Constitution.; Michael Cobb and James Kuklinski, "Changing Minds: Political Arguments and Political Persuasion," American Journal of Political Science 41, no. 1 (1997).; Michael Pfau and Henry Kenski, Attack Politics: Strategy and Defense (New York: Praeger, 1990). For the impact of negative political information on voter turnout see Stephen Ansolabehere and Shanto Iyengar, Going Negative: How Attack Ads Shrink and Polarize the Electorate (New York: The Free Press, 1995). 120 Jerit, "Survival of the Fittest: Rhetoric during the Course of an Election Campaign." See also George Marcus et al., "The Emotional Foundation of Political Cognition: The Impact of Extrinsic Anxiety on the Formation of Political Tolerance Judgments," Political Psychology 26, no. 6 (2005). Also, for a discussion of the role of emotions on equivalency frames, see James Druckman and Rose McDermott, "Emotion and the Framing of Risky Choice," Political Behavior 30, no. 3 (2008). 47 in mobilizing the party’s base as well as attracting the support of the uncommitted. These emotional appeals signal that the stakes are high, rousing citizens from inattention. Finally, she highlights that the media is more likely to cover emotional appeals than sober factual formulations, due to its preference for drama and excitement. Consequently, elites are more likely to use arguments evoking fear, anxiety and anger, which will be more enduring than those that do not elicit an affective response, all else being equal.121 To summarize, these studies suggest that vivid, concrete, image-provoking, emotionally compelling frames that contain negative information are most successful in affecting individuals’ opinions by increasing the salience of the particular dimension they emphasize. Furthermore, credible sources and resonance with cultural values contribute to success. The literature has also looked into the distinction between concrete negative argumentation and abstract pro argumentation in political campaigns. First, scholars have studied the difference between hard and easy arguments.122 While hard arguments are long, complex, more technical and relatively new to the political agenda, easy arguments are short, simple, symbolic and familiar.123 Cobb and Kuklinski expect easy and con arguments to prevail in most real-world circumstances as they catch the average citizen’s attention more easily.124 Second, Jerit in her analysis of Canada-United States free trade agreement in the 1988 Canadian federal election, found that the No campaign argumentation was more effective as they used appeals to anger (United States takeover), or to fear (economic impact).125 While most people could identify with the fear of losing their jobs, the threat posed to the local wine industry seemed more remote. Similarly, Darcy and Laver argue that especially in campaigns concerning issues which can be presented in a non-technical form, campaigners can easily play on voter uncertainties by conjuring up the image of an abyss into which the society

121 Jerit, "Survival of the Fittest: Rhetoric during the Course of an Election Campaign." 122 Cobb and Kuklinski, "Changing Minds: Political Arguments and Political Persuasion," 92. 123 While negative arguments work better in all cases, the results on hard-easy dimension is mixed, depending on the issue at hand. 124 Cobb and Kuklinski, "Changing Minds: Political Arguments and Political Persuasion," 115. 125 Jerit, "Survival of the Fittest: Rhetoric during the Course of an Election Campaign." 48 would fall if the proposal is passed.126 The situation was similar in the French and Dutch cases in 2005, as I will show in detail in the individual case chapters. The vivid, concrete, emotionally compelling No argumentation connected with the voters easily, whereas the Yes arguments remained more technical, institutional, abstract and thus remote. It is important to note that the strength of a frame is not based on its intellectual or moral content; rather, it is only strong in its appeal to audiences.127 A frame analysis should therefore separate considerations of fairness or relevance from resonance. Persuasiveness of a frame does not depend on its accuracy, validity or its relation to evidence.128 An effective frame can involve exaggeration or lies playing on the fears of the public. Strong frames often rest on symbols, endorsements, and links to partisanship and ideology, and may use heuristics rather than direct information on the substance of a policy.129 Individual-level and contextual factors are also important in mediating framing effects. First, individual predispositions such as values play a role. Individuals holding strong values are less open to frames that contradict those values.130 Furthermore, knowledge on the issue is important. However, the literature has provided mixed results on the subject so far.131 While some studies found stronger framing effects on less knowledgeable individuals, some others emphasized the opposite as the comprehensibility of the frames increase with higher level of knowledge. Second, concerning the context, Chong and Druckman

126 R Darcy and Michael Laver, "Referendum Dynamics and the Irish Divorce Amendment," Public Opinion Quarterly 54, no. 1 (1990): 18. Studying various campaigns around the world, they suggest that this dynamic may be seen as a result of the ‘community conflict’ model. In contrast to more conventional interest group politics, in which conflict is contained by a set of established political institutions and a network of entrenched elites who resolve matters according to well-understood norms and conventions, community conflict develops around sweepingly defined issues with a potential for appealing to broadly based groups within the community at large. Such issues tend to be denned in terms of strongly held abstract symbols, tend to be of fundamental social relevance, and tend either to be nontechnical or to be capable of presentation in a nontechnical form. 127 Chong and Druckman, "Framing Theory," 111. 128 ———, "Framing Public Opinion in Competitive Democracies," 652. 129 ———, "Framing Theory," 111. Nonetheless, they warn that the literature lacks a general theory that can predict which frames are more likely to emerge as the most applicable. 130 Ibid. 131 See Druckman and Nelson, "Framing and Deliberation: How Citizens' Conversations Limit Elite Influence." 49 distinguish situations when a politician is allowed to frame the issue alone, from competitive environments where individuals receive multiple frames representing alternative positions on the issue.132 Counter-framing and heterogeneous discussion groups thus moderate issue frames.133 Also, a weak frame may backfire in competitive contexts, especially among motivated individuals, causing their opinions to move in the opposite direction.134 In other words, competition and motivation place a premium on strong frames.135

Framing in the social movements literature The social movements literature studies mobilization comprehensively, focusing on political opportunities, mobilizing structures, and framing.136 Political opportunities refer to the broader political system in structuring the opportunities for collective action. Mobilizing structures imply informal and formal collective vehicles, through which people mobilize and engage in collective action; in essence this concept focuses on the organizational dynamics of movements. Lastly, framing signifies the shared meanings and definitions that people bring to their situation. Within this literature, I am interested in the framing component, particularly the activities of opinion leaders, political activists and entrepreneurs who mobilize political support via political rhetoric. The study of framing in this literature originated in the 1980s, showing that the way people interpreted their grievances was critical to their participation in movements.137 In the absence of frames specifying the necessity and viability

132 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments." 133 Druckman and Nelson, "Framing and Deliberation: How Citizens' Conversations Limit Elite Influence." 134 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 111. 135 ———, "Framing Public Opinion in Competitive Democracies," 652. 136 See Doug McAdam, John McCarthy, and Mayer Zald, Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structures and Cultural Framings (Cambridge: Cambridge University Press, 1996).; Doug McAdam, Sidney Tarrow, and Charles Tilly, Dynamics of Contention (Cambridge: Cambridge University Press, 2001). 137 Framing research started in response to the domination of the field by resource mobilization models. See David Snow et al., "Frame Alignment Processes, Micromobilization, and Movement Participation," American Sociological Review 51 (1986).; William Gamson, "Political Discourse and Collective Action," in International Social Movement Research, ed. Bert Klandermans, 50 of protest, political opportunities and powerful mobilizing networks would not be sufficient. Frames are schema of interpretation that enable individuals to locate, perceive, identify and label occurrences within their life space and the world at large.138 In other words, it is ‘meaning construction’ as frames help people notice their grievances. Social movements are therefore deeply involved in the work of naming grievances, connecting them to other grievances and constructing larger frames of meaning that will resonate with a population’s predispositions and communicate a uniform message to power holders and others.139 Frames studied by the social movement literature can take on a cultural as well as strategic conceptualization.140 Frames are derived from ideologies, but they are also oriented to the strategic demands of making claims effectively because activists are both ideological and strategic actors.141 I am particularly interested in the argumentation of these strategic actors who seek to mobilize political support. As the public opinion literature suggests, such rhetoric works by linking political choice to predispositions based on attitudes, interests, and values. Thus the concept of frame resonance is directly linked to the effectiveness or mobilizing potency of the frames.142 Two sets of interacting factors account for

Hanspeter Kriesi, and Sidney Tarrow (Greenwich: JAI Press, 1988). For an earlier work, which formed the conceptual source for these studies, see Erving Goffman, Frame Analysis: An Essay on the Organization of Experience (New York: Harper and Row, 1974). 138 Donatella della Porta and Mario Diani, Social Movements: An Introduction (Oxford: Blackwell, 1999), 69. 139 Sidney Tarrow, Power in Movement: Social Movements and Contentious Politics (Cambridge: Cambridge University Press, 1998), 110. 140 For a study that focuses on this dichotomy, and attempts to bring rationality and values together in the conceptualization of framing, see Dennis Chong, Rational Lives: Norms and Values in Politics and Society (Chicago: University of Chicago Press, 2000). 141 Francesca Polletta and M. Kai Ho, "Frames and their Consequences," in The Oxford Handbook of Contextual Political Analysis, ed. Robert Goodin and Charles Tilly (New York: Oxford University Press, 2006), 192. 142 David Snow and Robert Benford, "Ideology, Frame Resonance, and Participant Mobilization," International Social Movement Research 1 (1988). Social movements scholars have identified four dimensions of frame variation; the problems addressed, the flexibility in terms of openness to new themes, the interpretive scope referring to the breadth of the targeted group, and lastly the degree of resonance. I focus on frame resonance as this is the key dimension for this research. 51 variation in frame resonance: credibility and salience.143 First, the credibility of any framing is a function of three factors:144 i. Frame consistency: How congruent are the beliefs of the social movement organizations with its actions? ii. Empirical credibility: Is there a clear fit between the frames and the events in the world? In other words is there evidence of the claims embedded in the framing? Can the claims be empirically verified? iii. Credibility of the frame articulators: Credible speakers are more persuasive, so are the frame articulators credible depending on their status or the knowledge of the issue? Second, the salience to targets of mobilization also has three dimensions:145 i. Centrality: How essential are the ideas raised in the frames to the lives of targets? ii. Experiential commensurability: Are the frames congruent and resonant with the personal, everyday experiences of the targets? Or are the frames too abstract and distant from their lives? iii. Narrative fidelity: To what extent are the frames culturally resonant? Therefore to resonate with the audience, they should accord with available evidence, with people’s experiences, and with familiar stories, values and belief systems.146 Regarding credibility, speakers’ own credibility, the consistency of their past behavior with the proposed frames, and the fit between the frames and the real world is emphasized. Concerning salience, the centrality of the issues, the personal/everyday experience, and cultural relevance are presented as important factors. These findings parallel the public opinion literature’s stress on availability, accessibility, and applicability of frames, as well as its finding that frames involving cultural values and those coming from credible sources resonate more. The only addition the public opinion literature makes, mostly based on

143 Robert Benford and David Snow, "Framing Processes and Social Movements: An Overview and Assessment," Annual Review of Sociology 26, no. 1 (2000): 619. 144 Ibid. 145 Ibid. 146 See Polletta and Ho, "Frames and their Consequences."; Chong and Druckman, "Framing Theory." 52 experimental data, is that vivid, concrete, image-provoking, emotionally compelling frames that contain negative information are also more successful. Both literatures call for more research on identifying the sources/production of frames, and the factors that contribute to frames’ effectiveness/strength.147 The only difference between the two sets of findings concern the normative implications. While the public opinion literature discusses framing effects as a negative phenomenon, the social movements literature considers it to be positive.148 The public opinion literature emphasizes how political elites can strategically manipulate individuals, which then creates the need for means to counteract framing effects. On the other hand, the social movements literature defines it as a way to overcome collective action problems by developing shared frames. The other concepts of the social movements, political opportunities and resource mobilization are important as well. Nonetheless, as I explained in the previous chapter, all four cases have similar political opportunity structures as the institutional structures are the same. All four cases held non-compulsory and binding referenda, which allowed all interested actors, political parties and civil society groups to campaign without any political constraints. On the other hand, resource mobilization is an important aspect which affected the strength of the campaigns, as there is variation in the organizational structure and resources of the campaigns. Similarly, the public opinion literature has identified party resources as an important factor contributing to campaign success. I focus both on the importance of framing and mobilizational structures in the following sections in detail. In short, I am interested in how campaign actors framed the political opportunity they were given by the announcement of referenda, and what resources and organizational structures they used in this process.

147 Polletta and Ho, "Frames and their Consequences," 188. 148 Chong and Druckman, "Framing Theory," 120. 53

Strength of Political Campaigns In sum, political actors use strategies of agenda-setting, priming and framing to mobilize political support. Within this framework, I bring together the findings of the public opinion literature on frame strength and those of the social movements literature on frame resonance and resource mobilization. Therefore I define campaign strength combining the insights from the public opinion and social movements literatures. The more that campaigns engage in strategic agenda- setting, priming and framing, especially if they can back their campaign communications with credible sources and powerful mobilizational structures, the stronger they are. Recent developments demonstrate the importance of these factors. In recent years, political parties have adopted techniques of strategic communications, defined as the professionalization or modernization of campaigning. Successful strategic communication means running a professional campaign with clear objectives which: control the agenda, identify target voters, set consistent themes and images which touch a chord with their supporters, and present coordinated messages across all party publicity.149 Parties that achieve these are assumed to stay ‘on message’, and not be deflected by the attacks of their opponents. As there is proliferation in electoral and market research, political parties increasingly use surveys and focus groups in the preparation stage of a campaign.150 In line with the framing literature above, new campaign strategies involve stereotyping, association, demonizing, and code words to influence the electorate.151 Also, visually enticing ads such as catchy graphics, music and color capture viewer attention easily. Another important issue lies in resources. Campaigns that have more resources may conduct better public opinion research

149 Norris et al., On Message: Communicating the Campaign, 67. 150 de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue, 46. De Vreese and Semetko categorize campaign activities and tools as meetings, canvassing, information material, direct mail, internet, ads, PR firms, and polls. 151 Darrell West, Air Wars: Television Advertising in Election Campaigns, 1952-2004 (Washington D.C.: CQ Press, 2005). However, he points to the need of analyzing ads with candidate behavior and media information. 54 and thus identify frames that appeal most to the public.152 Similarly, they may advertise their themes more frequently and to a wider audience, and enlist representatives and supporters who can more credibly deliver their messages to the public. The literature about campaign strategies is still emerging and there is need for further research on the impact of campaign communications.153 Strategic communications is becoming an important goal in today’s political campaigns. Next, I move on to referendum campaigns and show how political campaigns matter more in referenda than in regular elections.

Referendum Campaigns In this section I first discuss peculiarities of the referendum campaigns and then look more closely at the literature on campaign intensity, issue salience, and level of knowledge in EU referenda. Below, I argue that not only are referendum campaigns more influential than regular election campaigns but that the EU referenda literature has not yet studied the referendum campaigns comprehensively so far. LeDuc, starting from Zaller’s theory that opinion formation is an interaction of information and predisposition, argues that in referendum campaigns voting behavior exhibits greater volatility than regular elections.154 Only when a referendum question concerns an issue in a longstanding debate, follows clearly understood ideological lines, and the positions of the political parties on the issue are well-known, does voting behavior conform to predictable patterns and resemble election campaigns.155 In other words, if voters’ predispositions such as partisanship or ideology are reinforced by the campaign,

152 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments." 153 See Nadeau, Petry, and Belanger, "Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election." 154 LeDuc, "Opinion Change and Voting Behaviour in Referendums." See also Zaller, The Nature and Origins of Mass Opinion.; Hobolt, "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums."; Ian McAllister, "Elections without Cues: The 1999 Australian Republic Referendum," Australian Journal of Political Science 36, no. 2 (2001).; Darcy and Laver, "Referendum Dynamics and the Irish Divorce Amendment." 155 LeDuc, "Opinion Change and Voting Behaviour in Referendums," 711. 55 referenda are similar to elections. On the contrary, when ideological alignments are unclear or the issue is unfamiliar to the mass public, referendum campaigns are more influential than the regular election campaigns. When the issue of the referendum is entirely new and the parties line up in a non-traditional way, the campaign becomes critical in determining the outcome.156 Campaign pamphlets, television advertising, and direct mailings from campaign organizations give voters cues, serving as short-cuts, helping them make sense of conflicting information.157 Similarly, in referenda on complex international treaties or large packages of constitutional provisions, voters do not have well-formed opinions, and public opinion shows greater movement during the campaign culminating in an outcome not predicted by pre-campaign polls.158 In such referenda, voters’ predispositions are weaker and they need the campaign to form an opinion, therefore the campaign strategies are often critical to the voting decision.159 In the EU referenda literature, scholars have recently started looking into campaigns more closely. Starting from the same point that a referendum provides a more unfamiliar context, Hobolt explores campaign effects, but from a different perspective, based on campaign information.160 The campaigns supply information, through which elites provide the voters with cues, and help voters rely on their attitudes towards the EU. In her comparative study of 19 EU referenda, she finds that the campaign environment acts both as an informer and a

156 Ibid.: 714. See also de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue.; Claes de Vreese, The Dynamics of Referendum Campaigns: An International Perspective (Basingstoke: Palgrave MacMillan, 2007). 157 LeDuc, "Opinion Change and Voting Behaviour in Referendums," 714. 158 Ibid.: 717. See also Richard Johnston et al., The Challenge of Direct Democracy: The 1992 Canadian Referendum (Montreal: McGill - Queen's University Press 1996). Johnston et al. study the 1992 referendum on the Charlottetown Accord, a package of proposed amendments to the Canadian Constitution. They find that even though voters outside Quebec were initially prepared to approve the Accord, the campaign shifted the emphasis on elements to the disadvantage of the whole package, culminating in a rejection. 159 LeDuc, "Opinion Change and Voting Behaviour in Referendums," 719. See also Lawrence LeDuc, "Opinion Formation and Change in Referendum Campaigns," in The Dynamics of Referendum Campaigns: An International Perspective, ed. Claes De Vreese (Basingstoke: Palgrave MacMillan, 2007). The timing of vote decision shows the extent to which an ‘opinion formation’ process takes place during the campaign. I have discussed this above with survey data concerning timing of vote decision in each case. 160 Hobolt, "Campaign Information and Voting Behavior in EU Referendums."; ———, "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums."; ———, Europe in Question: Referendums on European Integration. 56 mobilizer. Campaign intensity is measured as a combination of partisan polarization, perceived closeness of the race, and news coverage.161 When campaign intensity is high, it not only increases the quantity of information available but it also provides an environment which encourages people to process information. In turn, when people are exposed to more information, they are less likely to abstain, and they vote more in line with their issue-preferences (attitudes towards European integration). While this measure of campaign intensity is crucial, in effect her analysis equates intense campaign with higher levels of information. In this reasoning, when campaign intensity is high, more information is available to the voters, and they are more motivated and better able to make sophisticated judgments in line with their underlying attitudes, as these are the most relevant political dispositions.162 By increasing information, the campaign intensity helps voters be more aware of the location of their ‘ideal point’ concerning European integration, and relate this to the ballot proposition. However, this approach neglects the color of the information provided in these campaigns. Not all types of information are equally effective because of framing effects. As such, my theoretical perspective takes a step further. It does not automatically follow that the more information the voters receive in a campaign, the more they will rely on their attitudes towards the European integration. First, the elite framing of the campaign issue is not necessarily based on accurate political information. Instead of providing voters with facts, each actor

161 In her formulation, partisan polarization is the percentage of parties who recommend a no vote in the referendum weighted by these parties’ share in the parliament, thus an estimation of the opposition to the ballot proposal in parliament. The perceived closeness of the race is the average difference between intended Yes and no campaigners in the opinion polls, in the last six weeks leading up to the referendum. Finally, the news coverage is the number of daily articles monitoring the referendum issue during the three months leading up to the referendum in the two mainstream national daily newspapers. 162 See also Sara Hobolt, "Taking Cues on Europe? Voter Competence and Party Endorsements in Referendums on European Integration," European Journal of Political Research 46, no. 2 (2007). In this article, she analyzes whether party cues can function as substitutes for detailed knowledge. She finds that partisan endorsements can aid competent voting behavior – issue-voting – to the extent that voters are sufficiently knowledgeable about party positions on the EU. This study does not consider the party framing of the referendum issue. 57

‘frames’, presents a particular interpretation of the issue strategically.163 Citizens delegate the responsibility of assessing the alternative policies to journalists, experts, and elected officials. But if elites provide information that is inaccurate, incomplete or misleading, citizens may have mistaken evaluations of policy alternatives.164 Therefore, citizens do not necessarily make more sophisticated judgments. Second, these particular framings can use both domestic and European- level factors together, therefore it cannot be assumed that the information provided in the campaign increases voting in line only with attitudes towards European integration. In other words, the most relevant political predisposition may not be the underlying attitudes towards the EU, if the stronger frames link domestic problems to the European measures. The distinction is blurred, and domestic contentious issues may come to be considered relevant, precisely because of the campaign. Thus I propose a frame analysis to understand which types of information disseminated in the campaign resonated more with the public. Hobolt discusses framing effects to some extent, as part of the role of political parties in referendum campaigns.165 Using a spatial model of voting behavior, she argues that parties can convince voters to accept the ballot proposal by framing the proposal as close to the ideal point of the median voter and the reversion point – policy entailed by a No vote – as extreme. Voters essentially decide whether ratification is better or worse than the failure to ratify the treaty. Moreover, whether or not voters are certain about the positions of these points is

163 See Arthur Lupia, "Busy Voters, Agenda Control, and the Power of Information," The American Political Science Review 86, no. 2 (1992). Lupia presents a detailed spatial analysis of how different types of information affect political behavior and electoral outcomes. He emphasizes that there are many types of information cues but they are often not perfectly credible. Those who have the resources to provide political information sometimes have an incentive to mislead voters. 164 Jerit and Barabas, "Bankrupt Rhetoric: How Misleading Information Affects Knowledge about Social Security," 280. In their experiments, when the typical respondent is exposed to an environment composed entirely of misleading rhetoric (with words such as bankrupt), s/he has a 47% chance of incorrectly stating that social security will run out of money completely. Alternatively, when the same person is exposed to a benign rhetoric (with words such as reform) the likelihood of giving the incorrect response declines to 32%. 165 Hobolt, "How Parties Affect Vote Choice in European Integration Referendums."; ———, Europe in Question: Referendums on European Integration. 58 important. If they are more certain about the location of the proposal than the location of the reversion point, they are more likely to vote in favor. This is exactly where framing matters but she uses framing theories to assess only two types of frames: Consequences frames and government endorsement frames. The first are issue frames that emphasize the negative consequences of the referendum proposal, or of a No vote. The second are frames that emphasize government’s support for the referendum proposal. Based on two survey experiments involving hypothetical referenda, she finds both frames to be effective on public opinion.166 When the negative consequences of a No-vote/Yes-vote are highlighted, more people favor/oppose the proposal. While this model proves the importance of political parties and framing, the two frame categories are defined very broadly and it does not systematically analyze which attributes of a frame move a proposal closer to median voter preferences.167 The Yes and No campaigns compete for alternative framings and in a competitive environment weak frames can backfire. What factors determine the effectiveness of frames? The model does not explain the whole mechanism at work as this question remains unanswered. I take this framework a step further by looking into specific campaign messages, strategies and materials. Similarly de Vreese and Semetko, in their single-case study of the Danish 2000 referendum on the Euro, find that the campaign mattered in the vote choice along with ideology/party support, government disapproval, economic expectations, Euroscepticism, and political cynicism.168 Importantly, regarding the campaign effects, they investigate the role of issue framing. While the Yes

166 Individual-level moderators mediate this effect; more politically aware voters are more susceptible to issue frames, and those supporting the government are more susceptible to government endorsement frames. 167 Hobolt uses detailed case studies to show the difference between successful and unsuccessful framing. For instance the first Danish referendum campaign on Maastricht Treaty focused on political aspects of the integration project – more unpopular among the Danes – and removed the ballot proposal from the median voter’s ideal point. The divided and weak Yes campaign did not highlight the consequences of a no vote, and let the no campaign frame the vote as a vote for the European Communities but against the Union, and move the reversion point closer to the median voter. She highlights the domain (political vs. economic) on which the frames put emphasis, and the consequences specified by these frames. But she neither studies the individual campaign frames in detail nor systematically integrates the frame strength into her theoretical framework. 168 de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue. 59 side emphasized the economic aspects of the issue, the No side broadened the issue to cover fundamental questions on the future of Europe such as enlargement, social welfare or national sovereignty. They identify this broadening as the key winning strategy for the No campaign, yet their analysis of the news frames only distinguishes between issue-specific frames and broader multiple-topic generic frames.169 Thus, once again, this study does not examine individual campaign frames, nor does it answer the question of what makes a frame stronger. Put differently, it does not systematically analyze what distinguishes a strong frame from a weak one. Finally, Garry et al. also assess the campaign intensity in the two Irish referenda on the Nice Treaty, and argue that the public voted in line with their EU attitudes – issue-voting – in the more salient and more intense referendum.170 While this study points out that the Yes campaigners ran a more vigorous campaign, emphasized advantages of the Union, and sought to decouple domestic party politics from the referendum issue in the second campaign, they do not study the campaign dynamics systematically. The only evidence presented for the higher salience of the second campaign is survey data showing that citizens found media sources to be more useful in the second campaign, and that their level of knowledge was higher. In other words, there is no in-depth analysis of the mentioned campaign vigor, strategies, frames, or effectiveness.171 As this synopsis shows, the EU referenda literature has not studied the campaign effects thoroughly. Given the key finding that referendum campaigns are critical in opinion formation, this constitutes a significant gap in the literature.

169 They further divide the generic frames into conflict, human interest, attribution of responsibility, morality, and economic consequences frames. 170 Garry, Marsh, and Sinnott, "'Second-order' versus 'Issue-voting' Effects in EU Referendums: Evidence from the Irish Nice Treaty Referendums." 171 For a different, individual-level analysis of framing in the 2005 Dutch referendum, based on cognitive connectionist perspective, see Christian Baden and Claes de Vreese, "Making Sense: A Reconstruction of People's Understandings of the European Constitutional Referendum in the Netherlands," Communications 33, no. 2 (2008). Based on focus group data, they find that people shared considerable parts of the knowledge underlying their constructions, but used this information quite differently. They reconstruct how people use information available from media and societal discourse selectively and creatively for their own purposes. As such, the referendum question was interpreted very differently by different voters. But this study does not analyze the campaign dynamics. 60

To recapitulate, the real impact of referendum campaigns depends on the redefinition of the subject matter through campaign discourse. The framing literature also emphasizes that individuals who have a prior opinion about an issue are less-likely to be influenced by issue frames.172 On new issues where individuals do not have settled interpretations, even those with strong positions are susceptible to framing.173 Thus, framing effects are less likely on established issues with clear terms of debate.174 In this sense, the 2005 EU constitutional referendum campaigns were especially important and influential due to several factors. First, studies find that in cases where individuals have only vague notions on a political topic, the survey question at best elicits an imperfect representation of feelings based on the subset of feelings that are accessible at that moment.175 This finding parallels the initial positive public opinion polls on the TCE in fall 2004 in all four countries. These early polls were conducted at a time when the public did not yet form an informed opinion, that is, an educated interpretation of the subject, due to the unfamiliar content of the TCE. Thus, it can be assumed that these early survey questions on the referendum proposal reflected broad support for European integration. Second, the TCE was a technical issue, and a great majority of the public did not know the content of the treaty. The public in all four cases was considerably susceptible to the campaign framing, due to the absence of informed opinions. LeDuc also acknowledges that a referendum on an issue such as the TCE introduces an element of uncertainty because constitutional matters are the

172 Druckman and Nelson, "Framing and Deliberation: How Citizens' Conversations Limit Elite Influence," 732. They discuss the conflicting results of the literature concerning the impact of knowledge level on framing effects. They separate knowledge from prior opinions, and argue that elite frames will exhibit a greater impact on more knowledgeable individuals, and a smaller impact on individuals more likely to have prior opinions. 173 Chong and Druckman, "Framing Theory," 112. Chong and Druckman also discuss the impact of competition, suggesting that in these environments individuals choose the alternative that is closer to their values. However they acknowledge that this applies only when individual preferences on an issue are anchored by a core value dimension, such as freedom versus social order. For issues that are new to the agenda, the public is uncertain and therefore individuals are not sure how the new issue relates to their values. 174 ———, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 107. 175 ———, "Framing Theory," 105. 61 exclusive preserve of political elites, and voters need a complex learning process to become familiar with the underlying arguments and issues.176 Furthermore, it was an issue that was not anchored by a core value dimension such as freedom versus social order, which decreases the importance of individual-level value moderators. Third, the TCE constituted a relatively broad topic for a referendum. Because it was meant to be a constitutional document, it covered the entire body of legislation the EU had passed since its foundation. This broad topic opened an opportunity for the political elite to frame the issue with great flexibility, bringing in a wide variety of arguments ranging from social-political-economic concerns to immigration policies. Usually in referendum campaigns, the overall issue is defined well in advance of the campaign, as opposed to election campaigns where multiple issues might be placed on the agenda such as the economy, healthcare or national security.177 However, due to the broad content of the TCE, this was not the case in the 2005 constitutional referenda. Overall, given the unfamiliar, technical and broad nature of the TCE, and the absence of informed opinions on the subject, it can be assumed that the initial polls reflected broad non-specific positive attitudes towards the Union. Thus, politician’s framing during the referendum campaigns defined the extent to which public opinion shifted in all four countries. Initially, the TCE was uncharted territory in the minds of voters but it was no longer so at the end of the campaigns. I focus on the long campaign in each case to understand the opinion formation process, starting from the early positive opinion polls in fall 2004.

176 LeDuc, "Opinion Formation and Change in Referendum Campaigns." Also see ———, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," 183. In this article, LeDuc explicitly warns that a referendum on the TCE may not necessarily be about Europe as the way issues are framed, together with the dynamic of the campaign is likely to determine what voters believe they are voting on. 177 de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue, 15. 62

2005 EU Constitutional Referendum Campaigns: The Analytical Framework A strong campaign uses strategic agenda-setting, priming and framing. I argue that the stronger a state's No campaign relative to its Yes campaign, the greater the increase in the magnitude of the No vote. The campaigns mattered, as Eurobarometer survey data shows that the percentage of the voters that decided ‘at the time the referendum was announced’ is fairly low and similar across all cases. However, not all campaigns had the same intensity. Hobolt’s measurement of campaign intensity combines partisan polarization (based on the opposition to the ballot proposal in parliament), the perceived closeness of the race (based on the difference between intended Yes and No campaigners in the polls), and the news coverage (number of daily articles monitoring the referendum issue during the three months leading up to the referendum).178 Accordingly, the following is the ranking of campaign intensity in 2005 TCE referenda, in decreasing order: France, the Netherlands, Luxembourg, and Spain. The low intensity of the Spanish campaign can be seen as a result of being the first case and having a weak No campaign. Scholars argued that turnout can also be considered a campaign effect in referenda, and that it is linked to campaign intensity.179 While my dependent variable is the vote choice and not the turnout, it is interesting to observe that the turnout rates correlate with campaign intensity. Luxembourg is not comparable with the rest due to compulsory voting but the turnout levels in the others were, in decreasing order: France 69%, the Netherlands 63%, and Spain 42%. The fact that the negative cases had more intense campaigns and that these campaigns possibly mobilized more voters further justifies the necessity of a campaign analysis. Table 2.1 summarizes the analytical framework of the independent variable, relative strength of the No campaign. I measure campaign strength in

178 Hobolt, Europe in Question: Referendums on European Integration, 95. 179 LeDuc, "Opinion Formation and Change in Referendum Campaigns," 28. Le Duc argues that the referendum turnout levels show more variation due to more volatility in public opinion, moreover, turnout itself is partially a campaign effect. Where parties fail to mobilize their supporters on behalf of an issue, or where non-party groups succeed in mobilizing theirs, the outcome of a referendum can be more directly subject to a differential turnout effect. See also Hobolt, "Campaign Information and Voting Behavior in EU Referendums." Hobolt argues that in intense campaigns, people are exposed to more information and are less likely to abstain. 63

two stages, agenda-setting and priming/framing. In the first stage, agenda-setting, timing matters as defining the campaign issues in the eyes of the public necessitates early and careful planning. In the second stage, priming/framing, three components explain the success of the process: Strength of the campaign frames, credibility of the speakers, and mobilizational structures. As such, these factors are all components of campaign strength. A campaign that is stronger in all these categories will be more effective than its opponents. Stage Factors Sub-factors Agenda-setting Timing

Priming Strength of the campaign frames Availability Framing Accessibility Applicability Concrete Vivid Image- provoking Emotionally compelling Contain negative information

Credibility of the speakers Government popularity Scapegoating the European Union Disagreements within the party

Mobilizational structures Financial resources Civil society support Table 2.1: The analytical framework of the independent variable – Relative No campaign strength

While framing is the key component I focus on, agenda-setting and priming are also crucial as they pave the way for more effective framing. Political elites who were able to set the agenda and prime their themes had a better opportunity to frame the issue in their preferred way. Similarly, within framing, the strength of frames is the most important component. Yet once again if the campaigners had better credibility and resources, this amplified the power of these frames by giving the speakers more standing and opportunities to advance them. Frame strength is thus the central dimension in this analysis; all else can be seen as enhancing the opportunities and standing of the campaign actors to attain their preferred framing of the subject.

Regarding credibility, I argue that factors such as government popularity, previous tradition of scapegoating the EU, and disagreements within the parties affected the referenda outcomes, through campaign dynamics. The literature shows that in referendum campaigns, smaller and strongly ideological parties usually campaign with a consistent No message, and thereby are more successful.180 The large catch-all parties on the other hand, face a stronger challenge as they usually have high-profile dissidents within their ranks, coupled with a heterogeneous electorate. As I will show in the individual case chapters, for the TCE referenda it was mainly the center-left political parties which were divided, despite their declared official pro-TCE stance. In France, the Netherlands, and to a smaller extent in Luxembourg, the mainstream left-wing parties – Social Democrats or Greens – and/or the particular trade unions associated with them, were hesitant to vote for the TCE due to inner discussions and this was clearly reflected in their campaigns, decreasing their credibility.181 This problem was highly visible in the French case, where certain key figures and party activists of the Socialist Party and the Greens joined the No side. Not surprisingly, these parties have particularly low levels of success in keeping their voters in line with their party’s recommendations. Regarding mobilizational structures, I argue that financial resources and civil society support were important in understanding the mobilization capacity of campaigns. Whereas the Yes campaigns in all countries had significantly more financial resources than the No campaigns, the French and Dutch No campaigners gathered better financial resources and mobilized more civil society support than the ones in the two positive cases. The fact that the No campaigners could succeed, despite having fewer resources than the Yes side, show that resources do

180 de Vreese, "Political Parties in Dire Straits? Consequences of National Referendums for Political Parties." Also see Gabel and Scheve, "Mixed Messages: Party Dissent and Mass Opinion on European Integration." 181 While the Socialist Party and The Greens in France had an extreme split, the Labor Party and the GreenLeft in the Netherlands and Luxembourg Socialist Workers' Party and The Greens in Luxembourg only had inner discussions. The Spanish Socialist Worker’s Party who was in the government did not express a similar hesitation, however, this has to be read in light of the absence of a real no campaign. 65 not lead to success without the right frames.182 Better credibility or resources do not lead to success without effective frames. They only enhance the strength of the frames and should thus be analyzed in relation to the campaigns. The No side can prevail, if it conducts careful research on what parts of the issue voters would not like, have effective commercials playing to these themes and sufficient money for advertising.183 In a nutshell, I argue that the French and Dutch No campaigners were more successful than the Yes campaigners in setting their themes on the agenda, priming the key contentious themes, and framing the issue effectively. Thereby they succeeded in linking the contentious problems to the vote on the TCE. In contrast, the Spanish and the Luxembourgish No campaigns were not as strong. I present my findings on the full detail of campaign dynamics and argumentation in the individual case chapters. This analytical framework captures the dynamism of EU referendum campaigns. The literature has so far identified important variables such as the determinants of people’s general attitudes towards the EU, positioning of the political parties, or the popularity of the incumbent government. But they need to be considered in relation to referendum campaigns, as I have discussed in great detail, none of these variables account for the results on their own. Importantly, this analytical framework studies government unpopularity, the key variable of the second-order model, on two levels. First, directly, popular governments have greater credibility which strengthens their campaign frames. Second indirectly, unpopular governments provide the No campaigners with more contentious issues to campaign on. The French and Dutch incumbent governments were just as unpopular when the fall 2004 opinion polls demonstrated positive attitudes towards the TCE. Similarly, despite its popular government, in Luxembourg the

182 De Vreese and Semetko reach a similar conclusion in their comprehensive study of the 2000 Danish referendum on the Euro. Despite having fewer resources, the No campaigns win if they employ stronger frames. The most important factor is therefore the frames, not the resources. de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue, 58. 183 Ibid., 56. See also David Magleby, "Opinion Formation and Opinion Change in Ballot Proposition Campaigns," in Manipulating Public Opinion: Essays on Public Opinion as a Dependent Variable, ed. Michael Margolis and Gary Mauser (Belmont: Brooks/Cole Publishing Company, 1989). 66

No vote was only 7% short of the winning threshold. Thus dissatisfaction with the government should be broken down into speaker credibility and contentious issues to understand how these aspects interacted with the campaign process.

DIFFUSION The problem of cross-case influences, also known as ‘Galton’s problem’, is crucial in the analysis of social phenomena. In a globalizing electronic world in which local events are often monitored daily on the other side of the planet, Galton’s problem is becoming more important.184 In contrast with the assumption in most social scientific analyses, the cases are not independent from each other. Therefore, the goal of the conditioning variable is to identify the impact of sequencing and networking on the referendum outcomes. The sequencing of the 2005 TCE referenda matters as the later cases were exposed to the previous campaigns and their results. The campaigns were thus not all created equally. I argue that while diffusion did not independently influence the referendum outcomes, it could condition the magnitude of the campaign strength in the second-mover states. The later a country held its referendum, the more the previous referenda results and campaigns in the other states influenced its campaign dynamics. Nevertheless, this impact was not automatic and needed specific mechanisms and channels to work. Where these channels existed, and where the campaign of the first-mover state was intense, diffusion amplified the strength of referenda campaigns in second-mover states. In this section, I discuss diffusion effects based on interview, media content and public opinion data. The individual case chapters present my findings in detail. Table 2.2 summarizes the analytical framework of the conditioning variable. Two main factors define the influence of diffusion; strength of diffusion channels and campaign intensity. The stronger the diffusion channels between first-mover and second-mover states, the more the second-mover campaigns were

184 Mark Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions," Perspectives on Politics 5, no. 2 (2007): 260. 67

affected. Also, the higher the campaign intensity in the first-mover state, the greater its influence on second-mover campaigns.

Determinants Facilitators Sub-factors Strength of diffusion Shared language/culture channels Common media channels

Collaborative European Parliament groups networks/transnational European anti-globalization network linkages Ad hoc European networks (Institutional networks Existence of mobile communities and personal connections)

Campaign intensity Intense campaign

Table 2.2: The analytical framework of the conditioning variable – Diffusion

Regarding the strength of diffusion channels, only France and Luxembourg shared substantive channels among the 2005 EU Constitutional referenda cases. While institutional networks were common to all four cases and led to a limited amount of diffusion across them, France and Luxembourg were connected by all of the specified channels resulting in a high level of diffusion among these two cases. This interaction was so dense that the main Luxembourgish No campaigner took its lead from the French No campaign. Concerning the campaign intensity, France had the most intense campaign and influenced all the other cases through the available channels. Spain, in contrast, having had a quiet campaign, did not have an impact on subsequent cases. Therefore campaign intensity was necessary to activate the diffusion channels even though the level of diffusion depended on the strength of these channels. Figure 2.1 shows diffusion effects based on these two dimensions.

Spain France The Netherlands Luxembourg

Figure 2.1: Diffusion effects among the 2005 TCE referenda

There is an extensive body of work studying diffusion however theoretical accounts of this process currently have little internal coherence, as the proposed mechanisms are grounded in different theories.185 The literature has so far has focused on two main venues: Policy diffusion and modular phenomena such as collective action/democratic revolutions.186 Policy diffusion literature studies the temporal and spatial clusters of policy reform. Governments are independent in making their own decisions but interdependent in that they factor in the choices of other governments.187 While the findings of this literature are valuable, they have little relevance for the study of referendum campaigns as they focus mainly on the interaction between governments based on material or reputational competition. The modular phenomena literature is more relevant for the analysis of diffusion in referendum campaigns. Modular action is defined as action that is

185 Dietmar Braun and Fabrizio Gilardi, "Taking 'Galton's Problem' Seriously: Towards a Theory of Policy Diffusion," Journal of Theoretical Politics 18, no. 3 (2006): 300. 186 On policy diffusion see Beth Simmons and Zachary Elkins, "The Globalization of Liberalization: Policy Diffusion in the International Political Economy," American Political Science Review 98, no. 01 (2004).; Daniel Brinks and Michael Coppedge, "Diffusion Is No Illusion: Neighbor Emulation in the Third Wave of Democracy," Comparative Political Studies 39, no. 4 (2006).; Chang Kil Lee and David Strang, "The International Diffusion of Public-Sector Downsizing: Network Emulation and Theory-Driven Learning," International Organization 60 (2006).; for a spatial model of diffusion see Jeffrey Kopstein and David Reilly, "Geographic Diffusion and the Transformation of the Postcommunist World," World Politics 53, no. 1 (2000). On modular phenomena see Sidney Tarrow, The New Transnational Activism (Cambridge: Cambridge University Press, 2005).; Donatella della Porta, Globalization from Below: Transnational Activists and Protest Networks (Minneapolis: University of Minnesota Press, 2006).; Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions."; Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions."; Sarah Soule, "Diffusion Processes within and across Movements," in The Blackwell Companion to Social Movements, ed. David Snow, Sarah Soule, and Hanspeter Kriesi (Malden: Blackwell Publishing, 2003). 187 Zachary Elkins and Beth Simmons, "On Waves, Clusters, and Diffusion: A Conceptual Framework," The Annals of the American Academy of Political and Social Science 598, no. 1 (2005): 35. 69 based in significant part on the prior successful example of others.188 In other words, these actions draw inspiration and expertise from the previous cases. It must be noted that each case is based on local initiative and local dissatisfaction, however each takes previous cases as a model for action. This way, some groups that might be less structurally advantaged to engage in successful action gain advantage by riding the influence of the prior examples.189 Modularity has been applied widely in the collective action literature, focusing on the borrowing of mobilizational frames, repertoires, or modes of contention across cases. Referendum campaigns are built on the interaction between the campaign leaders and the public. Therefore, by nature such campaigns carry the logic of the modular collective action model. They share the core task of persuading and mobilizing people. In the later cases, if diffusion was present, it directly amplified the campaign effectiveness. Campaigners learned from the experience of the previous campaigns. There are several in-depth studies of diffusion in the modular action literature. From the social movements perspective, early work considered diffusion negatively as ‘contagion effects’.190 In the recent years, the analysis of diffusion became more positive especially because of the interest in globalization and increasing interdependency among actors and events in disparate locations.191 For instance, Tarrow analyzes how transnational processes bring new ways of framing domestic issues and new identities that may combine domestic with international contention.192 Another strand of work on diffusion originates from the postcommunist democratization literature. Beissinger shows that the four democratic revolutions – Bulldozer, Rose, Orange, Tulip Revolutions – which took place in Eastern Europe between 2000 and 2006 were examples of modular

188 Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions," 259. 189 Ibid.: 260. 190 For an excellent review of the literature see Soule, "Diffusion Processes within and across Movements." 191 Donatella della Porta and Hanspeter Kriesi, "Social Movements in a Globalizing World: An Introduction," in Social Movements in a Globalizing World, ed. Donatella Della Porta, Hanspeter Kriesi, and Dieter Rucht (London: Macmillan, 1999).; Tarrow, The New Transnational Activism. 192 Tarrow, The New Transnational Activism. 70 political phenomena.193 Similarly, but broadening the focus, Bunce and Wolchik study the wave of democratization through electoral revolutions in the postcommunist world as a case of international diffusion.194 Importantly, with this new focus, scholars sought to better specify the mechanisms through which diffusion takes place. Most definitions of diffusion include the following four elements: a transmitter, an adopter, an innovation that is being diffused, and a channel along which the item may be transmitted.195 I am specifically interested in this last component within this literature, in order to identify the factors that enabled diffusion in the 2005 TCE referenda. The literature specifies two basic models of diffusion; hierarchical and proximal.196 The hierarchical model argues that diffusion occurs in a top-down fashion from actors of higher status to actors of lower status, whereas the proximal model suggests that actors mimic others who are spatially or culturally relevant to them. Given the equal standing of the four referendum countries, proximal models are more important in the analysis of the 2005 referendum campaigns. Furthermore, the literature distinguishes between direct and indirect channels that transmit diffusion.197 Direct channels of diffusion can be highly effective as individuals are in direct and frequent contact. Yet there may also be indirect channels of diffusion, either through a sense of shared identification between activists or common mass media sources.198 Tarrow essentially makes the same distinction, between relational and nonrelational diffusion pathways.199 While relational diffusion functions through social bonds and personal networks, nonrelational diffusion works among people who have few or no social ties,

193 Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions." 194 Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions." 195 The transmitter implies the source, the adopter implies the receiver. See Rogers, Diffusion of Innovations.; Soule, "Diffusion Processes within and across Movements." 196 Soule, "Diffusion Processes within and across Movements."; della Porta and Diani, Social Movements: An Introduction.; David Collier and Richard Messick, "Prerequisites versus Diffusion: Testing Alternative Explanations of Social Security Adoption," American Political Science Review 69 (1975). 197 Soule, "Diffusion Processes within and across Movements." 198 For an excellent review of studies focusing on these types of channels see Ibid.; Doug McAdam and Dieter Rucht, "The Cross National Diffusion of Movement Ideas," Annals of the American Academy of Political and Social Science 528, no. 36-59 (1993).. 199 Tarrow, The New Transnational Activism, 103-06. 71 essentially through the mass media and electronic communication. Below I identify three channels that transmitted diffusion in the 2005 TCE referenda: Shared language/culture, common media channels, collaborative networks/transnational linkages. While the first and the third present a combination of direct and indirect channels, the second one is a uniquely indirect channel. I discuss each I turn.

Shared Language/Culture Studies show that modular political phenomena is made possible by a sense of interconnectedness across cases – produced by common institutional characteristics, histories, cultural affinities, or modes of domination – which allow agents to make analogies across cases and read relevance into developments in other contexts.200 The more similar the sending and receiving countries are, the more likely the innovation will travel.201 Monitoring of the activity in other cases is promoted by the existence of these shared characteristics, such as perception of common needs, capacities and benefits, common contexts and common identities. Such characteristics may be institutional – e.g. electoral fraud – as in the postcommunist transition cases. Among the 2005 EU referenda cases, it was the cultural and linguistic similarities which performed as an important enabling factor. These similarities can in general be seen as an indirect channel based on a sense of shared identification. But shared language also strengthens direct channels as it facilitates personal contact and communication. Almost all of my interviewees mentioned the importance of language and culture in determining their level of contact with campaigners in other countries. Among the four cases, it was France and Luxembourg which were linked culturally and linguistically.202

200 Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions." 201 Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions," 288. 202 In a similar fashion, the Dutch campaigners mentioned their close ties to the British parliamentarians in Europe in general, also stressing the importance of language. 72

Common Media Channels Mass media can serve as an indirect channel of diffusion. Studies on race riots repeatedly found that televised civil rights activism played an important role in creating a black solidarity that transcended boundaries of physical communities.203 The media served as an indirect channel of diffusion by creating a cultural linkage between African Americans in different metropolitan areas. Among the 2005 EU referenda cases, once again only Luxembourg and France shared media channels.

Collaborative Networks/Transnational Linkages There are two facets to this channel: first, there are institutional networks which serve as indirect channels, and second, personal connections which serve as direct channels. I review each in turn.

Institutional networks The literature suggests that transnational linkages connecting the cases are crucial. Diffusion is more likely when there are collaborative networks crossing national boundaries, which promote the diffusion of the model and provide incentives for actors to embrace transplantation.204 These networks can be provided by NGOs, civil society activists or other more established institutional frameworks. Although cross-national networks are critical to most processes of diffusion, these networks can be formal or informal, long-in-place or new.205 The institutional networks were common to all four cases, and were mentioned in the interviews as important facilitators. First, both the Yes and No campaigners pointed to the European Parliament (EP) groups and parties – to

203 See Soule, "Diffusion Processes within and across Movements."; Seymour Spilerman, "Structural Characteristics of Cities and the Severity of Racial Disorders," American Sociological Review 41, no. 771-793 (1976).; Daniel Myers, "The Diffusion of Collective Violence: Infectiousness, Susceptibility, and Mass Media Networks," American Journal of Sociology 106, no. 1 (2000). 204 Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions," 288. See also Beissinger, "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions," 262. 205 Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions," 287. 73 which their parties are members – as platforms to share ideas with other similar European political parties. This formed a regular networking opportunity, where most of the campaigners said they discussed their experiences. However, there is significant variation in the type of experience and advice that was gained. While these networks constituted a platform for political parties, the extent to which they were influential in shaping the campaigns is not clear. Interestingly, concerning the institutional linkages, the far left and far right took extreme positions. The far left mentioned close contacts with other far left parties in Europe, whereas far right parties rejected any links to other such parties. Second, institutional linkages were more visible and prominent in the analysis of the left-wing civil society groups.206 The strong anti-globalization network in Europe provided a platform for these groups to come together, discuss the agenda, analyze the new developments, and form positions. The Spanish, Dutch and Luxembourgish left-wing activists mentioned that in preparation for the campaign they first read about the discussion in France, and were inspired by the French debate in their argumentation. They frequently referred to the Association for the Taxation of Financial Transactions for the Aid of Citizens (ATTAC) network, and the European Social Forum (ESF) as important active platforms.207 The ESF meetings bring together the left-wing social movements with political parties. Particularly, the activists stressed the 2003 ESF meeting in Paris and the 2004 ESF meeting in London as sources, where the TCE was debated intensely. The Dutch and Luxembourgish No campaigners also noted the legitimization impact of the French debate and results, as they lent credit to their arguments, strengthened their case, and contributed to their visibility. However, the activists also highlighted the ad hoc nature of these informal networks. There was no structure apart from the ATTAC international network, the annual ESF meetings, and the temporary European No Campaign (ENC) network established

206 There was no equivalent networking among the far right no campaigners, which was confirmed by almost all of my interviewees. 207 On the formation of anti-globalization network in Europe see Andy Mathers, Struggling for a Social Europe: Neoliberal Globalization and the Birth of a European Social Movement (Aldershot: Ashgate, 2007). On the impact of the anti-globalization demonstrations on European summits see Élise Féron, "Anti-globalization Movements and the European Agenda," Innovation: The European Journal of Social Sciences 17, no. 2 (2004). 74 by British businessmen at the time.208 The ENC was a Europe-wide, independent, cross party network of political parties and NGOs which brought together both right and left-wing No campaigners. It was set up to coordinate ‘anti-EU Constitution’ activities, and maximize the No vote in every EU referendum on the TCE.209 Through this institutional channel, campaigners borrowed campaign frames, riding on the influence of prior examples. The social movements literature identifies master frames, which are larger frames used by a cluster of movements. The anti-globalization network in Europe, built around the ATTAC and ESF structure, created an anti-globalization master frame, which opposed the global spread of ‘neoliberal’ capitalism and the subjugation of political and social decision-making to the priorities of the market and global finance. It was the Paris ESF meeting and then the French campaign which linked this master frame closely to the vote on the TCE, advocating an alternative ‘social’ EU. Left-wing activists in all four cases echoed these arguments. It is important to note that this mechanism worked both ways. Where the borrowed frames were not adapted to the local reality, this influence became problematic. The literature states that even if the borrowing is deliberate and planned, the local conditions supporting diffusion vary.210 Frames need to be used strategically to improve the chances for adoption.211 While borrowing from the French case was mainly an advantage, it could also have had negative effects if these frames were not adapted to local reality. The Dutch far left political party strongly emphasized the importance of

208 Personal interview with Thomas Rupp, director of the ‘European No Campaign’, on April 18th, 2008. This platform was discontinued after the French and Dutch rejections. 209 But due to its cross-party character, this platform was not very active, especially since the French far left did not want to be part of a broader network that brought in far right campaigners as well. 210 Bunce and Wolchik, "International Diffusion and Postcommunist Electoral Revolutions," 287. 211 See Soule, "Diffusion Processes within and across Movements."; David Snow and Robert Benford, "Alternative Types of Cross-National Diffusion In the Social Movement Arena," in Social Movements in a Globalizing World, ed. Donatella Della Porta, Hanspeter Kriesi, and Dieter Rucht (London: Macmillan, 1999). Snow and Benford give the example the spread of the Sokagak-kai/Nichiren Shoshu Buddhist movement to the United States and other countries from Japan. Part of the reason that this movement has been so successful worldwide is presented as its ability to cloak its beliefs in culturally specific traditions and symbols that resonate with the particular society that the leaders are targeting. 75 the work that had to be done nationally, to fit them to the national framework.212 However, the Spanish or the Luxembourgish far left campaigners did not do the same, which resulted in less successful framing. The Dutch far left political party was the only one that strategically linked these anti-globalization concerns to their central No frame: The loss of sovereignty in a European ‘super-state’.213 Thus, borrowing frames worked both positively and negatively depending on the degree to which these borrowed arguments fit the local reality, in line with the expectations of the framing literature.

Personal connections Personal connections provide a direct and strong channel of diffusion as they are based on face-to-face interaction. A highly interesting study found that information about rebellions diffused through communication networks of travelers along transportation routes in the UK and France between 1730 and 1848.214 Another work similarly showed that the diffusion of the Swedish Social Democratic Party between 1894 and 1911 followed the travel routes of political agitators at that time.215 Among the 2005 EU referenda cases, only France and Luxembourg shared a personal bond. This channel is based on Luxembourg’s peculiar geographic, institutional and demographic setup. These characteristics, which I will discuss in detail in the Luxembourg case chapter, created a personal channel of diffusion carrying the influence across the border on a daily basis. These channels, taken as a whole, connected the Luxembourgish case to the French one closely. Apart from the intense interaction between the

212 This strategic move was not compatible with the French far left critique of the TCE. Thomas Rupp, the director of the ‘European No Campaign’, has also pointed out that the French far left and the Dutch/Scandinavian far left were very different in this regard, as the latter strategically tied the left-wing argumentation to the ‘super-state’ argument. French far left campaigners were resistant to share a common platform with the far right campaigners, whereas the Dutch far left was not. Personal interview with Thomas Rupp, director of the ‘European No Campaign’, on April 18th, 2008. 213 They not only underlined the ‘neoliberal’ character of the TCE but also argued that Brussels would have more control over the Dutch social policy. 214 See George Rude, The Crowd in History, 1730-1848 (New York: John Wiley, 1964).; for a detailed literature review see Soule, "Diffusion Processes within and across Movements." 215 See Peter Hedstrom, Rickard Sandell, and Charlotta Stern, "Mesolevel Networks and the Diffusion of Social Movements: The Case of the Swedish Social Democratic Party," American Journal of Sociology 106, no. 1 (2000). 76 campaigners, the public was also strongly exposed to the French campaign argumentation through common media channels and the mobile communities. Focus group data show that Luxembourgish voters conflated their country’s problems with the French ones, which is a clear indicator of the high exposure. Importantly, Luxembourg shared these strong diffusion channels not just with any case, but with the most intense case. Campaign intensity matters, I argue that the campaign needs to be intense in order to activate the diffusion channels. McAdam distinguished between initiator movements, which set a protest cycle in motion, and spin off movements, which are sparked by the initiators.216 Among the 2005 TCE referenda, France had the most intense public debate and became the initiator movement. Spain, despite being the first case, did not influence any of the subsequent campaigns with its low-profile and quiet campaign. Because the French discussion on the subject pre-dated the Spanish campaign, it was instead the French campaign that influenced the Spanish No campaign through the European anti-globalization network. Campaign intensity thus determines the amount of pressure on the diffusion channels; the more intense the campaign, the greater its influence on later cases. Overall, due to the exceptional diffusion channels between France and Luxembourg, the main Luxembourgish No campaigner took the French left-wing No campaign as a model.217 The shift in Luxembourgish Yes campaign strategy, essentially its shift to stronger campaign framing, was also related to sequencing as it did so only after the French and Dutch rejections. In Luxembourg, diffusion

216 Doug McAdam, "“Initiator” and “Spin-off” Movements: Diffusion Processes In Protest Cycles," in Repertoires and Cycles of Collective Action, ed. Mark Traugott (Durham: Duke University Press, 1995). 217 Concerning the role of agency in the diffusion process, the literature specifies four situations: Reciprocation (where both the transmitter and the adopter are interested/have an interest in diffusion), adaptation (where there is an active adopter and a passive transmitter), accommodation (where the transmitter promotes the diffusion of a relatively alien practice by tailoring the innovation to the needs of a fairly passive adopter), and finally simple contagion (where neither the adopter nor the transmitter have any interest in the item being diffused). I argue that in the 2005 TCE referenda, the diffusion was mainly based on reciprocation as the campaigners had an interest in sharing campaign strategies and materials, in order to promote their preferred referendum result all across Europe. Only the personal connections between France and Luxembourg could be considered as having less agency than the other channels, as this community unintentionally carried such influences in their daily interactions. See Snow and Benford, "Alternative Types of Cross-National Diffusion In the Social Movement Arena." 77 conditioned the campaign strength, as it led to both an increase in the No vote and its final containment in the last stage of the campaign. Where the specified conditions were met, diffusion effects were strong. Through networking and learning processes, diffusion conditioned the campaign strength and the choice of campaign strategies in the second-mover countries. It amplified the campaign strength directly, and in turn indirectly influenced the referendum outcomes.

Preview of the Case Findings Below I present a brief summary of the case findings. In terms of agenda-setting, in the French, Dutch, and Luxembourgish cases, the No campaigners came into the debate earlier than the Yes campaigners. This gave them the agenda-setting power and the opportunity to set the tone of the debate. In these countries, the Yes campaigners suffered greatly from taking the defensive position, and were forced to follow the No campaigners’ agenda. In terms of priming and framing, the French and Dutch No campaigners had more success in framing issues in their preferred way. They primed contentious issues, relating both to the domestic and European levels of governance such as increasing immigration, rising unemployment, market- friendly reforms, decline in the welfare state model, and various unpopular EU measures such as enlargement or the common currency Euro. Their frames blamed the TCE for these problems, linking them to broader concerns such as loss of national sovereignty and identity, ‘neoliberal’ agenda, and the lack of social protection. These themes were available and accessible as there was already contention around them in the society. The No frames also enjoyed greater strength and applicability because they were more concrete, vivid, image- provoking, and contained negative information. These arguments were also emotionally compelling. The No campaign arguments evoked, in essence, a fear of globalization by dramatizing the negative consequences. Both left-wing No arguments pointing at decreasing welfare, and the right-wing No arguments stressing the influx of immigrants (both culturally and economically, due to cheap

78 labor), could appeal to the fear of globalization. The wall around the welfare state could complement the wall around the country in a coordinated effort to stop unwelcomed aliens and welfare claimants.218 In both France and the Netherlands, there were recent reforms of the welfare system, and social movements against these public spending cuts. Similarly in both countries increasing immigration was a highly salient issue, and far right political leaders had shocking levels of electoral success in 2002.219 The perceived fit between these frames and the reality was greater but the key point is that these issues were equally contentious in fall 2004 when the polls showed positive attitudes towards the TCE. Therefore, in the absence of effective framing these problems remained unlinked to the referendum proposal. Also, in referendum campaigns the shift in public opinion can shift back again once the turbulence of the referendum campaign is out of the way.220 This was indeed the case in France, the Netherlands, and also Luxembourg. The French and Dutch Yes campaigners’ arguments on the other hand, were technical and institutional, and very much tied to the content of the constitutional treaty, which were not available or accessible for most of the society. The pro arguments were also abstract, referring frequently to the importance of the Union in providing peace in the aftermath of the Second World War. This concern was too abstract to be considered applicable especially for the young generations, as Europe did not have any imminent threat of war and as the TCE was not linked to security concerns. Most of the Yes frames presented the TCE as an institutional step towards a better Europe. These frames were not related to people’s daily lives, and did not have any substantial linkage to the salient issues in domestic or EU politics of the time.

218 This formed a bridge as the two arguments worked together to protect the country, the country’s identity in an increasingly globalizing world. Frank Lechner, The Netherlands: Globalization and National Identity (New York: Routledge, 2008), 202. 219 Instead of the more loaded term ‘extreme-right’, I use the tem ‘far right’ to include a wider area of right-wing political identitification where pronounced positions (conservatism, nationalism, authoritarianism, cultural separatism etc.) shade into extremism. See J.G. Shields, The Extreme Right in France: From Pétain to Le Pen (Abingdon: Routledge, 2007), 320. 220 LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," 179. 79

Spain and Luxembourg were able to sustain the favorable public opinion on the subject. Yet treating these two countries simply as ‘positive’ cases is misleading as they have nuanced campaign dynamics. Despite popular governments and the relatively lower level of contention relating to the EU, the No campaigns made a difference where they succeeded in tapping into existing concerns. The positive outcome in Spain is primarily explained by the absence of strong No campaign framing; the yes campaigners did not have a difficult task in sustaining the already positive public opinion on the subject. However, regional No campaigns were stronger where the No vote percentages (33% in the Basque Country and 28% in Catalonia) were indeed higher than the national average (17%). The No campaigners in the Basque Country and Catalonia blamed the TCE for not recognizing their self-determination, and particularly their regional languages, which was a salient concern in these regions. The fact that the regional No vote figures were higher confirms the significance of the independent variable, relative No campaign strength, in explaining the results. In Luxembourg, on the other hand, the No vote reached 43%. A careful analysis of the campaign frames and strategies reveal the reasons behind such a high No percentage in a strongly pro-EU county. Interestingly, this campaign started in much the same way as the French and Dutch ones. The No side was able to engage in successful agenda-setting, and had limited success in linking some contentious issues such as unemployment and decreasing protection of public services to the referendum proposal. The reason behind its limited success is the diffusion process. Due to the exceptional diffusion channels that existed between France and Luxembourg, the main Luxembourgish No campaigner relied mainly on campaign frames it borrowed from the French No campaign. These borrowed frames, while resonant to a certain extent, were not tailored to fit the local reality, which would have increased their strength. On the other hand, the Yes camp decided to change its campaign strategy after the French and Dutch rejections, which was critical to contain the increasing No vote. The revitalized Yes campaign successfully framed the TCE as a crucial step for Luxembourg’s survival in Europe, presenting the treaty as a matter of national interest. The new

80 framing was much more concrete and centered around the specific benefits received from Europe. Overall, the No vote was much higher in Luxembourg because of the No campaign’s relative strength and diffusion effects. In the absence of both, the Luxembourgish vote totals would likely have paralleled Spain’s figures closely. Overall, negative messages are more influential on individuals. As I explained in detail above, the literature has similar findings on the distinction between concrete negative argumentation and abstract pro argumentation in political campaigns.221 This dynamic took place easily in Europe as the EU is a new and distant level of governance, despite the fact that many features of European integration are visible in the domestic sphere. Negative and more immediate linkages to existing and thereby available frames won over positive, abstract, non-immediate, non-applicable benefits. Specific arguments which put Europe in a bad light were considered more plausible or persuasive than arguments which give Europe a star role. Negative concrete points prevailed over vague benefits and ideals.222 Once again, whether or not the No campaign arguments were accurate is irrelevant in framing analysis. An effective frame can be based on exaggeration or lies playing on the fears of the public. This must be separated from an evaluation of accuracy or fairness. The strength of campaign frames was amplified if they were backed by credible sources and powerful mobilizational structures. In the case chapters I show how these factors amplified or decreased the campaign strength with detailed empirical evidence using interview, media content and public opinion data. The case chapters follow the chronological order of the 2005 EU Constitutional referenda: Spain, France, the Netherlands, and finally Luxembourg.

221 See Cobb and Kuklinski, "Changing Minds: Political Arguments and Political Persuasion."; Jerit, "Survival of the Fittest: Rhetoric during the Course of an Election Campaign." 222 Dekker and Wennekers, "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen.", cited in Paul Dekker and Albert van der Horst, Europe's Neighbours: European Neighbourhood Policy and Public Opinion on the European Union (The Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis, 2008). They argue that if people have any doubts, they mostly prefer to be realistic or cynical rather than naïve. 81

CHAPTER 3 SPAIN: THE QUIET CASE

Spanish voters were the first to vote on the European Constitutional Treaty, and they approved it on 20 February 2005 by 76.73%. The No vote was limited to 17.24%, the lowest rate among the four states. The percentages suggested a great success, paving the way for a common European constitution. Spain’s forceful ‘Yes’ was thus different from the results in the subsequent cases. However the Spanish campaign started in the exact same way: The public initially held positive opinions towards the TCE, and while the political mainstream supported the TCE, the parties on the extremes of the political spectrum campaigned against it.223 In October 2004, four months before the referendum, opinion polls showed that Spaniards were in favor of the TCE.224 When they were asked ‘How would you vote if the referendum was tomorrow?’, 44.6 % were in favor while only 4% were against the TCE.225 Similarly, 56.7% said they would like the February referendum to approve the TCE, whereas only 5.5% said they would not. Also, the fall 2004 standard Eurobarometer showed that 72% of the public supported the general idea of a constitution for the EU.226 The November 2004 special Eurobarometer further demonstrated that 56% of Spaniards supported the TCE specifically.227 The Spanish political mainstream formed the Yes campaign; the governing Spanish Socialist Workers’ Party (PSOE), the main opposition party Popular Party (PP), the Catalan Nationalist Convergence and Unity party (CiU), the Basque Nationalist Party (PNV), and the Canary Islands liberal nationalist party Canarian Coalition (CC) were all in favor of the TCE. Only the far left, and

223 There was no far right party in the Spanish parliament, therefore the No campaign was confined to far left and green parties. 224 The Centro de Investigaciones Sociológicas (CIS) survey, available at: http://www.cis.es/cis/opencms/-Archivos/Marginales/2560_2579/2577/Es2577.pdf 225 30.4% said they did not know, 13.6% said they would abstain, and 5.2% said they would vote blank. 226 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004," European Commission: Brussels (2004). 227 ———, "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution." 82 regional green/far left parties were against the referendum proposal. The No campaigners were the United Left (IU), the Republican Left of Catalonia (ERC), the Initiative for Catalonia Greens (ICV), the Basque nationalist social democratic Basque Solidarity (EA), the Galician Nationalist Block (BNG), and the Aragonese Council (CHA). Yet, the turnout level was relatively low at 42.32%.228 This rate was not only the lowest among the four referenda cases but also the lowest turnout in a national poll since Spain’s return to democracy nearly 30 years ago.229 As this rate suggests, the campaign was not intense and did not generate much public debate. Being the quietest case of all four, the Spanish case has not attracted sufficient scholarly attention. Nevertheless, it is actually important to examine Spain as a control case. Unlike in the three other cases, the evolution of public opinion on the TCE in Spain did not show any unexpected fluctuations or sudden dips. The initial positive public opinion, a shared characteristic in all four countries, was sustained. Thus the key questions of this chapter are the following: First, and most critically, why was Spain uniquely able to maintain a high, favorable public attitude towards the TCE? Furthermore, why did the Spanish vote not influence the subsequent referendum campaigns? I argue that the absence of a strong and strategic No campaign resulted in pro-TCE public opinion remaining at its initially high levels all the way through to the referendum date. Without a powerful No campaign, the public was not exposed to effective anti-TCE arguments. Even though Spanish society was fairly concerned about issues such as unemployment, terrorism, immigration or crime, these contentious issues remained unlinked to the referendum proposal. In the absence of a strategic No campaign, these problems were not connected to the TCE in the public’s mind. Catalonia and the Basque Country were the exceptions. In these regions, the No campaigns were stronger and successfully tapped into the national self-determination issue by blaming the TCE for not recognizing the

228 For a detailed analysis of the causes of low turnout see Eva Anduiza Perea, "Who Abstained and Why? Voter Turnout for the Referendum on the Treaty to Establish a European Constitution," Elcano Royal Institute Working Paper 34 (2005). 229 Corinne Deloy, "The Spanish Approve the European Constitution by a Wide Majority," Robert Schuman Foundation Report (2005): 10. 83 regional languages. Indeed, the No vote levels were higher in these regions, proving the significance of the No campaigns. Second, Spain’s Yes campaign benefited from holding the first referendum and avoided following a negative case. Spain shared limited diffusion channels with the subsequent cases. Moreover, as the Spanish campaign was of low intensity, it did not activate the existing channels to transmit any influence. It was instead the intense French campaign which had an impact on all other campaigns, Spain included, as the French debate pre-dated the Spanish one. Through the institutional networks that were common to all four cases, Spanish campaigners were exposed to a limited amount of French anti-TCE campaign frames. But given the low level of diffusion, it did not condition the campaign strength. Overall, the Spanish vote is explained by the relative weakness of the No campaign as the Yes campaigners did not have a challenging task to sustain the initial pro-TCE public opinion. The chapter will proceed as follows. First, I will present the shifts in voting intentions over time to show the evolution of the public opinion. Second, I will discuss the Spanish public’s discontent with both domestic and EU levels of governance to discern the salient contentious issues of the period. This will highlight the societal problems that constituted potential No campaign material. Third, I will analyze the long campaign period using extensive interview data. Here I will assess the Yes and No campaign strategies focusing on the strength of the campaign frames, the credibility of the speakers, and the mobilizational structures (financial resources and civil society support) available to both sides. This section will explain in detail how and why the No campaign was virtually non-existent (with the exception of the Basque Country and Catalonia), making the Yes campaign’s task relatively easy. Fourth, I will use media content analyses and public opinion data to analyze the referendum vote, demonstrating the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. As such, the media and survey data will verify the preceding campaign frame analysis. Last, I will discuss why diffusion was so

84 limited in Spain by exploring both the diffusion channels between Spain and the other cases and the intensity of the campaign.

Evolution of the Public Opinion Figure 3.1 shows the shifts in voting intentions over time, combining both European and national polls:230

Source: Elcano Royal Institute

Figure 3.1: Evolution of voting intentions in Spain

The graph illustrates a steady line for the intention to vote No, with a slight increase towards the end. The Yes intention shows a dip in October 2004, but then fluctuates between 35-45%. It must be noted that the last month is not covered here, during which the No percentage increased to the final 17%, and the Yes percentage to the final 76%. Spaniards did take the campaign into account, as will be made clear below in the assessment of the public opinion data. But the graph demonstrates that the No campaign did not succeed in increasing the No vote, in contrast with the pattern in the other three campaigns.

230 José Ignacio Torreblanca and Alicia Sorroza, "Spanish Ratification Monitor," Elcano Royal Institute Working Paper 8 (2005): 8. 85

Discontent with Domestic and EU Governance In this section I discuss Spanish public opinion concerning domestic and EU levels of governance in the period leading up to the referendum. Spain was fairly satisfied with its young government. Moreover, the public attitude towards European integration was positive and above the EU average. This general optimism contributed to the initial favorable opinion on the TCE. Nonetheless, there were still contentious issues at both levels that potentially provided powerful No campaign material.

Contention concerning domestic politics Spain held general elections in March 2004, following two terms (1996-2004) of center-right Popular Party government under José María Aznar. The Spanish Socialist Workers' Party, led by José Luis Rodríguez Zapatero, won the elections and became the governing party. The main contentious issues of the period were unemployment, terrorism, immigration and crime, which I will discuss in detail below. The March 2004 elections were unusual, leading to an important switch, both in terms of ideology and foreign policy orientation. The right-wing Aznar administration had followed an explicitly pro-US policy, strongly advocating the global anti-terrorist campaign of US President George W. Bush.231 In contrast, the PSOE openly advocated for a European foreign policy. This difference played a key role in the election. Strikingly, the 2004 general elections were held only three days after the terrorist train bombing in Madrid. The way the PP government handled the incident was highly controversial as it immediately suggested that the Basque separatist Basque Homeland and Freedom (ETA) was responsible for the attacks.232 Yet, soon it was discovered to be an Islamist terrorist attack, which signaled a potential response to Spanish involvement in the War on Terror. The public reacted to the situation severely by various demonstrations. The pro-US vs.

231 Carlos Closa and Paul Heywood, Spain and the European Union (New York: Palgrave Macmillan, 2004), 219. 232 Raj Chari, "The 2008 Spanish Election: A Balancing Game," West European Politics 31, no. 5 (2008). 86 pro-EU distinction was used strongly by the social democrats in opposition. The PSOE presented Aznar’s policy as a retreat from traditional Europeanism and adopted the slogan ‘return to Europe’ during the election campaign.233 A year later it successfully used the same theme again during the TCE referendum.234 Monthly surveys by the Center of Sociological Investigations (CIS) monitor Spanish public opinion regularly. The surveys explicitly ask the public to rate the PSOE government. Between July 2004 and January 2005, the Spanish public was moderately satisfied with the Zapatero government. There was a slight deterioration over time, but overall 2-4% ranked the government ‘very good’, 34- 39% ranked it ‘good’, 33-44% ranked it ‘neutral’, 9-10% ranked it ‘bad’, and only 2-3% ranked it ‘very bad’. More specifically, another survey question asked individuals how much they trusted Prime Minister Zapatero personally. Once again the answers got less positive over time, but the balance was affirmative: 9- 11% expressed ‘a lot of confidence’, 41-45% expressed ‘sufficient confidence’, 25-31% expressed ‘little confidence’, and only 10-13% expressed ‘no confidence’. Interpreting the survey data, Font suggests that the immediate honeymoon period with the newly elected PSOE was over and that the support levels were back to normal by February 2005.235 This suggests that at the time of the TCE referendum, the early enthusiastic mood relating to the newly elected government was over and that the satisfaction levels were not too high anymore. Yet, a popular government does not imply that Spanish society was immune from contentious issues. These same surveys between June 2004 and

233 As promised, after coming to power, the PSOE immediately withdrew the Spanish troops from Iraq. Carlos Closa, "Business as Usual: EU Policy Under Zapatero -The EU Policy of the Socialist Government 2004-2008," in American Political Science Association (Toronto, Canada 2009). 234 Spaniards have historically been sceptical towards cooperation with the US. Originating from the military defeat of 1898, traditional anti-Americanism was reinforced by the events of the 20th century. In contrast with other European countries, Spain had not been liberated from fascism by US troops, nor had it been a part of the Marshall Plan. On the contrary, the Franco regime had had US support. Spain's mild anti-Americanism became visible in the discussion on NATO membership referendum. While consensus on EC membership was widespread as I will discuss below, for the left-wing political parties NATO membership became problematic. The PSOE government in the 1980s, under Felipe Gonzalez, accepted to support the NATO membership only if it was a prerequisite for the EC membership. For a detailed discussion see Closa and Heywood, Spain and the European Union, 18.; Otto Holman, Integrating Southern Europe: EC Expansion and the Transnationalization of Spain (London: Routledge, 1996), 96-124. 235 Font and Rodriguez, "The Spanish Referendum on the EU Constitution: Issues, Party Cues and Second-Order Effects." 87

February 2005 asked the public to specify the most pressing problems in Spain. The problems mentioned, in decreasing order of significance, were unemployment (56-59%), terrorism and ETA (45-46%), housing (20%), immigration (17-21%), crime (16-18%), economic problems (9-11%), drugs (6-9%), political problems (6-7%), and the public health system (5-6%).236 Eurobarometer data from fall 2004 found similar concerns. When asked about the two most important problems facing their country, the Spanish selected terrorism first and unemployment second, followed by immigration, crime, housing, the economic situation, and rising prices.237 Overall, these problems suggest that the public had concerns that could have constituted potential mobilizing ground for the No campaigners. The No campaigners could have tapped into problems relating to unemployment, immigration and crime to mobilize the society by linking these issues to the referendum proposal. As the figures demonstrate, Spanish society was sensitive in these issue areas. While terrorism could be added to the list of potential No campaign materials, this would have been harder to link to the EU. As I explained, the PSOE used the ‘return to Europe’ card strongly by associating Aznar’s pro-US policy with the recent terrorist attacks. In the other three referendum campaigns, welfare state and immigration were the two main issues that were raised by the No campaigns. Such arguments blamed the TCE for deteriorating the welfare state or for increasing immigration, and resonated well as these three cases had contention around these two topics. Concerning the first, the Spanish No campaign adopted the European anti- globalization movement’s main criticism that the TCE was not providing sufficient social protection. Because Spanish society had relatively fewer problems in this domain, these arguments were not very successful. Regarding immigration, there was no right-wing No campaign in Spain. This issue was therefore not brought up in the campaign, which arguably had the potential to mobilize people.

236 The percentages from the June 2004 and February 2005 surveys are used for illustrative purposes, and for the months in between the numbers remain within the same range. All surveys can be found at: http://www.cis.es/cis/opencms/EN/1_encuestas/estudios/anual.jsp 237 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 88

Concerning the welfare state model, the Spanish welfare state only began to be constructed after the transition to democracy in the 1980s, in an attempt to mirror the European social and democratic model.238 As opposed to the other three referendum countries, which traditionally have had very strong welfare systems, Spain is still in the process of building its up. While the right-wing PP government between 1996 and 2004 prioritized economic liberalization over social spending, this period still saw some limited expansionary measures.239 The most important restriction to the welfare model was the reform of unemployment subsidies, which could relate to the high-ranking of unemployment problem in the polls. Therefore, as I will discuss in greater detail below, the Spanish No campaign’s arguments blaming the TCE for not being social enough did not resonate much. Spanish public opinion on immigration is more complicated. As shown above, the public listed immigration as one of the top five problems. Spain became an immigration country in the 1980s, although the number of foreigners remained below that in other countries such as Germany, France, or the Netherlands.240 The proportion of the population born abroad is relatively small in Spain, at around 4%.241 Yet the number of non-EU immigrants is increasing rapidly; by 2003 their proportion has risen to 65%.242 In contrast with other European states, Spaniards discount the notion that their country’s national

238 Paloma Villota Gil-Escoin and Susana Vazquez, "The Welfare State in Spain: Unfinished Business," in The Handbook of European Welfare Systems, ed. Klaus Schubert, Simon Hegelich, and Ursula Bazant (Abingdon: Routledge, 2009). 239 Further, the reform of the pension system and the labor market followed the general reform trends across the EU to gain efficiency and sustainability of social protection programs. See Joseph Harrison and David Corkill, Spain: A Modern European Economy (Aldershot: Ashgate, 2004).; Ana Guillen, Santiago Alvarez, and Pedro Adao e Silva, "Redesigning the Spanish and Portuguese Welfare States: The Impact of Accesion into the European Union," in In Spain and Portugal in the European Union: The First Fifteen Years, ed. Sebastian Royo and Paul Manuel (London: Frank Cass, 2003). 240 Antonio Escribano, "The EC and Spanish Immigration Policy," in Spain and EC Membership Evaluated, ed. Amparo Barbado (London: Pinter Publishers, 1993), 293. 241 This number is composed mainly by Latin Americans, Moroccans and Europeans. Sjef Ederveen and Paul Dekker, Destination Europe: Immigration and Integration in the European Union (The Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis, 2004), 35. 242 Wayne Cornelius, "Spain: The Uneasy Transition," in Controlling Immigration: A Global Perspective, ed. Wayne Cornelius, et al. (Stanford: Stanford University Press, 2004), 388. 89 identity is threatened by immigration.243 An interesting measure in this regard is ‘ethnic distance’ by the European Social Survey. It measures the degree of resistance to social interaction with members of ethnic groups in various domains such as residential neighborhood, work, and personal relationships.244 The ethnic distance in Spain is lower than the EU average, and lower than the French and Dutch figures. Similarly, Spain does not fit into the rising far right voting in Western Europe; there is no far right political party in the parliament. As such, Spanish society has not experienced a xenophobic backlash to the increasing immigration. Relatively speaking, immigrants have not yet overwhelmed social services, or caused unemployment by competing against Spaniards for desirable jobs.245 Nevertheless, the public locates immigration among the top five problems in Spain. Surveys show that Spaniards are hardening their opinion since the late 1990s with the increasing size and visibility of the non-EU immigrant population as well as rising media attention to negative immigration events.246 Using the monthly CIS surveys conducted between 1998 and 2005, Perni shows that Spaniards became more concerned about immigration specifically during Aznar’s government.247 The early 2000s also saw several killings by immigrants of North African origin, which received significant media coverage and were followed by racist violent manifestations.248 In May 2004, following the Islamist terrorist attacks of March 2004, the Elcano Royal Institute barometer investigated the perception of Muslims in Spanish society, and found that Islam was increasingly being associated with religiosity, authoritarianism and violence.249 Indeed, since 2000, PP politicians have used ‘anti-immigration’ appeals, and the Aznar government revised the national immigration law in 2001,

243 Ibid., 391. 244 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 74. 245 Cornelius, "Spain: The Uneasy Transition," 390. 246 Ibid., 421. 247 Orietta Perni, "Migration in Spain from Aznar to Zapatero: A Real Change of Direction?," in International Studies Association 47th Annual Convention (San Diego, California, USA). 248 Ibid. 249 The barometers are available at: http://www.realinstitutoelcano.org/wps/portal/rielcano_eng/BarometeroftheRIElcano 90 restricting the labor rights and civil liberties of immigrants.250 Associating immigration with crime, social conflict and imported poverty, Aznar gave his restrictive policy an EU ‘cover’, presenting it as complementary to the EU’s attempts to control unwanted immigration. Therefore, immigration – unlike the welfare state – was a sensitive issue for Spaniards and provided potential campaign material for the No campaign. In sum, the main problematic issues of 2004 in the domestic sphere were unemployment, terrorism (concerning both ETA and the Islamist attacks reacting to the Spanish involvement in the Iraq War), immigration and crime.251 These issues were potentially useful for the No campaign. If they blamed the TCE for these existing problems, such arguments would have resonated easily. Terrorism would not have been easily linked to the referendum proposal as the public associated the terrorist bombings in Madrid with the pro-US policy of the Aznar government. The PSOE tapped successfully into this contention in its election campaign by using the ‘return to Europe’ argument, and later on once again at the 2005 TCE referendum campaign. However unemployment or immigration formed potential No campaign material, which were not used by the No side. Instead, in line with the European anti-globalization movement, the Spanish No campaign raised welfare state-related issues, which were not contentious in Spain as much as they were in the other three cases.

Contention concerning the EU Spain is the only one of the four case-study countries that is not a founding member of the EU, having joined the Union in 1986. Shortly after General Franco’s death in 1975, the transition to democracy began and Spain applied for membership in the European integration project. The Spanish transition to democracy and economic modernization therefore became closely associated with

250 Wayne Cornelius et al., eds., Controlling Immigration: A Global Perspective (Stanford: Stanford University Press,2004), 35-36. For a detailed analysis of Aznar’s migration policy, see Perni, "Migration in Spain from Aznar to Zapatero: A Real Change of Direction?." 251 According to OECD unemployment statistics, Spain’s unemployment rate has declined from 17.8% in 1996, to 10.6% in 2004. Within this period the lowest rate was 10.3% in 2001, followed by 11.1% in 2002-2003. The statistics are available at: http://stats.oecd.org/index.aspx?queryid=21760 91 the EU, leading to relatively positive attitudes towards European integration.252 Furthermore, Spain has been a net recipient of EU funding, particularly due to EU redistributive policies such as the structural funds designed to contribute to the development of the poorer regions in Europe, or the agricultural, training and cohesion programs. In absolute terms, Spain is the country that has benefited the most from structural and cohesion funds; for example, in 2001 it was allocated nearly 63% of the EU’s structural funds budget.253 Table 3.1 summarizes Spain’s benefits and contributions between 2000 and 2005:254

Total benefit from the EU Total contribution to the EU (Euro million) (Euro million) 2000 10,803.7 5,527.5 2001 13,557.9 5,686.9 2002 15,210.9 5,965.9 2003 15,875.1 6,676.2 2004 16,356.7 7,428.9 2005 14,823.6 8,377.1 Table 3.1: Spain’s benefit and contribution in the EU budget (2000-2005)

Apart from the evident budgetary benefits, the EU has also influenced Spain’s welfare state. Spain has strengthened its financial and budgetary capacity and its social expenditures, in order to meet European standards.255 The funds received

252 Closa and Heywood, Spain and the European Union, 15. See also Juan Carlos Castanares and Antonio Juste, "Spain: In the Center or on the Periphery of Europe?," in Southern Europe and the Making of the European Union, ed. Antonio Pinto and Nuno Teixeira (New York: Columbia University Press, 2002). 253 Sebastian Royo, Varieties of Capitalism in Spain: Remaking the Spanish Economy for the New Century (New York: Palgrave Macmillan, 2008), 48. 254 Detailed information can be reached at the EU’s official website on financial programming and budget. Available at: http://ec.Europa.eu/budget/budget_glance/where_from_en.htm 255 In order to adapt to the prevailing model in the European Community, Spain deeply reformed its fiscal policies (both direct and indirect taxation) which helped in the introduction of universalizing measures in the 1980s. This move towards a welfare state model was seen as the social aspect of democratization, and was labeled the ‘European Project’. See Guillen, Alvarez, and Adao e Silva, "Redesigning the Spanish and Portuguese Welfare States: The Impact of Accesion into the European Union," 245.; Royo, Varieties of Capitalism in Spain: Remaking the Spanish Economy for the New Century, 50.; Ana Arriba and Luis Moreno, "Spain - Poverty, Social Exclusion and 'Safety Nets'," in Welfare State Reform in Southern Europe: Fighting Poverty and Social Exclusion in Italy, Spain, Portugal and Greece, ed. Maurizio Ferrera (Abingdon: Routledge, 2005). Ariba and Moreno suggest that high rates of unemployment, demographic imbalance and decline of domestic care threaten the stability of the welfare system. 92 from Europe have also helped in developing social policies and constructing related infrastructures. How do Spaniards perceive Europe? In extensive research on Spanish attitudes towards European integration, Medrano finds several key themes:256 Modernization, status, and opposition to isolationism. The Spanish public believes that membership in the EU contributes to the modernization of the country, enhances Spain’s international prestige, and breaks with Spain’s tradition of isolation. It is not surprising to see a combination of these three themes because since the Second World War, Spaniards have witnessed their European neighbors develop at a rapid pace within the European integration project, whereas Spain remained a traditional rural society under Franco’s authoritarian rule.257 As such, Europe symbolized modernity and democracy. Today most Spaniards are highly satisfied with EU cohesion and the structural funds that have been flowing into Spain. These funds enjoy great visibility as all the infrastructural projects carry large signs mentioning ‘built with European funds’. An analysis of the Eurobarometer polls shows that Spaniards have a positive attitude towards identifying themselves as both Spanish and European, 10% above the EU average.258 Yet the perceptions of membership benefits have been fluctuating. While 1986-1991 saw enthusiasm, 1992-1996 were years of disenchantment, followed by 1997-2003 which were characterized by rather conditional support for the Union.259 The literature suggests several reasons for this fluctuation.260 First, there seems to be a clash between deep-rooted support for the Union and more instrumental considerations regarding the consequences of the free circulation of goods and workers. I discuss these concerns below. Second, there is also a correlation between the economic cycle and these instrumental considerations, mediated through national priorities. As in other

256 Juan Diez Medrano, Framing Europe: Attitudes to European Integration in Germany, Spain, and the United Kingdom (Princeton: Princeton University Press, 2003), 43. 257 Ibid., 44. See also Castanares and Juste, "Spain: In the Center or on the Periphery of Europe?." 258 Closa and Heywood, Spain and the European Union, 33. 259 Antonio Menendez-Alarcon, The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom (Westport: Greenwood Publishing Group, 2004), 58. 260 Closa and Heywood, Spain and the European Union, 39. 93 member states, domestic factors such as unemployment rates and general satisfaction with democracy are shown to be empirically correlated. These fluctuations suggest that the support for the Union is not unconditional and that concerns in the domestic and European domains can be linked. But the overall Spanish attitude to the Union is mainly above the EU average. Figures 3.2 and 3.3 show the trends in the Eurobarometer surveys:261

Figure 3.2: Spanish support for EU membership (1984-2004)

261 Eurobarometer, "Standard Eurobarometer 61: Public Opinion in the EU, Spring 2004," European Commission: Brussels (2004). 94

Figure 3.3: Spanish public’s perceived benefit from EU membership (1984-2004)

There are certain EU-related issues that Spaniards consider problematic. Spain’s net benefit from EU funding was affected by the 2004 enlargement of the Union towards poorer countries of Eastern Europe.262 The public feared lower EU subsidies for farmers, a loss of social benefits, and that the EU might ban the ‘ñ’ in the Spanish alphabet or bull-fighting.263 The Eurobarometer fall 2004 survey, which asked citizens of every member state about their fears concerning the building of Europe, also shows similar concerns.264 Paralleling the public in the three other cases, Spaniards stated mild concerns relating to the transfer of jobs to other member countries which have lower production costs, more difficulties for Spanish farmers, and their budget contribution. Further, most Spaniards are

262 For a detailed analysis see José Ignacio Torreblanca, "Farewell to Funds? Keys to Understanding Spain’s Position When Negotiating the 2007-13 EU Budget," Elcano Royal Institute Working Paper 21 (2005). 263 Closa and Heywood, Spain and the European Union, 33. See also Medrano, Framing Europe: Attitudes to European Integration in Germany, Spain, and the United Kingdom, 45. Medrano specifies the Common Agricultural Policy as a source of dissatisfaction as well, due to the quotas on certain farm products. 264 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 95 concerned that Spain has little say in the Union.265 Finally, in the Elcano Royal Institute’s November 2004 survey only 44% of the respondents supported Turkish accession to the EU.266 Potential Turkish EU membership was a crucial issue (both in cultural and economic terms) in the three other referendum campaigns. These poll results suggest that there could have been the mobilization ground for similar arguments in Spain, especially given the recent sensitivity concerning immigration and the negative impact of enlargement on Spain’s benefits. These contentious issues, as in the domestic field, could have served as potential No campaign arguments.

Political Party Attitudes towards Europe Looking at the political parties, Spain parallels the rest of European countries where Euroscepticism is confined to the political extremes. Furthermore, as opposed to France and the Netherlands, European integration has not led to significant internal splits within parties.267 The two main political parties, the PSOE and the PP, have always supported Spain’s membership in the EU.268 In recent years, under Aznar’s leadership, the PP debated certain EU policies but this mild opposition never turned into open Euroscepticism. The only party which has a Eurosceptic tendency is the United Left (IU), which is a coalition built around the Communist Party of Spain (PCE).269 The PCE was in favor of Spain’s EU membership, however over time, it has started to defend the social dimension of

265 Menendez-Alarcon, The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom, 69.; See also Medrano, Framing Europe: Attitudes to European Integration in Germany, Spain, and the United Kingdom. 266 The barometers are available at: http://www.realinstitutoelcano.org/wps/portal/rielcano_eng/BarometeroftheRIElcano 267 Closa and Heywood, Spain and the European Union, 36-37. 268 See Menendez-Alarcon, The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom, 61. Alarcon argues that the PP is relatively divided on the EU, while being supportive of the EU membership, the party members voice concerns about going too far in the integration process. 269 The Basque Independentist Left (ETA) and the minor radical right Spanish Falange groups can be seen as hard Eurosceptics but they did not play any role in the 2005 referendum campaign. For a detailed discussion see Marga Gomez-Reino, Ivan Llamazares, and Luis Ramiro, "Euroscepticism and Political Parties in Spain," in Opposing Europe? The Comparative Party Politics of Euroscepticism: Volume I, ed. Aleks Szczerbiak and Paul Taggart (Oxford: Oxford University Press, 2008). 96

European integration and criticize the Economic and Monetary Union (EMU) for generating market-friendly policies.270 There are also regional parties in the Spanish political system. All of these parties see the EU as a new political arena and resource for their goals, as the EU regional policy empowers regions by assigning an active political role to them and by providing structural funds for the development of poorer regions. However, the parties' ideologies affect how opposed they are to the market-friendly policies of the Union.271 Neither the Catalan Nationalist Convergence and Unity nor the Basque Nationalist Party have problems with the free-market focus of the EU policies as they are both located at the center-right of the political spectrum with a Liberal/Christian Democrat ideology. The other regional parties that are left-wing and nationalist (in the regional sense) such as the Galician Nationalist Block, the Republican Left of Catalonia, the Initiative for Catalonia Greens, and the Basque nationalist social democratic Basque Solidarity (only in the regional parliament), have occasionally criticized the economic dimension of the European integration.272 These Eurosceptic patterns were reflected clearly in the TCE referendum campaign. Overall, the Spanish public was satisfied with domestic and European governance, with generally positive attitudes towards European integration. Yet, as I have shown, there were problematic issues both in the domestic and European domains which could have been used to mobilize the Spanish public against the TCE. Concerning the domestic sphere, the issues consisted mainly of

270 Concerned that the welfare state model was declining, the IU explicitly opposed the Maastricht Treaty in 1993. This was the first time Euroscepticism became evident in Spanish party politics. Similarly, despite being strongly committed to the integration project, the two main trade unions (the socialist UGT – General Union of Workers, and the communist CCOO – Workers Commissions) have been critical of the Union’s market policies since the Maastricht Treaty. See Closa and Heywood, Spain and the European Union, 51.; Menendez-Alarcon, The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom, 63. 271 Closa and Heywood, Spain and the European Union, 52. See also Montserrat Guibernau, "Nations without States in the EU: The Catalan Case," in European Integration and the Nationalities Question, ed. John McGarry and Michael Keating (Abingdon: Routledge, 2006).; Gurutz Jauregui, "Basque Nationalism: Sovereignty, Independence and European Integration," in European Integration and the Nationalities Question, ed. John McGarry and Michael Keating (Abingdon: Routledge, 2006). 272 Menendez-Alarcon, The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom, 63. 97 unemployment, immigration and crime; concerning the EU domain, the loss of social benefits due to the costly enlargement process, subsidies for farmers, and issues such as a potential ban on ‘ñ’ in the Spanish alphabet or bull-fighting formed potential No campaign material. For instance, polls show that a rather small section of society already associated the unemployment problem to the EU. In 2000, 18% believed that the Union contributed to increased unemployment.273 However most people mentioned that the EU was not helping solve this problem: 39% stated having little trust in the EU in this regard, and 16% stated having no trust at all.274 In other words, the No campaign could have capitalized on these concerns. Below, in the analysis of the campaign, I discuss why these issues did not become part of the debate on the TCE.

The Long Campaign In this section, I discuss the strategies and framing of the Yes and No campaigns. In order to analyze the campaign, I will consider the agenda-setting and priming/framing stages in turn, based on personal interviews with the campaigners, campaign materials, television clips, and party strategy papers.275 I will show that the No campaign was weak despite the availability of contentious issues with which to work (with the exception of the regions), and that as a result, the Yes campaign had a relatively easy time sustaining the already positive public opinion towards the TCE. The No campaign did not use strong frames. It did not have strong mobilizational structures either as the Yes side had much better resources. Below, I first provide a brief overview of the campaign and the campaigners. Prime Minister Zapatero announced in July 2004 that a referendum would be held, but did not set the 20 February date until January 2005. The campaign activities began in November but the real Spanish campaign was short as it did

273 Ibid., 68. 274 Ibid. 275 A full list of the campaigners I interviewed is presented at the end of this chapter. 98 not get intense until the last two weeks.276 The referendum campaign was thus low profile. The various parties supporting both the Yes and the No votes all chose to run individual campaigns rather than uniting their forces, a general weakness on both sides that in practice, advantaged the status-quo-oriented Yes campaigners. The Yes campaign was formed by the political mainstream: The PSOE, the PP, the CiU, the PNV, and the CC. To build positive momentum, the PSOE government wanted to be the first EU member to hold a referendum on the treaty, and was a strong advocate throughout the process.277 But far from campaigning together, the two main parties, the PSOE and the PP, criticized each other. Mariano Rajoy, the leader of the PP, criticized the Prime Minister for rushing into a referendum without taking the time to inform the Spanish electorate, in order to be the first in Europe to vote on the Treaty.278 The PP’s critical attitude was also confirmed in my interviews. PP Senator Alejandro Muñoz Alonso suggested that the PP had not supported the referendum idea because it was premature and was on a very complex text.279 In return, the PSOE chided the opposition party for their tentative campaign for a Yes.280 The PP's lack of enthusiasm was mentioned many times by the other political campaigners in my interviews. The PP initially remarked critically on the absence of any reference to Christianity and Spain’s decreasing veto power in the Council.281 These issues could have formed potential No campaign material, yet they disappeared from the agenda quickly as neither the PP nor the other parties pursued them. Despite their official Yes campaign, some leaders of the PP hinted their preference for a negative vote whereas Aznar never stated his support for the TCE. Instead of a genuine opposition towards the TCE, this could be read as a strategic move to avoid supporting the PSOE

276 Corinne Deloy, "Referendum on the Ratification of the European Constitution in Spain," Robert Schuman Foundation Report (2005). 277 Juan Carlos Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," Mas Democracia and Democracy International Monitoring Reports (2005): 4. 278 Quoted in Deloy, "The Spanish Approve the European Constitution by a Wide Majority." 279 Personal interview, 2 October 2008. 280 Quoted in Deloy, "The Spanish Approve the European Constitution by a Wide Majority." 281 Closa, "Business as Usual: EU Policy Under Zapatero -The EU Policy of the Socialist Government 2004-2008." See also Arantxa Capdevila, Lorena Gómez, and Laia Aubia, "Partisan Strategies in Light of the European Constitution in Spain," EpsNet Electronic Journal 5 (2006). 99

government. Given these tensions, the Yes campaign was far from being united. The PSOE held around 700 public meetings before the vote, while the PP held around 500.282 On the No side, the campaigners were the main far left party IU, and regional green/far left parties such as the ERC, the ICV, the BNG, the EA, and the CHA. The IU was the main nation-wide No campaigner, as the regional parties concentrated their campaign in their respective regions. While the No arguments were coherent as a whole, each political party campaigned on its own. Only the civil society No campaigners formed a common platform, which I will discuss in detail below. Apart from the Yes and No campaigns, the government ran an official information campaign, aiming to inform the citizens on the content of the treaty. This information campaign had major resources and was very visible. The rest of the campaigns, both the Yes and No campaigns, were not high-profile mainly because the political parties had had two major campaigns in the last year: the March 2004 general elections and the June 2004 EP elections.283 After having heavily invested in those campaigns, the parties did not devote significant resources to the TCE campaign.

Agenda-setting Unlike the other three states in this study, Spain did not have a long campaign. Timing is an important tool for campaigners since getting an early start can allow the first mover to define the terms of the debate. The Yes and No campaigns began preparing for the campaign in November 2004. But the campaigns were not highly visible at this stage. In the other three cases the No campaigners started earlier than the Yes campaigns and thereby managed to set their themes on the agenda. The Spanish No campaign neither had an early start nor did they have sufficient resources. As a result, they could not define the debate around their

282 Corinne Deloy, "Referendum on the European Constitution in Spain on 20th February: An Assessment Just a Few Days Before the Election," Robert Schuman Foundation Report (2005). 283 Personal interview with Orestes Suárez Antón, the international secretary of the PSOE, 3 October 2008. 100

themes such as the deterioration of the welfare state or the militarization of Europe. They could not force the Yes campaigners to take the defensive position.

Priming and Framing I have previously argued that three factors matter in the priming and framing stage: the strength of the campaign frames, the credibility of the speakers, and mobilizational structures. The campaign was of low intensity, and neither side was highly successful in priming their themes in the media. The media content analyses in the next section will demonstrate that the Yes campaign primed their campaign themes relatively better. In terms of framing, the No campaign suffered both from weak frames and weak financial resources/civil society support. The Yes campaign not only had stronger frames, but better resources.

Content of the campaign frames According to the definition adopted from the framing literature, the Yes frames were stronger than the No frames. In this section, I explain the content of the Yes and No campaign messages followed by a discussion of their strength. The Yes campaign put forth both constitutional and instrumental arguments.284 Such instrumental arguments rendered the pro-TCE frames stronger than the anti-TCE frames. Below, I list the Yes and No campaign frames. I identified the campaign frames primarily based on written campaign materials.285 Next, I categorized them according to their subjects such as peace, economy, democracy, institutions, and other salient issues like welfare state, immigration, or regional languages. The main Yes campaign frames can be categorized as: 1. The TCE represents a fundamental step towards European unification. (This frame was used in combination with ‘bringing Spain back into Europe’ after the pro-US foreign policy of the Aznar government.)

284 Closa, "Business as Usual: EU Policy Under Zapatero -The EU Policy of the Socialist Government 2004-2008." 285 Where available, one to two page leaflets listing the campaign arguments provided the best source. If not, I used interview data and secondary sources to complement the available materials. 101

2. The TCE is an important step forward in the European integration, which brought clear benefits to Spain, such as democratic consolidation, peace, prosperity, economic growth, trade, civil and social rights, and regional funds. Thus, the TCE would be in Spain’s national interests. 3. The TCE institutionally enhances the integration project, especially with a common bill of rights for EU citizens. It would further a common foreign policy, environmental policy, and immigration policy, along with majority voting and fight against terrorism at the European level. 4. The TCE includes measures that would bring the Catalan language into EU institutions. The PSOE, the main Yes campaigner, used the first three frames strategically. Its main slogan was ‘Se trata de Europa’, meaning that the referendum was about Europe. The logic was to present the referendum as more than just a simple judicial step; instead it was an essential step for European unification, without which Europe would be weak and divided. The key argument of the PSOE campaign was therefore to confirm Spain’s return to the heart of Europe. Zapatero used this argument to emphasize the switch from his predecessor Aznar’s pro-US policy.286 In the interviews this strategy was mentioned widely. This first frame was then combined with the second frame, the benefits received from Europe, in the following way: Zapatero stated that a positive vote would be a contribution to the construction of Europe, ‘to which we owe so much and from which we have also benefited so much’.287 He also declared, ‘we know how we were and we imagine how we would be if we had not entered the Union in 1986’.288 The third frame paralleled that of the Yes campaigners in the three other cases, and highlighted the TCE as an institutional step towards a better Europe. Hence this argument had more technical content. Nonetheless, the PSOE strategy paper explicitly discusses the need to focus mostly on the first two frames as they

286 Quoted in Deloy, "The Spanish Approve the European Constitution by a Wide Majority." 287 Quoted in Ibid. 288 Quoted in ———, "Referendum on the Ratification of the European Constitution in Spain." 102

concern matters that interests citizens the most. In turn, it warns against using bureaucratic and institutional arguments that only interest policy-makers. Driven by this concern, the PSOE chose to build the third frame mainly around the idea of common citizenship rights for the EU citizens, promoted by the Charter of Fundamental Rights. As such, they stressed the common rights the TCE was introducing for all citizens of Europe. Some of the PSOE campaign material showed these three frames clearly: ‘Se trata de Europa, se trata de España, se trata de tus derechos’ (It is about Europe, it is about Spain, it is about your rights). The PP, in its less rigorous campaign, used the second and third frames. It used the slogan, ‘Sí a Europa porque Europa Sienta bien’ (Yes to Europe because Europe feels good), referring to all the benefits Spain has received from Europe. Nonetheless the PP television commercial clearly reflected the party’s reservations as well. Although calling for a Yes vote, the party leader Rajoy openly stated that the TCE was ‘probably not the best possible treaty’. The pro-TCE regional parties, the Basque PNV and the Catalonian CiU, adopted pro-TCE stances only after internal debates. The PNV initially announced that although its vote would be positive, its position was not favorable.289 PNV Member of Parliament (MP) José Ramon Beloki Guerra acknowledged that there was an internal discussion because the TCE did not recognize Basque self-determination rights in Europe.290 Yet, he emphasized the Basque movement’s commitment to Europe. The PNV focused on the second and third frames, using the slogan ‘Más Europa, más Euskadi’ (More Europe, more Basque Country). The same slogan type was adopted by other regions as well, suggesting that Europe was increasing their region’s power at the expense of Spain’s by hinting at ‘less Spain’.291 Similarly, CiU MP Jordi Xuclà i Costa as well as the international secretary of the CDC Carles Llorens i Vila stated that their party had initial concerns with the expression of Catalan identity in the TCE, but explained that during the treaty negotiations, they were able to secure the right

289 Jauregui, "Basque Nationalism: Sovereignty, Independence and European Integration," 256. 290 Personal interview, 13 October 2008. 291 As a liberal regional party, Canarian Coalition also adopted ‘Más Canarias, Más Europa’ (more Canary Islands more Europe) as its slogan. 103 of petition (to the EU institutions with a request or complaint) in Catalan and the right to speak Catalan in some European committees.292 They remarked that this switch formed the basis of the CiU support for the TCE, and that they used this argument (the fourth frame) in their campaign as the language issue was important in Catalonia. Accordingly, the CiU combined the second, third and fourth frames, and campaigned with the slogan, ‘Sí Ara. Més Europa, Més Catalunya’ (Yes now, more Europe, more Catalonia). On the other side, the No campaign frames can be categorized as follows: 1. The TCE includes market-friendly provisions without sufficient social rights and protection. 2. The TCE does not recognize the self-determination of European nations and does not introduce a federal Europe. 3. The TCE is not democratic enough. 4. The TCE is not environmentally friendly. 5. The TCE subordinates the EU’s security system to North Atlantic Treaty Organization’s (NATO) and militarizes Europe by increasing military expenditures. 6. The TCE promotes a Fortress Europe with restrictive immigration policies. The main No campaigner, the IU, used all six frames in its campaign. It campaigned with the slogan ‘Otra Europa es posible, No a esta constitución’ (Another Europe is possible, No to this constitution). As the slogan shows, it tried to stress its pro-EU yet anti-TCE stance. This was a major concern for all the No campaigners. The left-wing anti-globalization civil society organizations, which formed a common platform ‘No to the Constitution’, also used all of these frames. This platform included various groups such as the ATTAC, Alternative Space, Red Current and Ecologists in Action. They adopted the same slogan as the IU. Catalan and Basque regions have been historically pro-EU, expecting the integration project to weaken the Spanish state and thus strengthen and recognize their regions as nations. Yet political parties from both regions found problems

292 Personal interviews, 8 October 2008 and 15 October 2008, respectively. 104

with the TCE, as it did not provide the expected self-determination rights or recognition for their regional languages. In the Convention process that drafted the TCE, the Catalan Convention specifically demanded such recognition, and the lack of this acknowledgement became a source of contention concerning the TCE. Similarly the historically pro-EU Basque Country also had problems with the lack of recognition for Basque language Euskera in the TCE.293 Among the regional parties, the Catalan left-wing nationalist ERC focused on the first three frames, prioritizing the assertion of Catalan identity. Using the slogan ‘Per moltes raons aquesta constitució No’ (For many reasons No to this constitution), the ERC emphasized that the petition rights the CiU secured were not sufficient. The Basque nationalist social democratic EA similarly stressed the second frame for the Basque identity. The Catalan Greens ICV combined the first five frames, and said ‘Cap a la dreta No, No a aquesta constitució Europea’, emphasizing their opposition to this particular constitution as it was a move towards political right. The Galician BNG followed the same left-wing regional pattern and adopted a mix of all six frames, using the slogan ‘É posíbel unha Europa mellor’ (Another better Europe is possible).294 A striking element of the Spanish No campaign in comparison to the other cases was the absence of far right anti-immigrant frames. On the contrary, the Spanish No campaigners emphasized regional identities and advocated for a federal Europe in which the Spanish state would be weaker. This was in clear contrast with the other three referenda campaigns. For instance, in the Dutch campaign, the far left No campaigners wrapped their left-wing criticism around the loss of national sovereignty theme. Besides the Yes and No campaigns, the government held an information campaign. Even though this campaign was officially only ‘informative’, it was de facto part of the Yes campaign. The government’s official campaign slogan was, ‘Los primeros con Europa’ (The first with/in Europe), which became

293 Jauregui, "Basque Nationalism: Sovereignty, Independence and European Integration," 256. 294 Along the same lines, the Aragonese Council CHA used the slogan ‘Sí A Europa, Pero No a Este Tratado’, Yes to Europe but no this treaty. 105

controversial due to complaints questioning its impartiality.295 It was argued that the wording implicitly assumed a positive result. While the Electoral Commission accepted this complaint and stated that the government should remain impartial, it did not ban the government from using that slogan.296

Relative strength of the campaign frames Which of these frames were stronger according to the strength definition of the framing literature? As explained in Chapter 2, frames that involve available, accessible, and applicable concerns are more effective. Furthermore, vivid, concrete, image-provoking, emotionally compelling frames that contain negative information are most successful in affecting individuals’ opinions by increasing the salience of the particular dimension they emphasize. Both the first and second frames of the Yes campaign centered on clearly accessible, available and applicable concerns for Spaniards. The PSOE’s ‘return to Europe’ theme was an existing concern in society as the public linked the Madrid bombings to Aznar’s pro-US policy. Mild anti-Americanism and the belief that Aznar’s pro-US policy led to the terrorist attacks helped the PSOE mobilize its electorate only one year before the referendum. Similarly, due to the strong public association of the EU with democracy and prosperity, the second frame emphasizing the benefits that Spain received from Europe was also highly accessible, available and applicable. The first two Yes frames were also successful because they invoked vivid, concrete, image-provoking, and emotionally compelling features. The first frame stressed a return back to Europe from a detour that had caused terrorist attacks. The attacks concretely demonstrated why Europe was needed. Relating to the fear of terrorism, it was emotionally compelling as well. The second theme, similarly used concrete examples of how Spain materially benefited from the Union. For a society which strongly associates democracy and prosperity with the Union, this frame could also tap into the negative memory of the Franco regime.

295 Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," 5. 296 Ibid.: 5-10. 106

The third Yes frame referred to the technical and institutional aspects of the draft treaty and presented the TCE as an institutional step forward in the integration project. As such, this frame was abstract and difficult for the public to relate to, as most people were not familiar with the technical and complex nature of the EU. The PSOE’s strategic solution to this problem was to emphasize common EU citizenship rights within this frame, which were being introduced with the Charter of Fundamental Rights. Only this emphasis on common citizenship rights provided a concrete example to the public. The final frame concerning the new presence of the Catalan language in EU institutions was an undoubtedly contentious and concrete issue in Catalonia. On the other hand, the No campaigners were not as able to tap into accessible, available, and applicable concerns. Of the six No frames, the second frame was the strongest according to the relevant criteria. This frame dealt specifically with the issue of regional identities and languages, and many Catalans and Basques fear an EU which ignores stateless nations. Despite being the official language of three regions and the tenth most spoken language within the Union, Catalan is not an official EU language.297 The language issue has been a central concern for the otherwise strongly pro-European Catalonians, and led to fears about the TCE as it did not recognize the Catalan language.298 This issue was similarly an important concern for the Basque identity. This frame was hence concrete and emotionally compelling for Catalonians and Basques. Moreover, the framing literature argues that frames containing negative information are more effective than those containing positive information. As it blamed the TCE for not recognizing the Catalan identity, this frame was more influential than the positive corresponding Yes frame used by the CiU. The rest of the frames in the No campaign did not center on particularly important contentious issues in Spanish society. Neither of them built on the problems specified in the previous sections such as domestic issues as unemployment, immigration, crime or other EU related concerns as the costly

297 Guibernau, "Nations without States in the EU: The Catalan Case," 219. 298 Ibid., 220. 107

enlargement process, subsidies for farmers, or issues such as a potential ban on ‘ñ’ in the Spanish alphabet or bull-fighting. The No frames carried negative information but these frames did not address emotionally gripping topics. The No campaign tried to use some concrete examples, such as the possibility of death penalty in the new treaty, or the increasing of military expenditure. Another example was a count of certain words in the TCE. While the terms ‘free market’ and ‘competition’ existed more than 300 times, the word ‘social’ was used only about 15 times.299 Yet these examples did not tap into existing concerns, as neither a decline in the welfare system, nor environmental issues were major controversial issues. Immigration was indeed a sensitive issue for some Spaniards but the No frame on immigration was not anti-immigrant. Quite the opposite; it favored less control on immigration. Overall, the No campaign frames were weaker, apart from the second frame on recognition of regional identities. The campaign materials – such as posters and leaflets – of the Yes and No sides reflected their arguments in a simple and dull manner. Figures 3.4-3.10 below show the campaign posters. Only the posters in Figures 3.9 and 3.10, used by the Catalonian No campaigns (the Catalonian left-wing nationalist ERC and the Greens ICV), were more concrete. They stated several reasons to vote against the TCE, and presented the TCE as a move towards the political right.

299 Personal interview with José Manuel Fernández Fernández, 7 October 2008. 108

Figure 3.4: The PSOE’s poster Figure 3.5: The PP’s poster

Figure 3.6: The information campaign’s poster Figure 3.7: The IU’s poster

Figure 3.8: The CiU’s poster

109

Figure 3.9: The ERC’s poster Figure 3.10: The ICV’s poster

However, the No campaign materials were not very visible – except in the regions. The No campaign suffered from a lack of resources, as I will discuss below. Therefore their campaign remained low profile and low-budget. On the other side, the government information campaign, which was commonly blamed for supporting the TCE rather than being impartial, used celebrities such as football stars or artists who read selected articles of the treaty in TV ads, as well as distributing a soft drink called ‘referendum plus’.300 The text of the draft treaty was distributed to citizens through the main national newspapers in January 2005.301 But this text only included Parts I-II of the treaty, excluding the other sections.302 This move was fiercely criticized by the No campaigners, who argued that the most problematic ‘market-friendly’ policies were embedded in Part III. The No campaigners argued that the information campaign primed the more aesthetic parts of the treaty such as the Charter of Fundamental Rights. Interviewees from the No camp brought this up as a problem.303

300 Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," 10.; Closa, "Business as Usual: EU Policy Under Zapatero -The EU Policy of the Socialist Government 2004- 2008." 301 Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," 7. 302 Ibid.: 10. Part I laid out the general provisions, Part II covered the Charter of Fundamental Rights, as opposed to concrete policies. 303 Personal interview with Ricardo Gómez Muñoz, 6 October 2008; personal interview with Jaime Pastor, 8 October 2008. 110

Based on the assessment above, it can be clearly seen that the Yes campaign did not face a challenging No campaign. The No campaign framing was not strategic in the sense that it did not capitalize on the most problematic issues in society such as unemployment, immigration, or crime. Importantly, both the Yes and No campaigners acknowledged that the No campaign did not succeed in building strong arguments. As a result, the Spanish Yes campaign did not share the difficulties the French, Dutch and Luxembourgish Yes campaigners faced. In Spain, with the exception of the regional campaigns, the Yes campaign did not find it difficult to win the arguments. The Yes campaigners shared the idea that the campaign was easy. Juan Fernando Lopez Aguilar, then Minister of Justice and Interior Affairs, stated that the Yes campaign was ‘not risky business’ given all the positive aspects associated with Europe, such as democracy, economic and social progress.304 However he added: ‘this does not mean that we did not have to campaign’, especially referring to the regional No campaign arguments. The Yes campaign’s success was therefore closely related to the lack of strategic No campaign framing. As the regional No campaigns demonstrated, where the anti-TCE frames were strong, the Yes campaign had a harder job. Juan Moscoso del Prado Hernández, MP of the PSOE, mentioned that it was overall an easy campaign because the No side was mainly limited to the IU and that their arguments were not very strong.305 Enrique Baron Crespo, Member of European Parliament (MEP) of the PSOE, having participated in the French campaign as well, emphasized the lack of ‘Polish plumber’, xenophobic arguments and a strongly Eurosceptic communist movement in Spain, in contrast with the French campaign.306 In turn, he characterized the No campaign as lacking grip. Given the lack of strong anti-TCE arguments, the Yes campaigners mainly tapped into the broad positive image of European integration. José Ignacio Torreblanca of the Elcano Royal Institute for International Affairs explained that the yes campaigners ‘appealed to the intuitive nature of what Europe stands for’ and that

304 Personal interview, 16 October 2008. 305 Personal interview, 7 October 2008. 306 Personal interview, 16 October 2008. 111

they did not have a difficult job.307 Orestes Suárez Antón, the international secretary of the PSOE said that the argument ‘return to Europe’ was the key, and that Spaniards knew they received modernization and infrastructural funds from Europe.308 These comments by the Yes campaigners confirm the weakness of the No frames. They also support the above assessment that the first and second frames were the most effective, given the absence of effective anti-TCE framing. The remarks of the No campaigners also parallel the previous framing analysis. Some interviewees, such as the coordinator of the IU parliamentary group, José Manuel Fernández Fernández, stressed that the Spanish context was different from the French or the Dutch ones, as there were less economic problems.309 Yet as I have shown above, Spanish society was not without concerns. Even though it might have been difficult, the No campaign could have drawn upon the above-specified contentious issues. But the No side failed to build strong frames, with the exception of the regional No campaigns. Carlos Girbau Costa of both the Social Forum and the IU, characterized the No campaign arguments as too ideological and ineffective in linking the TCE to people’s lives.310 He added specifically that anti-TCE frames on NATO or the loss of social protection and the privatization of public services did not resonate with people. Marc Giménez Villahoz, the ICV European Politics Coordinator, explained that although – as a Green party – their priority was the social Europe argument, the Catalan language issue was of significant sensitivity for the Catalan people, and that this argument gained more ground in the campaign.311 Similarly, Oriol Duran Torres, the spokesperson of the ERC parliamentary group, stressed that the language argument became the most resonant one among all their arguments.312 Jordi Vaquer i Fanés of the Barcelona Centre for International Affairs (CIDOB) confirmed that the Convention process which drafted the TCE was not popular in Catalonia, as society was highly mobilized and the resulting

307 Personal interview, 6 October 2008. 308 Personal interview, 3 October 2008. 309 Personal interview, 7 October 2008. 310 Personal interview, 8 October 2008. 311 Personal interview, 15 October 2008. 312 Joint personal interview with Joaquim Puig Vilamala, the ERC Coordinator, 13 October 2008. 112

text was disappointing in terms of identity recognition.313 The idea that Catalan people were not taken into account gave the Catalan nationalists a strong mobilizing ground. Apart from the regional self-determination frame, the No campaign frames were weak as they failed to link their arguments to tangible issues and examples from people’s lives. In the other three referendum campaigns, the No frames on two topics – decline in welfare models and increasing immigration – were very strong. This is in remarkable contrast to the Spanish case. Concerning the frames on the deteriorating welfare model, both the Yes and No campaigners specifically stated that they did not resonate well in Spain. The interviewees noted that Spain was catching up with the European welfare model rather than cutting down, as opposed to the other West European countries. Juan Moscoso del Prado Hernández, PSOE MP, highlighted that Spain adopted the welfare model from Europe in the 1980s, unlike the experience of the Dutch or French governments.314 Furthermore he remarked that despite the IU’s criticism of the market-friendly EU policies, nobody could observe a decline in Spain’s social standards. Following the same reasoning, PP MP Ignacio Cosidó Gutiérrez, also emphasized that the Spanish welfare state was weaker than the traditional European model and that Spain identified Europe with the welfare model.315 Similarly, the frames on increasing immigration or the costly enlargement process were absent in Spain. Referring to the more recent experience of Spain with immigration as opposed to France or the Netherlands, most of my interviewees mentioned that the increasing immigration anxiety in Spain was not specifically related to Europe; however a majority of them stated immigration as an emerging problem. A strategic No campaign could therefore have built on this existing concern and presented it as a Europe-related phenomenon. Turkey’s potential membership to the Union in particular could have been easily linked to concerns around immigration, as it was in the three other cases. Even if Spaniards are more supportive of Turkish accession than the French or Dutch populations,

313 Personal interview, 14 October 2008. 314 Personal interview, 7 October 2008. 315 Personal interview, 1 October 2008. 113

as I explained above, they are sensitive about Spain’s decreasing benefits due to the enlargement process. Thus, Turkey’s membership could have been presented as having a potential negative impact on Spain’s benefits from the Union. In sum, the Spanish Yes campaign frames were stronger than the No campaign frames. Yes frames built successfully on the positive link between Europe and Spain’s democratic, economic and social development. On the other hand, contentious issues such as unemployment, terrorism or immigration, as well as themes which are known to have strong mobilizing appeal such as a potential EU ban on ‘ñ’ in the Spanish alphabet or bull-fighting were not employed in the No campaign. Apart from the regional campaigns using the identity frame, the No campaign did not use such volatile issues in their campaign.

Credibility of the speakers The framing literature argues that credible sources enhance the strength of campaign framing. I argue that when applied to referendum campaigns, the role credibility plays can be assessed on three dimensions: Government popularity, scapegoating of the EU, and disagreements within political parties. To start with government popularity, as I have discussed above, the regular CIS surveys show that the public was generally satisfied with the Spanish government at the time and trusted Prime Minister Zapatero. This can be interpreted as a contributing factor to the strength of Yes campaign framing as Zapatero himself and the government constituted credible sources. Second, in some European countries such as France and the Netherlands, politicians regularly blame the EU for unpopular measures to achieve electoral gains. This tendency decreases the speakers’ credibility in a Yes campaign as the positive frames conflict with the previous critical ones. In Spanish politics, using the EU as a scapegoat was not as common as it was in the other cases, essentially because of the strong positive association of the EU with democracy, peace and prosperity in post-Franco Spain. However, the Aznar government’s framing started to emphasize protecting Spanish national interests in the European integration project. The PP’s pro-TCE standing could have therefore been

114

somewhat in contradiction with the previous stance of Aznar, especially given the PSOE’s emphasis on ‘return to Europe’. Third, disagreements within the campaigning parties decrease their credibility by giving out conflicting signals. As we shall see, in France, the Netherlands, and to a smaller extent in Luxembourg, the social democratic and green parties suffered from internal divisions which resulted in both pro-TCE and anti-TCE messages. Yet, in the Spanish campaign, there were no lasting internal divisions. Even when parties such as the PNV, the CiU or the ICV had internal debates, these parties decided on their final position and coherently campaigned afterwards.

Mobilizational structures Finally, mobilizational structures are another critical factor in understanding the success of frames. I divide this concept into financial resources and civil society contributions. First, the better financial sources the campaigners have, the better they can support their campaign and advance their frames. Political parties and civil society organizations use both public and private sources for their campaign spending. Monitoring reports show that the Yes campaign political parties received far more resources from the state for this referendum campaign. While the parties supporting ratification received 8.1 million Euros, the parties opposing ratification received 0.9 million.316 Strikingly, 9 million Euros – the same amount that was distributed to the parties with parliamentary representation in total – was allocated to the government information campaign.317 As I discussed above, the information campaign was criticized for not being impartial, which in effect suggests that the Yes campaign had vast resources compared to the No camp. This imbalance was a crucial reason for the No campaign’s weakness. Civil society organizations did not receive any financial resources from the state. Among them, those advocating a Yes were able to raise more private

316 Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," 8. 317 Ibid.: 9. 115

funds than the No campaigners.318 Hence the No campaign civil society organizations suffered the most from the situation, and their activities were mainly limited to distributing leaflets on the streets and organizing several debates.319 Given these financial restraints, it is not surprising that in my interviews, most of the Yes campaigners characterized the No campaign as minor and marginal. The No campaigners, on the other hand, pointed out the difficulties they faced. José Manuel Fernández Fernández of the United Left explained that the IU was the only major party that campaigned against the treaty nation-wide, as there was no far right anti-TCE party in Spain.320 The No campaigns of the regional parties were based only in their own regions. Carlos Girbau Costa, from the Social Forum and the IU, pointed to the small size of the IU’s local network, as opposed to the strong networks of the far left parties (PCF or LCR) in France.321 Overall the No campaign not only suffered from lack of financial resources relative to the Yes campaign, but also at the national level, it was only supported by the left-wing political forces which were not powerful in terms of local implantation. In contrast with the French and Dutch far left parties, the IU did not have strong local connections with left-wing civil society organizations. Second, civil society’s contribution was limited in the campaign. To begin, when compared with other European countries, Spain's civil society is relatively weak, thin, and underorganized.322 The main civil society organizations such as the employer association (the CEOE – Confederation of Employers and Industries of Spain), and the trade unions (the socialist UGT – General Workers Union, and the communist CCOO – Workers Commissions) were in favor of the treaty but did not campaign strongly. Only some smaller NGOs campaigned against the TCE. Jaime Pastor, a member of the IU, the ATTAC, and the Alternative Space, stressed that it was a major problem for the No campaign that

318 Ibid.: 8-9. 319 Ibid.: 9. 320 Personal interview, 7 October 2008. 321 Personal interview, 8 October 2008. 322 Closa and Heywood, Spain and the European Union, 53.; Kerstin Hamann, "European Integration and Civil Society in Spain," in Spain and Portugal in the European Union: The First Fifteen Years, ed. Sebastian Royo and Paul Manuel (London: Frank Cass, 2003), 49. 116

there were no internal debates within the trade unions, in contrast with the situation in France.323 On the No side, civil society formed the left-wing ‘No to the Constitution’ campaign platform composed of groups such as the ATTAC, Alternative Space, Red Current and Ecologists in Action. However, this civil society-based platform did not have a strong campaign. My interviewees suggested several reasons for this weakness; The IU and this platform, despite advancing identical arguments, chose to run separate campaigns. Most of my interviewees, including those from the ATTAC, emphasized how the ATTAC Spain was not as active as the ATTAC France, and had no street mobilization potential. The ATTAC Madrid representative Ricardo Gómez Muñoz pointed to the lack of local networks in Spain, in contrast with the enhanced local organization of the ATTAC France that has more than 200 local branches.324 ‘Having 1,000 members and a very small budget, the ATTAC Spain is a weak one’, he added. Carlos Girbau Costa of both the Social Forum and the IU, referring to the ‘No to the Constitution’ platform in general, stressed the small and non-organized nature of the groups that participated. Summarizing the problem, he argued that ‘the left was too narrow and too ideological to show the problems in Europe’.325 Importantly, Luis González Reyes, the coordinator of Ecologists in Action, explained that there were strong social movements against the previous Aznar government but that the Zapatero government made reforms and inactivated these movements, by withdrawing the Spanish troops from Iraq, or legalizing same-sex marriage.326 All these various reasons confirmed the weak nature of the No campaign, both in terms of financial resources and civil society contributions. In sum, the low- profile No campaign focused only on major cities, did not have a strong local network, and accordingly failed to powerfully advance its campaign frames.

323 Personal interview, 8 October 2008. For a detailed comparison of Spanish and French civil society actors and unions, see José Luis González Vallvé, "España y Francia Frente al Referéndum Constitucional: Entre la Inquietud y la Esperanza," Elcano Royal Institute Working Paper 55 (2005). 324 Personal interview, 6 October 2008. 325 Personal interview, 7 October 2008. 326 Personal interview, 9 October 2008. 117

Given the weakness of the No side, it was not surprising that the public was uninterested in the campaign. Most of my interviewees acknowledged that the campaign did not spark enthusiasm and that the campaign meetings gathered only small crowds. As such, the campaign did not highly inform or mobilize the voters. Similarly, the CIS post-referendum survey asked the public how much they were interested in the campaign.327 Only 3% stated high level of interest, 25% stated sufficient interest, 44% stated low interest, and 27% stated no interest at all. But this does not suggest that the campaign did not matter at all, because more specific questions concerning whether Spaniards had seen TV programs, posters, and leaflets on the TCE, or spoken to their families/friends on the subject received mostly affirmative answers. Further, 24% suggested that the campaign helped them a lot or sufficiently to decide on how to vote, whereas 73% mentioned that it helped a little or not at all. Concerning the level of information, while the campaign did not increase the level of knowledge much, it contributed to reducing the number of those who knew nothing at all about the TCE from 29% in July 2004 to 19% after the referendum.328 Finally, at least 50% of the voters took their decision during the campaign and half of them, during the last week.329 The assessment of the campaign in this section shows not only that the Yes campaign frames were stronger than the No frames according to the strength definition adopted from the framing literature, but also that they had better financial resources. The No campaign (with the exception of the regions), suffered from weaker frames and weaker financial resources and civil society mobilization. Taken as a whole, the Yes campaigners did not have a difficult task in maintaining an already positive public opinion as they did not face a strong and strategic No campaign.

327 The results of the survey is available at: http://www.cis.es/cis/opencm/ES/1_encuestas/estudios/ver.jsp?estudio=4617 328 Font and Rodriguez, "The Spanish Referendum on the EU Constitution: Issues, Party Cues and Second-Order Effects." 329 Ibid. 118

Analysis of the Vote In this section I use media content analyses and public opinion data to illustrate the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. The assessment below confirms that the No campaign framing was weak except for the regional No campaigns, and that the Yes campaign frames were indeed more influential.

Shifts over Time As demonstrated in Figure 3.1 at the very beginning, the No vote intentions remained quite steady throughout the campaign, showing a slight increase only at the end. This was in clear contrast to the other three referendum campaigns. Below, media content analyses and public opinion data demonstrate that the No campaign was very weak.

Media Content Analyses The media provided extensive coverage only in the last two weeks leading up to the referendum. Only the information campaign and the free TV slots given to the political parties achieved significant visibility in the campaign.330 Pro-TCE parties represented more than 90% of the Congress and they were given TV time proportional to their parliamentary weight.331 The No campaign political parties were therefore represented only occasionally, while the civil society No campaign was almost invisible in the media coverage.332 Not surprisingly, given the pro- TCE bias of the information campaign, media coverage was overwhelmingly positive. Rosa Berganza conducted a detailed media content analysis of the last two weeks of the referendum campaign (4 February-19 February), which was the main campaign period for Spain.333 This analysis covered the three top newspapers, El País, El Mundo, and ABC, and confirmed that the content of the media coverage

330 Ibid. 331 Madroñal, "Spanish Referendum on the EU Constitution: Monitoring Report," 5. 332 Ibid.: 11. 333 Berganza, "Media Framing of the Spanish Referendum on EU Constitution." 119 was more positive than negative. 26% of the stories described the TCE as beneficial for the EU or Spain, while only 10% found it detrimental. Once again, 26% quoted an actor mentioning positive aspects, whereas only 9% included negative quotes. Table 3.2 shows the valence towards the TCE in the media in detail: Valence (%) Strongly favorable for the Yes side 33.2 More favorable for the Yes side than for the No side 8.8 Balanced 2 More favorable for the No side than for the Yes side 2.8 Strongly favorable for the No side 9 No valence 27.1 TCE not mentioned 17.1 Table 3.2: Valence towards the TCE in the Spanish media coverage

Table 3.3 demonstrates the topical distribution of the media coverage: Topical Newspaper Coverage (%) Orientation of voters, voters’ decision making process, vote 30.7 intentions of citizens Campaigning (style, finance, events) 16.8 Describing, explaining, introducing specific content of the TCE 13.1 Vote advice for referendum 12.8 EU evolution (enlargement, more/less integration) 5.3 Reforming EU institutions and decision making 2.5 Referendum as an instrument of direct democracy 2.5 Other referendum topics 2 Nationalist demands inside Spain related to the TCE 1.8 European culture, history, identity 1.5 Commerce, trade 1.3 Consequences of referendum outcome 1.3 EU USA relations 0.8 Civil rights of EU citizens 0.8 Economic consequences of a common Constitution 0.8 Level of information about the TCE 0.8 Euro/Monetary Union 0.5 Structural funds from the EU 0.5 Competition 0.5 Table 3.3: Topical distribution of the Spanish media coverage

120

This topical newspaper coverage illustrates that the media focused mainly on voter decision-making and intentions, as well as on campaign styles and events. Explaining the content of the TCE also received attention. Nonetheless, the small and dispersed coverage of the rest of the issues shows the low intensity of the campaign. This suggests that the campaigners were not very successful in priming campaign themes. Comparing the Yes and No campaign frames, mainly Yes campaign frames were visible such as the EU evolution, European culture, history, identity, civil rights, commerce, and structural funds. ‘Nationalist demands inside Spain related to the TCE’, or ‘economic consequences of a common Constitution’ could be read as reflecting the No campaign argumentation. Voting advice for the referendum also seemed to have received considerable attention, which is another advantage for the Yes campaign given the higher proportion of the Yes campaigners. Sampedro, Ruiz and Carriço found similar trends in their broader media content analysis, conducted first between May 2003 and October 2004, and then between October 2004 and February 2005.334 They also focused on the three key newspapers, El País, El Mundo, and ABC. They compare the topical media coverage in these two periods, finding that some issues disappeared over time. In 2003 and early 2004, the PP initially criticized the PSOE government’s negotiations over the draft treaty by emphasizing two issues: the Christian roots of Europe and the loss of Spanish power in the European institutions. The research shows that while these issues received media coverage in the first period, they disappeared in the second period, arguing that the PP did not pursue themes that could have mobilized a No vote. This proves that if the No campaign used and primed these issues, it could have successfully drawn the public’s attention to them. Spain was not without any concerns; instead it was the lack of a strong No campaign that helped sustain positive attitudes. The data shows that some issues existed in both periods. A key topic from both periods was regional language recognition (around 4% of the analyzed stories). This reflected the No campaign’s

334 Blanco, Jiménez, and Reis, "El Referendo del Tratado de la UE en la Prensa Española de Referencia: A Favor de la Constitución o en Contra de la Unión Europea." 121

moderate success in priming the regional language argument. As a more general observation, the analysis stresses that the media did not cover concrete themes explaining the reasons for voting for or against the TCE. Some such themes, covered with low intensity, involved abstract values such as peace, security, civil and ethical values of Europe. The Yes campaign message in the media was primarily that Spain received a lot from the EU and that a No vote could lead to the failure of the European integration project. In this sense, the study underlines the synonymous use of the EU and the TCE in the media coverage. Rather than using specific arguments on the TCE, the Yes messages equated the referendum proposal with broader European integration. Moreover, the study confirms the combination of a peace and security message with the ‘return to Europe’ theme, referring to the terrorist attacks. Other topics such as environmental or gender- related advances brought by the TCE received only marginal coverage. Overall, the media content analyses confirm that the debate on the TCE and its policy consequences was not substantive, and that the media coverage mostly focused on the formal aspects of the campaign and the campaigners, with a bias towards positive stories. While priming was not a key aspect as the campaign was of low intensity, the Yes campaigners better primed their themes. The No campaign themes were not very visible, with the exception of regional identity recognition.

Public Opinion Data To what extent did the Spanish public echo the campaign messages? The CIS conducted a detailed survey immediately after the referendum to analyze the reasons behind the Yes and No votes.335 Tables 3.4 and 3.5 show the motivations to vote in favor or against.

335 This survey does not provide specific data for each region. Technical information and the results of the survey is available at: http://www.cis.es/cis/opencm/ES/1_encuestas/estudios/ver.jsp?estudio=4617 122

Reasons to Vote Yes (%) It is essential for continuing the European 38.7 construction The party I vote for supported the TCE 24.2 It creates a European citizenship 14.6 Due to recommendations of friends and 7 family Other reasons 5.1 It strengthens the EU against the USA 4.5 It would be bad if the government had lost 4.1 the referendum Table 3.4: Reasons to vote Yes in Spain

The responses of the Yes voters echoed mostly the first frame of the Yes campaign, which stressed a return to Europe and presented the TCE as a fundamental step for European unification. Given the contention around Aznar’s pro-US policy, a hint of this frame can also be seen in a less frequent response, mentioning the strengthening the EU against the US. The responses on European citizenship can be linked with the third Yes campaign frame outlining the technical content of the TCE, but particularly with the strategic way the PSOE employed it. As I explained, concerned that the technical content would not interest the citizens, the PSOE chose to stress the common rights introduced by the Charter of Fundamental Rights as part of EU citizenship rights. An important percentage of the Yes voters stated that they followed their party’s recommendations, which can be read as a sign of lack of information and need for political guidance. As the campaign was quiet and somewhat lackluster, this response is not very surprising.

123

Reasons to Vote No (%) Due to lack of information 24.5 It has little social content 23 Other reasons 16 It is not a sufficiently democratic text 8 I am against the European construction 6.5 The party I vote for was against the TCE 7 Due to loss of national sovereignty 5 In order to punish the government 4.5 Due to recommendations of friends and 0.5 family Table 3.5: Reasons to vote No in Spain

Many of the No voters, 24%, mentioned lack of information as their motivation, while another 7% stated party recommendation. Similar to the reasoning above, this demonstrates that the Spanish public did not have a sufficient debate on the TCE. The first No campaign frame – that the TCE had insufficient social content – was mentioned by 23% of the No voters. This is a considerable percentage, and shows that this was a clear mobilizing ground for some Spaniards. Nonetheless, it should be kept in mind that this is No voters (17% of the total) only. The third No campaign frame, blaming the TCE for not being democratic enough, is also echoed by 8%. Furthermore, public opinion data shows a certain element of protest vote, around 4%. Interestingly, a section of the PP – the main opposition party – electorate abstained to avoid an overly spectacular victory for the Zapatero government.336 In Madrid, in districts such as Salamanca, Chamartín or Chamberí where the PP holds strong electoral basis, the No vote reached 30%. It should be noted that this survey did not provide respondents with a wide variety of choices. The high percentage in the ‘other reasons’ category, especially for the No voters, demonstrates this problem. The survey later includes more specific questions posed to the general population which provide better tests for campaign framing. When asked how good they believed the TCE would be for Spain, 55% answered very good or good, 17% answered neither good nor bad,

336 Deloy, "The Spanish Approve the European Constitution by a Wide Majority." 124

whereas only 8% answered bad or very bad. Similarly, on whether they thought that EU membership has benefited or hurt Spain, 65% responded that it has benefited, and only 7% responded negatively while 14% remained neutral. These answers show that the second frame of the Yes campaign emphasizing the benefits Spain received from Europe resonated clearly. Following the same lines, Table 3.6 shows the strength of Spain's association of EU membership with modernization and democracy. The question is: Do you think that EU membership had positive or negative effects in each of these dimensions? Very Positive Negative Very Positive (%) (%) Negative (%) (%) Modernization of Spanish society 12.5 64.1 5.9 0.5 Functioning of democracy 10 61.9 5.5 0.4 Role of Spain in the world 10.7 60.1 6.9 0.7 Table 3.6: Spain's association of EU membership with modernization and democracy

More specifically Table 3.7 shows how much Spaniards agree with specific campaign frames.

Very Sufficiently Little Not at all Much (%) (%) (%) (%) With the TCE, Spain loses weight 2.5 12.7 27.9 23.5 in Europe The TCE is a step forward in the 13.8 49.5 11.1 2.9 European integration process The TCE guarantees peace and 11 41.7 14.8 5.8 prosperity in Europe The TCE establishes Europe of 6.6 19.2 19.5 10.4 capital instead of a social Europe The TCE does not recognize the 4.5 13.8 22 16.5 identity of European peoples Table 3.7: Spaniards’ evaluation of the TCE concerning various campaign topics

The first three statements essentially bring together the first and second Yes campaign frames which suggest that the TCE is a step forward and that Spain has gained considerably from being a member of the EU. The agreement levels clearly illustrate that these two campaign frames were strong. The last two

125

statements were the first and second frames of the No campaign. Here, one can see that the social welfare argument had little persuasive effect for the general public. Finally, the last statement concerning the regional identities does not carry broad appeal either, but it becomes crucial when looking at the region-specific polls below. Table 3.8 presents the crosstabulation of Yes/No vote preference and agreement with the same statements:337 Vote Yes Vote No With the TCE, Spain loses weight 10 39 in Europe The TCE is a step forward in the 84 37 European integration process The TCE guarantees peace and 75 24 prosperity in Europe Spain has benefited from the EU 84 46 membership The TCE establishes a Europe of 17 69 capital instead of a social Europe The TCE does not recognize the 10 62 identity of European peoples Table 3.8: Spaniards’ evaluation of the TCE concerning various campaign topics – the Yes/No breakdown

In line with the expectations of the framing literature, Table 3.8 essentially demonstrates that the Yes frames were echoed by the Yes voters, whereas the No frames resonated more with the No voters. As such, it confirms the importance of a frame analysis. So far, the survey questions I assessed were posed to a non-region specific community. To understand the impact of campaign frames in the regions, the regular barometers of the Elcano Royal Institute provide a better test.338 The No vote was higher in the Basque Country (33.6%), and in Catalonia (28%).339 These

337 Joan Font, "El Voto de los Españoles en el Referendum sobre el Tratado Constitucional," Elcano Royal Institute Working Paper 22 (2005): 6. 338 The barometers are available at: http://www.realinstitutoelcano.org/wps/portal/rielcano_eng/BarometeroftheRIElcano 339 In Navarre, the no vote reached 29,2%. In Galicia, despite the efforts of the BNG, the Yes vote reached 80%. Nonetheless, neither the poll data nor my interviews include these regions. Thus the analysis here is limited to Catalonia and the Basque Country. 126

percentages show that in these regions, where the No campaigns were especially active, the No vote rates were indeed higher. When asked whether they believe that the TCE helps the economic, political and cultural development of their autonomous community (region), 54% of Spaniards answered positively. However in the field of culture and identity, the answers in Catalonia (46%) and the Basque Country (42%) were slightly below the Spanish average (50%). Specifically concerning the recognition of regional identities, 20% of Spaniards thought that the TCE does not provide the necessary recognition. Yet, the percentages were higher in Catalonia and the Basque Country reaching 30%. Further, 59% of Spaniards answered positively when asked whether or not they would like the EU to recognize the regional languages of Spain. Strikingly, this number reached 81% in the Basque Country, and 84% in Catalonia. This was highest among the ERC voters (100%), followed by the PNV (95%), the CiU (92%), the IU (80%), the PSOE (65%), and finally the PP (40%). These responses illustrate that in the Basque Country and Catalonia, particularly the second No campaign frame on non-recognition of regional identities was successful. This finding parallels the regional campaigners’ remarks in the interview data that the regional identity issue became an important mobilizer in the regions. Moreover, the November 2004 barometer compared percentages of voters in each party that shared their party’s position on the TCE; the PSOE ranked the highest, followed by the ERC, the PNV, the IU, the PP, and finally the CiU. This ranking demonstrates that among the Yes campaigners, the PSOE, and among the No campaigners, the ERC, did the best. The ERC’s success, coupled with the CiU’s relative failure also confirms that the regional identity was the important factor in mobilizing Catalan society. While the CiU campaigned for a Yes pointing to the recent deal they achieved that allowed for limited presence of Catalan in the EU institutions, the ERC took the hard line by campaigning against the TCE and stressing that this recognition was minor and insufficient. As such, the second No campaign frame (non-recognition of regional identities) was indeed stronger than the fourth Yes campaign frame (limited recognition of Catalan). On

127

the other side, the PSOE, as the main and the most strategic Yes campaigner, used both ‘return to Europe’ and the common EU citizenship rights frames, in addition to the second campaign frame used by all the Yes campaigners. The ranking shows that it indeed succeeded. The public opinion data, on the whole, confirms the predictions of the framing analysis presented above. The Yes campaigners, especially the PSOE, succeeded with their stronger frames that mainly tapped into the positive image of Europe in Spain. The No campaigners, on the other hand, gained more with the regional identity frame than they did with their other weaker frames.

Diffusion As the first country to hold a referendum on the TCE, Spain was not influenced by any previous referendum cases. I argue that the later a country held its referendum relative to other states, the more the previous referenda results and campaigns in the other states influenced its campaign dynamics and public. As such, Spain does not come into the scope of analysis. Nonetheless, the Spanish case is important once again as a control case. The central question of this section is therefore: Why did the Spanish case not influence the subsequent cases? The fact that Spain held the first referendum was not coincidence. José Ignacio Torreblanca pointed out that the sequencing among the four 2005 TCE referenda was coordinated.340 Spain’s strong pro-EU attitude was envisioned to build a positive momentum. Yet while Spain approved the treaty, as we shall see in subsequent chapters, it failed to strongly influence the other cases. In the following referendum cases, the campaigners said that they were primarily inspired by the French case – not the Spanish – and the citizens also had low awareness of the Spanish referendum outcome. I suggest two explanations. First, I argue that cross-case influences are conditional, and depend on the existence of transmission belts between the states such as shared language/culture, common media sources, and collaborative networks/transnational linkages. The more that these channels are open, the more the later cases can be influenced by previous

340 Personal interview, 6 October 2008. 128

ones. I will show that Spain did not share significant diffusion channels with later cases, with the exception of limited collaborative networks/transnational linkages. Second, I argue that campaign intensity mattered. As the Spanish campaign was a quiet one, it did not put pressure on the existing limited diffusion channels. To start with the first channel, shared language and culture, Spain did not share a common language with France, the Netherlands, or Luxembourg. My interviewees mentioned this as a problem blocking further cooperation. José Manuel Fernández Fernández of the IU explained that from the far left parties in the other three countries, they only invited a few campaigners who spoke Spanish to contribute to the Spanish campaign.341 Among them were important figures as Marie-George Buffet, National Secretary of the French Communist Party.342 Similarly Carlos Girbau Costa, from the Social Forum and the IU, pointed to the different cultures of north and south Europe, and mentioned that the Netherlands and Luxembourg were far from Spain, while France shared both cultures.343 French campaigners were confirmed as closer partners during most of my interviews. Yet, this closeness did not reach the same level of linguistic and cultural connection between France and Luxembourg. Second, concerning common media channels, Spain did not have any common television channels or other media outlets such as newspapers with the later cases. This also formed an impediment to its potential influence on the others. Through the third and final channel, collaborative networks and transnational linkages, Spain did have interaction with the other cases. I explore institutional networks and personal connections by looking at the European Parliament groups, the European anti-globalization network, ad hoc European networks and existence of mobile communities. Spain did not have any mobile communities as Luxembourg did, but it was influenced by the EP groups as well as the European anti-globalization network. Spanish political parties are well-

341 Personal interview, 7 October 2008. 342 See Deloy, "Referendum on the European Constitution in Spain on 20th February: An Assessment Just a Few Days Before the Election." 343 Personal interview, 8 October 2008. 129

integrated into the EP groups. Most campaigners from political parties mentioned discussion of the campaign with their EP party groups, which included similar French, Dutch and Luxembourgish political parties.344 The campaigners stated that they agreed with their counterparts on the arguments and the main campaign lines in occasional meetings but also warned that the campaigns were to run separately. Interestingly, following the pattern in the three other cases, the No campaigners mentioned stronger collaboration than the Yes campaigners. Also, just like in France, the Spanish greens ICV mentioned the problems they faced due to their anti-TCE position within a pro-TCE EP group, the European Greens.345 Interviewees from the IU and the civil society No campaign mentioned strong links to the European anti-globalization network. Once again paralleling the pattern in the other cases, civil society campaigners frequently referred to the ATTAC and the European Social Forum meetings as important platforms. They specified the November 2003 ESF meeting in Paris, and the 2004 ESF in London as sources, where the TCE was debated intensely.346 Importantly, they explained that the ATTAC France was especially a reference point as they had developed theoretical questions and arguments regarding the TCE as early as 2003. Specifically, the book by Yves Salesse, one of the key left-wing campaigners in France, was mentioned as a resource. This book was even translated into Spanish and published. These remarks demonstrate that the French No campaign debate started early despite the fact that their campaign was held later than the Spanish one. Through the ESF meeting in Paris, and the ATTAC France’s theoretical criticism of the TCE, the French left-wing No campaign debate influenced the Spanish left-wing No campaigners. An important aspect within the third channel is the local adaptation of the frames. The anti-globalization network in Europe spread the arguments that

344 Personal interview with Enrique Baron Crespo, 16 October 2008; personal interview with Alejando Muñoz Alonso, 2 October 2008; personal interview with Jordi Xuclà i Costa, 8 October 2008; personal interview with Marc Giménez Villahoz, 15 October 2008; personal interview with José Manuel Fernández Fernández, 7 October 2008. 345 Personal interview with Marc Giménez Villahoz, 15 October 2008. 346 Personal interview with Carlos Girbau Costa, 8 October 2008; personal interview with Marc Giménez Villahoz, 15 October 2008; personal interview with Jaime Pastor, 8 October 2008. 130

originated in the Paris ESF and the ATTAC meetings and thereby formed a left- wing master frame, advocating an alternative ‘social’ EU. The left-wing activists in Spain and the other three cases echoed these arguments. Nonetheless, apart from the Dutch far left political party, the left-wing No campaigners in the three other countries did not strategically fit this master frame to their national frameworks. Borrowing frames worked both positively and negatively depending on the degree to which these borrowed arguments fit the local reality. The Spanish campaigners did not adapt this left-wing master frame that the TCE was too market-friendly, to the concerns of Spanish society. In other words, they did not use the contentious issues of Spanish society to make these borrowed frames stronger. The only difference was that the IU stressed its wish to see a federal Union, which is the opposite of what the far left in France or the Netherlands argued. However, this peculiar emphasis was not visible in the Spanish media, neither it was echoed by the Spanish public in the survey data. In a nutshell, Spain held the first referendum and in so doing avoided being influenced by an earlier negative case. Further, the limited diffusion channels only allowed in a small number of anti-TCE arguments. Thus in the Spanish case, diffusion worked only minimally, amplifying the significance of the independent variable – the relative No campaign strength. Facing a weak No campaign and limited diffusion of anti-TCE frames, the Yes campaigners easily sustained the initial pro-EU public opinion on the TCE. More importantly, the quiet Spanish campaign did not affect the upcoming three referenda cases. Originally EU officials hoped that Spain’s strong pro-EU attitude would build positive momentum. To further this goal, the president of the European Commission, Barroso, and several European statesmen such as Chirac and Schroeder, participated in the Spanish Yes campaign. Yet due to weak diffusion channels and low campaign intensity, it did not happen. Instead, France turned out to be the momentum case with an intense campaign and strong diffusion channels.

131

The Quiet Campaign José Ignacio Torreblanca summarized the Spanish campaign in a few words: ‘Instead of having a policy debate on what sort of Europe we want, we had sportsmen reading Article 7 of the Charter of Fundamental Rights on television’.347 The extracts used in the campaign stressed the new rights for EU citizens, and he explained that this resonated very well with the idea that Europe had brought wealth, prosperity, and social modernization. Despite the lack of scholarly attention, the Spanish referendum is a key case to understand the dynamics of 2005 referenda campaigns. It serves as a control case, demonstrating how the early positive public opinion could be maintained in a quiet and uneventful referendum campaign. I argued that two factors mattered: The relative weakness of the No campaign, and the limited nature of diffusion. First and foremost, Spanish society was not exposed to strong anti-TCE frames put forth by a strategic No campaign. This argument is reinforced by the fact that the stronger regional No campaigns indeed raised the No vote, especially in Catalonia and the Basque Country. The national No campaign did not use contentious issues in society such as unemployment, immigration or crime to frame their arguments, nor did it stress topics which carried significant mobilizing potential such as a possible EU ban on ‘ñ’ in the Spanish alphabet or on bull-fighting, or Turkish accession into the Union (possibly linking it to recent Islamist terrorism or the negative impact of enlargement on Spain’s benefits). As such, the absence of a strong No campaign mattered, domestic contention remained unlinked to the TCE, and the Yes campaigners easily preserved the initial positive public opinion. Second, the sequencing of the 2005 TCE referenda that allowed Spain to go first also helped Spain sustain its positive attitude. Not only did it avoid following a negative referendum, but restricted diffusion channels only brought in a minor amount of anti-TCE framing. Moreover, the Spanish case demonstrates the significance of campaign intensity to facilitate diffusion. Having had a quiet campaign, Spain failed to build the strong positive momentum it was expected to

347 Personal interview, 6 October 2008. 132 generate. Given the minimal diffusion, the Spanish vote is primarily explained by the relative weakness of the No campaign.

133

PERSONAL INTERVIEWS

Political parties:

Yes campaigners: Juan Fernando López Aguilar (PSOE, MP, then Minister of Justice and Interior Affairs) Juan Moscoso del Prado Hernández (PSOE, MP) Enrique Baron Crespo (PSOE, MEP, the chairman of the Party of European Socialists Group in the EP between 1999 and 2004) Orestes Suárez Antón (PSOE, International Secretary) Alejando Muñoz Alonso (PP, Senator) Ignacio Cosidó Gutiérrez (PP, MP) Jordi Xuclà i Costa (CiU, MP) José Ramón Beloki Guerra (PNV, MP) Carles Llorens I Vila (CDC, International Secretary)

No campaigners: José Manuel Fernández Fernández (IU, Coordinator of the Parliamentary Group, and mayor of Bustarviejo) Joaquim Puig Vilamala and Oriol Duran Torres (ERC, Coordinator, and Spokesperson of the Parliamentary Group) Marc Giménez Villahoz (ICV, European Politics Coordinator) Mikel Irujo Amezaga (EA, MEP)

Civil society:

Jaime Pastor (IU, ATTAC, Alternative Space) Ricardo Gómez Muñoz (ATTAC) Carlos Girbau Costa (Social Forum, IU) Luis González Reyes (Ecologists in Action) José Ignacio Torreblanca (Senior Analyst for EU Affairs, Elcano Royal Institute for International Affairs, participated in the campaign by taking questions on the national public radio, and by touring the country for meetings) Jordi Vaquer i Fanés (Europe Programme Co-ordinator, CIDOB Foundation)

134

CHAPTER 4 FRANCE: THE MOMENTUM CASE

On 29 May 2005, only three months after the Spanish approval, the French public rejected the TCE by 54.68%. The Yes percentage was 45.32%, and the turnout rate was remarkably high at 69.34%. This negative result shocked Europe. The positive vote in Spain had raised hopes that the other countries would follow. President Jacques Chirac had suggested that ‘this vote in favor of the Constitution and Europe by the Spanish people is a major sign and shows the way to the other countries who will be ratifying this treaty over the next few months’.348 Moreover, France was not an ordinary EU member. It has historically been a key actor in driving the European integration project further. The very Convention that drafted the TCE was led by a former French President, Valéry Giscard d'Estaing. Contributing to the puzzle, just like the three other cases, was the fact that French public opinion was initially positive towards the TCE. In September 2004, the Advice on Survey Analysis (CSA) found that 69% of the population was in favor of the TCE, while only 31% was against.349 Also paralleling the three other referenda, the French political mainstream formed the Yes camp: The Union for a Popular Movement (UMP), the Union for French Democracy (UDF), the Socialist Party (PS), and The Greens (Les Verts).350 Only the far right and the far left political parties were officially against the TCE: The National Front (FN), the Movement for France (MPF), the French Communist Party (PCF), and the Revolutionary Communist League (LCR). Given all these similarities, this chapter asks: Why was the initial positive attitude not sustained in France; why did the French reject the TCE? Furthermore, why did they not follow the positive Spanish vote?

348 Quoted in Deloy, "The Spanish Approve the European Constitution by a Wide Majority," 59. 349 Available at: http://www.csa.eu/dataset/data2004/opi20040915c.htm 350 There were visible inner splits within three Yes camp political parties. While the Socialist Party, The Greens, and the UMP officially campaigned for a Yes, important figures from these parties joined the no campaign. This not only increased the strength of the no campaign but also decreased the Yes campaign’s credibility. I will discuss this point in great detail below, under the campaign analysis section. 135

I argue that the answer to the puzzle lies in campaign dynamics. France had the most intense debate among the four cases. The French No campaigners started their analysis of the Constitution considerably earlier than the Yes campaigners and their strong No campaign succeeded in linking existing contentious issues to the TCE. The studies on the subject commonly pointed to domestic dissatisfaction, suggesting that the French public may have punished the government. Yet the literature has ignored a key fact: The domestic and the EU- related contention existed while the public still supported the TCE in fall 2004. In the absence of the No campaign, the two remained unlinked in the public’s mind. Therefore, I argue that the main factor that explains the negative result is the relative strength of the No campaign. The Yes campaign failed to generate effective pro-TCE arguments to counter the anti-TCE frames. Second, regarding diffusion, the French referendum has built a momentum and became the most influential case among the four due to its intense campaign and the strong diffusion channels it shared primarily with Luxembourg. Spain’s quiet campaign and the limited diffusion channels between the two cases explain why the Spanish vote did not influence the French case. In this chapter I will first demonstrate the evolution of public opinion on the subject over the course of the campaign. Second, I will look at French public opinion to discern the salient contentious issues of the period, relating to both domestic and EU levels of governance. Third, using interview data, I will analyze the long campaign period and discuss the Yes and No campaign strategies. Here, I will focus on the strength of the campaign frames, credibility of the speakers, and mobilizational structures available to both sides. Specifically, I will explain how the strong No campaign linked the contentious issues to the referendum proposal and how the Yes campaigners could not build effective arguments to counter this No campaign framing. Fourth I will analyze the referendum vote, based on media content analysis and public opinion data, to show the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. This assessment will verify the campaign frame analysis. Lastly I will discuss why France became the most influential case in terms of diffusion, by

136 exploring the diffusion channels between France and the other cases, and the intensity of the campaign.

Evolution of Public Opinion The data on the evolution of voting intentions verify that public opinion was positive in fall 2004, and that the intention to vote No started to gain ground during the course of the referendum campaign. Figure 4.1 shows the shifts in vote intentions over time: 351

Source: CSA

Figure 4.1: Evolution of vote intentions in France (October 2003-May 2005)

The No vote intention has steadily risen throughout the campaign, which is closely linked with the success of the No campaign as I will show below.

351 Sally Marthaler, "The French Referendum on Ratification of the Constitutional Treaty " European Parties Elections and Referendums Network, Referendum Briefing Paper 12 (2005): 5. 137

Discontent with Domestic and EU Governance The French public was dissatisfied with the domestic level of governance in the few years leading up to the referendum. On the other hand, the public attitude towards the EU has been fairly positive. But the political elite frequently used the Union as a scapegoat for unpopular policies and thereby created some controversy in the public opinion. Below I outline the contentious issues in both domains. Importantly, this section demonstrates that these issues were equally contentious in fall 2004 when the polls showed positive attitude towards the TCE. The existing contention provided campaign material for the No side, yet it remained unlinked to the TCE before the No campaigners successfully capitalized on these issues in the referendum campaign.

Contention concerning domestic politics France held presidential and legislative elections in 2002. The winner of both was the center-right UMP, with President Jacques Chirac and Prime Minister Jean- Pierre Raffarin.352 The main contention during the UMP rule was around social policy and immigration. I will discuss these concerns in detail. The CSA regular polls ask the public whether they trust their President and Prime Minister to address the major problems in France effectively.353 Figure 4.2 shows the evolution of affirmative answers over time in percentages:

352 The UMP was called the Rally for the Republic at the time of the presidential elections, and the Union for the Presidential Majority at the time of the legislative elections. 353 Available at: http://www.csa.eu/dataset/data2005/cote20050516.htm 138

Source: CSA

Figure 4.2: Trust levels in Chriac and Raffarin’s ability to address problems (2003-2005)

As can be seen, the confidence levels of both Chirac and Raffarin reached its lowest point right before the referendum. It is important to note that the fall 2004 levels were similarly very low, when the polls reflected favorable positions on the TCE. What were the main reasons behind this unpopularity? The Raffarin government lost its popularity quickly due to rising unemployment and the proposed program of reforms to health care, pensions, unemployment benefits, and education.354 In 2004 unemployment was on the rise and the budget deficit breached the EU’s stability pact, leading to fierce criticism of the government’s economic performance.355

354 Shields, The Extreme Right in France: From Pétain to Le Pen, 295. 355 The election results further demonstrated this dissatisfaction; the Socialists, the Communists, and the Greens won overwhelmingly. Nonetheless, Chirac decided to retain Raffarin as Prime Minister despite the defeat of the government party in the 2004 regional and European elections and his unpopularity. See Ibid.; Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 10. 139

The regular CSA surveys ask the French public to choose the three domains that concern them the most.356 Between May 2004 and May 2005, the French selected the following themes the most, in hierarchical order: Health (44- 57%), pensions security (36-43%), employment (27-37%), purchasing power (18- 36%), security (21-30%), education (20-29%), social protection (18-30%), environment (19-27%), quality of life (15-23%), housing (5-13%), labor rights (6- 8%). Eurobarometer data from fall 2004 underlines similar concerns: When asked about the two most important problems facing their country, the French selected unemployment first and crime second, followed by the economic situation, rising prices, health care system, immigration, pensions, and terrorism.357 Unemployment exceeded 10% in March 2005, for the first time in five years.358 At the same time, the March CSA surveys found that more than half of the French felt that their purchasing power had decreased over the past three years.359 These survey results indeed show that the public ranked socio-economic and welfare- related concerns the highest, mentioning immigration as well. While immigration was not ranked particularly high, it was a sensitive issue. Below I discuss the contention around these issues. Concerning the welfare state, it is important to note that since the Second World War, France has built the world’s third most expensive health and welfare system.360 France has the most impressive record of social spending and labor law innovations as well as the second highest level of pension increases; but it also has the poorest record of job creation and saw a dramatic increase in unemployment in the last quarter of the twentieth century. In an extensive

356 Available at: http://www.csa.eu/dataset/data2005/opi20050601c.htm 357 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 358 Gaëtane Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," Notre Europe 44 (2005): 14. The OECD statistics show figures around 9% for 2005 but confirm that it reached the highest level in the last five years. Available at: http://stats.oecd.org/index.aspx?queryid=21760 359 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 60. 360 Between 1981 and 2002, socialist governments ruled France for 15 years, and in 2000 the World Health Organization rated the French health care system the best in the world. The French are therefore strictly committed to the idea of a big welfare state, commonly juxtaposing their system against a negative vision of the American system. Timothy B. Smith, France in Crisis: Welfare, Inequality and Globalization since 1980 (Cambridge: Cambridge University Press, 2004), 2. 140

analysis, Smith suggests that politicians were keen to blame outside forces for the problems of the French welfare state instead of homegrown reasons.361 I will discuss this pattern in detail relating to the European level of governance. The contention concerning the decline in the welfare state was not new. Between 1995 and 2003, France saw many demonstrations denouncing ‘neoliberal globalization’, against privatization of state-owned companies and cuts to social spending and pensions.362 Prime Minister Alain Juppé’s vast reform program of the welfare system in 1995-1996 led to major strikes, indeed the largest since May 1968. The strike activities continued until 2004, concerning different social rights such as pensions or retirement age. Particularly in 2003, Raffarin’s questioning of the 35-hour work week and the proposed pension reforms caused large trade union mobilizations, massive strikes and walkouts across the country in defense of the public sector and higher wages.363 Specifically on 10 March 2005, several unions demonstrated to protect working hours, purchasing power and employment; and on 16 May 2005, only 13 days before the referendum, there were strikes against the removal of Whit Monday as a bank holiday.364 The protestors, including ATTAC, blamed Raffarin for ‘destroying’ the welfare state.365 Immigration was another important source of contention in France in the early 2000s. In 2001, the proportion of persons born abroad was higher in France and the Netherlands at more than 10%, compared to Spain.366 Initially, due to

361 Ibid., 5. Smith suggests that the primary reason for this problem is that most social benefits go to the richest half of the society. He characterizes the French model as a loser-suffer-all labor market and a winner-take-all welfare state, as opposed to the winner-take-all North American market dominated by business elites and political allies. 362 Ibid., viii. Also see Camal Gallouj and Karim Gallouj, "The French Social Protection System: Current State and Future Prospects," in The Handbook of European Welfare Systems, ed. Klaus Schubert, Simon Hegelich, and Ursula Bazant (Abingdon: Routledge, 2009). 363 Smith, France in Crisis: Welfare, Inequality and Globalization since 1980, 63. See also Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second-order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia." 364 Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," 7. 365 Quoted in Smith, France in Crisis: Welfare, Inequality and Globalization since 1980, 63. Labor leader Claude Debons, of the French Democratic Confederation of Labor (CFDT), also openly pointed to the Thatcherism behind the Prime Minister’s policies. 366 Following decolonization, the pull of high French economic growth combined with open immigration policies made control difficult and created protected immigrants from former African 141

republican ideals, the French held an open immigration policy based on an expansive notion of citizenship rather than an ethnocultural one.367 The consensus for open immigration lasted until the 1973 major economic recession, after which successive governments have tried to control immigration.368 The issue exploded in 1984 when the far right FN won municipal elections in Dreux on a platform advocating a complete halt to immigration and encouraging the deportation of African immigrants. Immigrants were suddenly the cause of the economic and cultural decline.369 While in general immigrants were accused of taking jobs from French citizens, Muslims were also seen as inassimilable and hostile to republican values.370 European Social survey measures the ‘ethnic distance’, which is the degree of resistance to social interaction with members of ethnic groups in various domains such as the residential neighborhood, work, and personal relationships.371 The ethnic distance in France is close to the EU average, but along with the Dutch, it is higher than the Spanish and Luxembourgish figures.372 The issue gained utmost importance in the 2002 presidential elections. Much to the surprise of the French electorate, Le Pen won 17% of the vote in the first round of the presidential election, defeating a divided left and qualifying as one of the two candidates for the second round alongside Chirac. In the second round, Chirac won by 82%, the largest margin of victory – against the 18% FN vote. But Le Pen succeeded in linking insecurity and unemployment with illegal immigration.373 He argued that national identity was at stake, and as such his rise

colonies who became quasi-French citizens. Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 34. 367 James Hollifield, "France: Republicanism and the Limits of Immigration Control," in Controlling Immigration: A Global Perspective, ed. Wayne Cornelius, et al. (Stanford: Stanford University Press, 2004), 189. 368 Ibid., 190. 369 Ibid., 193. 370 The second half of the 1990s, under the Socialist government, tried to bring the French immigration policy back to its republican origins. Yet, these immigration reforms failed to contain the rising xenophobia. Ibid., 199. 371 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 74. 372 Ibid., 75. 373 Le Pen’s framing was clearly far right, combining diagnosis of decadence (AIDS, decline of family, cosmopolitanism, ethnic warfare), diabolic causality (scapegoating the North African immigrant as the bearer of all evils), restoration (need to reinstate tradition), and lastly trust in the leader who embodies ‘the healthy people’. See Hollifield, "France: Republicanism and the Limits 142

was closely connected to the fear of globalization.374 Moreover, December 2003 saw a highly charged public debate over the banning of Islamic headwear and other conspicuous religious symbols in state schools.375 In 2004, fear of terrorism increased with the threat of attacks on the French rail network, which overlapped with the Islamist terrorist bombings in Madrid.376 As a result, the 2004 regional elections showed that no region lay beyond the appeal of the FN, which recorded over 8% in every one of France’s 21 mainland regions.377 Therefore, within the domestic sphere, the main concern of the period was economic and welfare state-related anxiety as well as considerable sensitivity for increasing immigration. These themes, when combined, reinforce a fear of globalization. Immigration and the welfare system can be easily linked as restrictions in that both rely on the same logic – to protect the country against unwelcome aliens and welfare claimants in an increasingly globalizing world. This correlation can also be seen in the FN’s political program, which in the 1990s, switched to a pro-welfare agenda, increasing its support among the most economically vulnerable sections of French society.378 A similar combination could be observed in Dutch society when looking at the electoral success of Pim Fortuyn, which I will discuss in great detail in Chapter 5.379 In both France and the Netherlands, the 2005 anti-TCE campaigns successfully struck the same, already-sensitive nerve.

of Immigration Control," 210.; Pascal Perrineau, "The Conditions for the Re-emergence of an Extreme Right Wing in France: The National Front, 1984-1998," in The Development of the Radical Right in France: From Boulanger to Le Pen, ed. Edward Arnold (Basingstoke: Macmillan Press, 2000), 253-54.; Peter Davies, The Extreme Right in France, 1789 to the Present: From de Maistre to Le Pen (New York: Routledge, 2002). 374 Smith, France in Crisis: Welfare, Inequality and Globalization since 1980, 65. 375 Shields, The Extreme Right in France: From Pétain to Le Pen, 296. 376 Ibid., 297. 377 Ibid., 301. 378 Ibid., 8. 379 See Lechner, The Netherlands: Globalization and National Identity. 143

Contention concerning the EU French public opinion has been consistently positive towards the European integration but its support has been less enthusiastic than that of the Spanish public. Looking at general public opinion towards the Union, longitudinal surveys suggest that there is no reduction in support for European integration up to and including 2005 in France or the Netherlands.380 In comparative ranking of the member state public opinion, the French population is still relatively positive towards the integration project.381 Figures 4.3 and 4.4 show the trends in the Eurobarometer surveys:382

Figure 4.3: French support for EU membership (1984-2004)

380 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 1. 381 Alain Guyomarch, Howard Machin, and Ella Ritchie, France in the European Union (New York: St,Martin's Press, 1998), 101. 382 Eurobarometer, "Standard Eurobarometer 61: Public Opinion in the EU, Spring 2004." 144

Figure 4.4: French public’s perceived benefit from EU membership (1984-2004)

But when looking at a broader period, Figure 4.5 shows that French public opinion has shown a downward trend since the late 1980s:383

Source: Eurobarometer Figure 4.5: French support for EU membership (1973-2005)

383 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 12. 145

There are several reasons behind this downward trend.384 Since the late 1980s, the EU’s range of decision-making power and policies has increased drastically, provoking debate both among the public and the political parties. Especially the completion of the internal market was deemed to threaten French dirigisme and protectionism by opening up the markets.385 Furthermore, the French public was allowed to voice their discontent because of Mitterrand’s decision to ratify the Maastricht Treaty in a referendum. This referendum debate sparked political controversy on the EU, as I will explain below. The Eurobarometer fall 2004 survey asked citizens of every member state about their fears concerning the building of Europe.386 While the sets of concerns highlighted by all four countries are similar, France has the highest rank among the four – also among the EU 25 – in mentioning the transfer of jobs to other member countries which have lower production costs. France also ranks high when it comes to the loss of social benefits, more difficulties for French farmers, and the budget contribution. It is not a coincidence that these concerns were mentioned, because the French political elite have been raising these particular issues in relation to the European integration.

Political Party Attitudes towards Europe France is a founding member and has been a motor of European integration. French statesmen such as Jean Monnet, Robert Schuman, Jacques Delors, and Valéry Giscard d'Estaing played a massive role in pushing the integration forward. But French leaders have also been responsible for some of the major set- backs in the integration process. De Gaulle strongly resisted tampering with the sovereignty of member states, and accordingly advocated a limited community of interests and identity among European states.387 The French political elite’s attitudes towards European integration have been influenced by both of these

384 Guyomarch, Machin, and Ritchie, France in the European Union, 95. 385 Ibid., 75. 386 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 387 In an attempt to protect French national interests, he almost brought the integration project to a halt in the 1960s with the Empty Chair Crisis. Guyomarch, Machin, and Ritchie, France in the European Union, 21. 146

views, resulting in integrationist as well as cautious moments. Thus the French relationship with the EU is more complex than the other three cases. When looking at the three other referendum countries, the national benefits/contributions balance concerning the EU budget has been important in defining the political elite’s attitude towards the Union. In France, despite being a net contributor to the budget, this has not been a major source of contention. The difference between France’s contribution and benefits has been increasing since 2004. While contributing an important amount of funds to the budget, it has also been a major beneficiary of the Union’s agricultural funding. Table 4.1 summarizes France’s benefits and contributions between 2000 and 2005:388

Total benefit from the EU Total contribution to the EU (Euro million) (Euro million) 2000 12,377.8 12,866.0 2001 11,674.5 12,961.9 2002 12,155.7 13,202.7 2003 13,359.1 14,113.8 2004 12,944.9 14,878.6 2005 13,620.5 15,516.8 Table 4.1: France’s benefit and contribution in the EU budget (2000-2005)

Instead of the budget, the welfare system and social spending have been intensely debated in France in relation to the EU. The French political elite frequently linked the problems of the French welfare system to globalization and encouraged French citizens to seek culprits beyond their borders.389 Especially from the mid- 1990s onwards, French intellectuals, politicians and labor leaders attributed France’s problems to globalization, Thatcherite neoliberalism, and the drive toward the EU.390 The French political elite frequently justified the pro-market

388 Detailed information can be reached at the EU’s official website on financial programming and budget. Available at: http://ec.Europa.eu/budget/budget_glance/where_from_en.htm 389 See Vivien Schmidt, The Futures of European Capitalism (Oxford: Oxford University Press, 2002).; ———, "The Politics of Economic Adjustment in France and Britain: When Does Discourse Matter?," Journal of European Public Policy 8, no. 2 (2001).; ———, "Does Discourse Matter in the Politics of Welfare State Adjustment?," Comparative Political Studies 35, no. 2 (2002). 390 Specifically, under the PS Prime Minister Lionel Jospin (1997-2002), the French state was portrayed as a model in a world of ‘neoliberal’ politicians and organizations such as the International Monetary Fund and the World Trade Organization. At the grassroots level José 147

reforms with the need to remain in the European Monetary System. Prime Minister Juppé tied his vast and considerably unpopular reforms in 1995 to the Maastricht Treaty and the need to cut spending in order to meet the EMU criteria, even though they were necessary regardless of Europe.391 Hence in the eyes of the French public, Europe and welfare state reforms are connected. Not surprisingly, during the TCE campaign an EU directive on liberalization of services became highly contentious. This law also known as the Bolkestein Directive symbolized the European Commission’s ‘neoliberal’ agenda, providing cover for the ‘anti- social’ policies of the French government. Do these reservations regarding European integration translate into Euroscepticism in party politics? The answer is negative as hard Euroscepticism is limited to the far left and far right in the French system, along with the rest of the member states. While French mainstream political and administrative elite have always presented European integration as a useful platform to further French political and economic interests, they have also found it useful to use the EU as a scapegoat for unpopular policies even if they have actively been part of the decision-making process.392 However these positions never turned into outright opposition, and Euroscepticism in the mainstream parties was never high enough to lead to a realignment of the party system on the basis of positions towards the EU.393 While not insignificant, Euroscepticism has only added a complex

Bové, anti-globalization leader, became a national celebrity and received significant media coverage. Smith, France in Crisis: Welfare, Inequality and Globalization since 1980, ix. 391 Ibid., 39. For a detailed analysis also see Schmidt, The Futures of European Capitalism.; —— —, "Does Discourse Matter in the Politics of Welfare State Adjustment?."; Michael Loriaux, France after Hegemony: International Change and Financial Reform (Ithaca: Cornell University Press, 1991).; David Howarth, The French Road to European Monetary Union (Basingstoke: Palgrave, 2001).; Susan Milner, "Protection, Reform, and Political Will: France and the European Social Model," in French Relations with the European Union, ed. Helen Drake (Abingdon: Routledge, 2005). 392 Guyomarch, Machin, and Ritchie, France in the European Union, 31. Howarth, The French Road to European Monetary Union. For a detailed analysis of De Gaulle’s legacy concerning Europe, see Nick Startin, "Maastricht, Amsterdam, and beyond: The Troubled Evolution of the French Right," in French Relations with the European Union, ed. Helen Drake (Abingdon: Routledge, 2005). 393 Chris Flood, "French Euroscepticism and the Politics of Indifference," in French Relations with the European Union, ed. Helen Drake (Abingdon: Routledge, 2005), 42. 148 dimension to both the center right and left.394 The French political system is still mainly based on a strong left-right division and the political elite chose to neutralize the European issue so as to not to allow a new political division to weaken this key cleavage.395 Only the far left and far right have been openly sceptical towards the EU. Since the mid-1980s, the French far left parties such as the PCF, the LCR, the Workers’ Struggle (LO) have strongly questioned the economic consequences of the European monetary project. They attacked the EU for its commitment to liberal capitalism and deregulation which are adverse to the interests of the disadvantaged sections of the society.396 On the other hand, French far right parties such as the FN, and Bruno Mégret’s National Republican Movement (MNR, split from the FN) opposed the increasing immigration and potential membership of an Islamic country such as Turkey in the EU. The FN presented Europe as a historic civilization founded on Christian values and opposed all efforts to force economic or political unity.397 Interestingly the French far right also emphasized the adverse socio-economic consequences of the EMU project.398 The French referendum on the TCE was the third referendum on Europe; the first referendum was held in 1972 on the enlargement to the UK, Ireland, Denmark and Norway, and the second referendum took place in 1992 to ratify the Maastricht Treaty. Both of these previous referenda had positive results.

394 Ibid. Within the French mainstream, only certain politicians became openly Eurosceptic, such as Philippe Séguin, Charles Pasqua (founded his own party Rally for France, RPF), Phillipe De Villiers (founded his own party Movement for France, MPF), Nicolas Dupont-Aignan (founded his own party Arise the Republic, DLR) from the conservatives, and Jean Pierre Chevènement (founded his own party Citizen and Republican Movement, MRC) from the socialists. The literature has classified these parties as soft Eurosceptics rather than hard ones. See Paul Taggart and Aleks Szczerbiak, "The Party Politics of Euroscepticism in EU Member and Candidate States," Opposing Europe Research Network Working Paper 6 (2002). 395 Gerard Grunberg, "Euroscepticism in France, 1992-2002," in Opposing Europe? The Comparative Party Politics of Euroscepticism: Volume I, ed. Aleks Szczerbiak and Paul Taggart (Oxford: Oxford University Press, 2008). 396 Flood, "French Euroscepticism and the Politics of Indifference," 54. 397 Davies, The Extreme Right in France, 1789 to the Present: From de Maistre to Le Pen, 138. Thus the FN carries two irreconcilable postures, a pro-European notion of ethnic and cultural kinship, and an anti-European stance based on the Maastricht model of integration and political federalism. See Shields, The Extreme Right in France: From Pétain to Le Pen, 243. 398 Flood, "French Euroscepticism and the Politics of Indifference," 51. 149

Nevertheless, the 1992 Maastricht referendum was approved by a narrow margin of 51%. This referendum formed the first occasion where a concrete pro/anti- integration cleavage crystallized in French politics.399 Just like the campaign on the TCE, the far left and far right – the PCF, the LCR, the FN, and Philippe De Villiers (then an MP of the UDF) – were against the treaty. The center-left was only slightly divided. This campaign had its principal No campaigners from the deeply divided center-right wing (from the Rally for the Republic – forerunner of the UMP). Accordingly the main No campaign argument, based on a traditional Gaullist vision of Europe, was that the Maastricht Treaty was another step towards an undemocratic, remote, and uncontrollable European ‘super-state’ that threatened national sovereignty.400 In the electorate as well, the leading motivations specified by the No voters were sovereignty and technocracy.401 In clear contrast, the 2005 TCE campaign saw a deeply divided center-left and a slightly divided center-right, coupled with an electorate motivated to say No primarily because of social concerns. The Maastricht referendum debate therefore turned Europe into a political controversy and reinforced the growing lack of confidence among certain sections of society in the ability of the EU to address key problems. Since the 1990s, the mainstream elites have tried to conceal their divergence on European matters whereas the marginal far right and far left parties have capitalized on public anxieties about European integration.402 Specifically, the discussion on the domestic costs of meeting the Maastricht EMU convergence criteria in relation to French social spending has led to increasing concerns. However these were not significant enough to produce a major realignment of the party system on the

399 Gaëtane Ricard-Nihoul and Morgan Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," Elcano Royal Institute Working Paper 57 (2005): 5. 400 Howarth, The French Road to European Monetary Union, 170. For discussions of this referendum debate, see Lawrence LeDuc, The Politics of Direct Democracy: Referendums in Global Perspective (New York: Broadview Press, 2003).; Byron Criddle, "The French Referendum on the Maastricht Treaty September 1992," Parliamentary Affairs 46, no. 2 (1993).; Guyomarch, Machin, and Ritchie, France in the European Union, 97-102.; and Flood, "French Euroscepticism and the Politics of Indifference." 401 Criddle, "The French Referendum on the Maastricht Treaty September 1992." 402 Guyomarch, Machin, and Ritchie, France in the European Union, 103. 150

basis of positions towards the EU.403 In the public opinion there was some Euroscepticism especially in the attitude of a significant minority, but it was often an episodic scepticism of indifference or detachment rather than an outright antagonism.404 Overall, in the period leading up to the 2005 referendum, there were contentious issues concerning both the domestic and European levels of governance. Society had primarily socioeconomic and immigration related anxieties. Yet these issues, despite being just as contentious in fall 2004, were not linked to the referendum proposal in the public’s mind. Similarly, the post- referendum polls found that public attitudes towards the European integration were positive: 72% indicated to be in favor of continued European integration.405 Therefore what happened was an episodic dip rather than permanent opposition. The next section will show how the No campaigners caused this dip by connecting contentious issues strategically to the TCE in the referendum campaign.

The Long Campaign In this section I analyze the strategies and framing of the Yes and No campaigns. This analysis will focus on the agenda-setting and priming/framing stages, based on personal interviews with the campaigners, campaign materials, television clips, and strategy papers of some parties.406 I will demonstrate that the No campaign was very strong; they not only built effective frames but also had remarkable civil society support that helped them amplify their frames at the local level. Moreover, they started earlier and placed their campaign themes successfully on the agenda. In turn, the Yes campaign was forced to take the defensive position and did not succeed in putting forward strong frames. They further suffered from credibility problems as some Yes camp parties were visibly divided. While the Socialist

403 Though the issue of European integration has been more contentious since the early 1990s, it has not gained sufficient electoral salience in the presidential or parliamentary elections. Flood, "French Euroscepticism and the Politics of Indifference," 42. 404 Ibid., 43. 405 Available at: http://www.ipsos.fr/canalipsos/poll/8074.asp#07 406 A full list of the campaigners I interviewed is presented at the end of this chapter. 151

Party, The Greens, and the UMP officially campaigned for a Yes, important figures from these parties joined the No campaign. As a result, the No campaigners succeeded in linking the existing contention to the referendum proposal, and the initial positive public opinion could not be sustained. I first provide a brief overview of the campaign and the campaigners. President Jacques Chirac called the referendum and was in favor of the Treaty. Even though he set the date in March 2005, he had already stated in July 2004 that there would be a referendum.407 The preparation period was long; the public debate and the campaign started before March 2005. The government ran an information campaign, led by the Ministry of Foreign Affairs, using television ads, websites and local initiatives.408 There was No uniform Yes campaign among the UMP, the UDF, the PS, and The Greens. Only at a few occasions they have attended meetings together. The UMP had the leading position in the public debate. Sarkozy, as the new president of the UMP, participated in group telephone calls in their campaign. The UDF was actively campaigning in the last two months, however public polls placed the UDF leader Bayrou behind the UMP leaders as best spokesmen for the TCE.409 The PS was careful not to share a common platform with Chirac as the PS voters were still resentful of having had to vote for him in 2002 against Le Pen.410 Importantly, the PS and The Greens were officially pro-TCE but de facto divided between the Yes and No camps. The No campaign was not uniform either. The left and the right wing campaigns were carefully separated from each other. Within the right Philippe De Villiers’s MPF has been the most vocal, and it did not cooperate with Le Pen’s FN. Le Pen’s campaign was said to be strategically low-profile due to his secure support and the danger that his association with the No camp could alienate other

407 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 12. 408 Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," 2. 409 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 54. 410 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 5. 152

potential supporters.411 However, the left wing No campaign was highly united in itself; the PCF, the LCR, parts of the PS and The Greens, and the left-wing civil society groups came together to build a No Committee. This committee had astonishing local implantation, having approximately 900-1,000 local committees all over France. In the No camp, three parties officially called for a No vote but did not run active campaigns. Among the left, the LO opposed the treaty though they stayed in the background and did not campaign with the other groups.412 Among the right, Charles Pasqua’s RPF and Bruno Mégret’s MNR remained largely absent from the media during the campaign.413 As I will discuss in detail below, just like the other three cases, the Yes camp had better access to financial resources and media attention.414

Agenda-setting If the campaigners start early, they have a better opportunity to set the tone of the debate. Thus, timing matters. The official campaign started on 16 May 2005 but in reality the campaign was much longer. The No campaigners started as early as October 2004 when the ‘Appeal of 200 – Say No to the constitutional treaty to build Europe’, signed by 200 representatives of left-wing groups, was published.415 Similar civil society initiatives from the Yes side by business leaders or legal experts were published only in May, the last month of the campaign. The French government launched the neutral information campaign in November 2004.416 The UMP, the government party, similarly started its campaign in November 2004 by sending a large number of envoys around the country to

411 Ibid.: 6. See also Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second-order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia." 412 Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," 4. 413 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 55- 56. 414 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 22. 415 L’appel des 200. Ibid. 416 Ibid.: 20. 153

organize meetings between different federations.417 Nonetheless, the Yes campaigners commonly stated that their active campaign was in the last two months, specifying March 2004 as their starting point. Chirac came into the discussion only in mid-April, when he debated the TCE with young voters.418 An important turning point in the campaign was the internal referendum of the PS in December 2004. In the aftermath of this referendum, the pro-TCE campaign scaled down whereas the anti-TCE campaign became highly active.419 The ATTAC has announced its decision to campaign against the TCE on 12 December 2004, which was followed by the General Workers’ Confederation (CGT – a major trade union) internal vote against the TCE despite its leader’s position.420 Soon after these statements, the Bolkestein Directive caused a vibrant discussion. By mid-March – only after two weeks of active Yes campaign – public opinion polls started to indicate a possible victory for the No camp.421 The No campaign’s weight and the organization, very early on in the process, caused the balance to shift and the belated efforts of the Yes campaign were useful but inadequate.422 This observation was confirmed by both Yes and No campaigners. Olivier Ubéda of the UMP stated that the Yes camp was on the defensive and not the offensive, which helped the No campaign, adding that ‘it was too late, very bad arguments were already out there’.423 Similarly, Isabelle Sicart from the UDF said that the main Yes camp strategy had to be to counter the No campaign arguments which were everywhere and for which the Yes campaign was not prepared.424 The No campaigners also confirmed this pattern. Jacques Généreux of the PS explained that the No campaigners, particularly the ATTAC, started to analyze the subject in 2003 and that the Yes campaign had to counter

417 Ibid.: 6. 418 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 5. 419 Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," 2. 420 Ibid. 421 Ibid. 422 Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," 10. Also, personal interview, 24 September 2008. 423 Personal interview, 9 September 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 424 Personal interview, 23 September 2008. 154

the No campaign: ‘The Yes campaign was not for a Yes vote, it was only against a No vote’.425 Therefore the Yes campaigners operated reactively and were forced to develop defensive arguments that were less convincing than the No arguments.426 The next section will look into this second aspect.

Priming and Framing I propose that three factors mattered in the priming and framing stage; strength of the campaign frames, credibility of the speakers, and mobilizational structures. The No campaigners primed their themes successfully in the media by coming into the debate earlier and sustaining an active campaign until the referendum. The media content analysis in the next section will demonstrate this in detail. Regarding framing, the Yes campaign suffered both from weak frames and weak credibility. The No campaign had stronger frames as well as excellent civil society mobilization.

Content of the campaign frames The No campaign frames were more effective based on the strength definition of the framing literature. Below I explain the content of the Yes and No campaign messages followed by a discussion of their strength. I identified the campaign frames primarily based on the written campaign materials.427 Next, I categorized them according to their subjects such as peace, economy, democracy, institutions, and other salient issues like welfare state, immigration, accession of Turkey, or national identity. The main Yes campaign frames can be categorized as: 1. Approving the TCE is approving Europe. The European integration has guaranteed peace in Europe for the last 60 years. 2. The TCE would institutionally enhance the integration project, and make it more democratic, efficient, and powerful in the world. The

425 Personal interview, 19 September 2008. 426 Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," 7. 427 Where available, one to two page leaflets listing the campaign arguments provided the best source. If not, I used interview data and secondary sources to complement the available materials. 155

new EU President and the Minister of Foreign Affairs, clarification of competences of the Union and the member states, and increased use of majority voting are important steps forward. These reforms would also help in formulating common environmental and immigration policies as well as a common fight against organized crime. 3. The TCE would render Europe stronger, with clear borders and privileged relations with its neighbors. As such, Turkey would only have a privileged partnership with the Union, instead of full membership. 4. The TCE and the Charter of Fundamental Rights would bring a social dimension to Europe with employment, social progress, fair trade and gender equality, and would help fight against exclusion and discrimination. Moreover, as these social objectives would be in a constitutional document all the member countries would have to comply. Public services would be recognized and could be exempted from the competition rules. 5. The TCE would make Europe more democratic and easier to reform, therefore making Europe more social would be easier. The UMP was the main Yes campaigner and has used the first three Yes frames. Using the slogan ‘L’Europe mérite un oui’ (Europe deserves a Yes), it highlighted the importance of peace as well as the institutional advances the TCE would bring. Interestingly, using the third frame, the UMP tried to appease those in the public who were against Turkish accession. Similarly, the UDF used the first two frames, and adopted the slogan ‘L’Europe a besoin de notre oui’ (Europe needs our Yes). On the center-left, the PS and The Greens chose to emphasize the social advances brought by the treaty. The PS used all Yes frames except the third, and used the slogan ‘Avec la gauche européenne, l’Europe sociale passe par le oui’ (Along with the European left, social Europe will be achieved through a Yes). In short, the PS stressed not only Europe’s role in maintaining peace and the institutional advances but also the social protection provided by the new treaty. The Greens similarly used all but the third Yes frames, and highlighted the

156 environmental progress specifically, adopting the slogan ‘Prenons les avancées du traite, votons oui’ (Consider the advances of the treaty, vote Yes). The PS and The Greens used the last Yes frame in order to counter the left-wing criticism of the TCE, to emphasize that further social improvement could be achieved after the TCE was accepted. Among the Yes campaigners was President Chirac, who primarily used the first frame, the role of Europe as the sole guarantor of peace and democracy in Europe. The No campaign frames can be categorized as: 1. The TCE is not social enough; it includes market-friendly provisions without sufficient social rights and protection. As such, it poses a threat to the French social model and to the competition for French jobs because of social and fiscal dumping and liberalization of public services. The services directive, the Bolkestein Directive, seeks to establish free movement of services within the EU. Importantly, the TCE is consolidating these ‘anti-social’ rules in a constitutional document, thereby ‘carving it in the marble’ (writing it in stone)428 and making it very difficult to reform. Economic policies as a matter of political choice should not be included in the TCE. (This frame strongly emphasized the pro-EU but anti-TCE position that another, more social, Europe is possible.) 2. The TCE is not democratic enough; neither the process nor the result is democratic. 3. The TCE subordinates the EU’s security system to NATO’s and militarizes Europe by increasing military expenditures. 4. The TCE is not environmentally-friendly. 5. The TCE promotes a Fortress Europe with restrictive immigration policies. 6. The TCE means the end of French sovereignty in a European ‘super- state’.

(428 Phrase first used by Yves Salesse, the co-President of Copernic Foundation and the co-initiator of the Appeal of 200, personal interview, 17 September 2008. 157

7. The TCE would increase immigration as more decisions on the subject would be taken at the European level. Furthermore, it would advance Turkish accession into the Union. The TCE and Turkish accession are linked as Turkey actively participated in the negotiations for the TCE. Combined, the TCE would increase Islamization of France. The main No campaigner was the united left-wing No Committee. As I explained, the PCF, the LCR, sections of the PS and The Greens, as well as civil society organizations such as the Copernic Foundation and the ATTAC came together. This platform used the first five No frames, and adopted the following slogans: ‘Non à la constitution Giscard oui à une autre Europe’ (No to Giscard’s constitution, Yes to another Europe – PCF); or ‘Pour une Europe sociale et démocratique, pacifique, écologique et solidaire, non à la constitution libérale’ (For a social, democratic, peaceful, ecological Europe with more solidarity, say No to the liberal constitution – LCR). Slogans such as ‘Another Europe is possible’ or ‘We will construct another Europe’ were frequently used. On the right-wing the main campaigner was De Villiers’ MPF, which used the first, sixth, and seventh frames. Adding to the social argument, De Villiers emphasized protection of France’s sovereignty and the risks of Turkey’s accession. The MPF adopted the slogan ‘Dire non c’est sauver nos emplois, un non qui va tout changer’ (No is to save our jobs, a No which will change everything); or ‘Turquie, constitution, délocalisations, non merci’ (Turkey, constitution, delocalization – relocation of industry – No thanks). Despite being less vocal, Le Pen’s FN used the exact same combination of frames as the MPF, and used the slogan ‘Je garde la France, je vote non’ (I protect France I vote No), coupled with statements such as ‘Vote No twice on 29 May, No to the constitution, No to Turkey in Europe’, or ‘Saying Yes to the constitution is saying Yes to Turkey’. Lastly Nicolas Dupont- Aignan, an MP of the UMP joined the No side based on a combination of first, second and sixth frames, arguing against the anti-social undemocratic TCE and defending French sovereignty. His campaign slogan was ‘J’aime l’Europe je vote non’ (I love Europe I vote No).

158

Relative strength of the campaign frames The framing literature suggests that frames are stronger if they can invoke available, accessible and applicable concerns. Further, if they are concrete, vivid, image-provoking, emotionally-compelling, and contain negative information, they tend to be more effective. The first and second Yes frames were not strong based on this strength definition as they were both abstract and technical arguments. The first Yes frame concerning peace was not an applicable concern as there was no imminent threat to European peace. Similarly, the second Yes frame presented the TCE as an institutional step towards a better Europe, which was neither available nor accessible as technical content of the treaty was difficult to grasp for the citizens. Neither of the first two frames used recent contention in society nor did they carry emotional content. In contrast, the third, fourth, and fifth Yes frames were indeed based on contentious issues such as Turkish accession and the deterioration in the welfare system. These frames invoked available, accessible and applicable concerns in the society. Yet they were stronger No campaign frames on the subject, as will be seen below. Among the No campaign frames, the first, sixth and seventh frames were the strongest ones given the fact that they were based on recent contention in society. They used available, accessible, and applicable concerns by tapping into problematic issues such as the decline in the French welfare system and increasing immigration coupled with Turkish accession. As I mentioned above, while the Yes campaign sought to counter these frames, the No campaign frames were more effective because they carried negative information, because they were more concrete in terms of the examples they provided, and because they invoked the fear of globalization. The second, third, fourth and fifth No frames were more abstract and weaker as they did not involve contentious issues. The campaign documents reflected the same pattern. The government’s information campaign distributed copies of the TCE to libraries and schools, and

159 the TCE was sold in bookshops.429 Moreover, every voter received a complete text of the treaty. The Yes campaign slogans and posters remained more symbolic and abstract in contrast with the more concrete and specific No campaign material, as Figures 4.6-4.12 demonstrate. Figure 4.6 presents the UMP’s poster showing a child with the EU flag reflecting in her eyes, while the FN’s poster in Figure 4.9 uses an image of Turkish Prime Minister signing the TCE. Although all the candidate countries were part of the Convention process that drafted the treaty, the FN chose to use this image to imply that the TCE would advance Turkey’s accession into the Union.

429 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 20. 160

Figure 4.7: The MPF’s poster

Figure 4.6: The UMP’s poster

Figure 4.8: The PS’ poster

Figure 4.9: The FN’s poster

Figure 4.10: Posters of the UDF and PCF

161

Figure 4.12: The ATTAC’s poster

Figure 4.11: The FN’s poster

Overall, the No campaigners designed more vivid posters with actual key words listed on them as opposed to abstract statements. Interviewees from both sides confirmed this analysis, commonly pointing out that the Yes campaign remained too much within the institutional details of the treaty. In clear contrast with the Spanish Yes campaigners, the French Yes campaigners mentioned the difficulty of the campaign and their misperception that they would easily win. Furthermore, they admitted that their arguments did not mobilize the voters in the way the No campaign arguments did. Ubéda from the UMP mentioned that the No campaign was strong as they used the profound fears of people, whereas the Yes campaigners could not explain their arguments:430 The No arguments were easier to understand than the Yes ones … They had the ears of the people that we did not see. They went where we had not been for too long, and they convinced the people to vote No, we could not convince enough people to vote Yes. … They caught the country, they were heard. We were not heard, we were too institutional, not close enough to people, too classical, too much for justification, not too political. They started early, and they had really good No arguments, we did not have good Yes arguments. It was as if you had to say Yes, because you just had to. We could not explain, that helped the No camp. People could remember their arguments and not our arguments. … They could talk to everybody, their arguments were about everything. Ours were more institutional, we had a gap between the Yes arguments and the

430 Personal interview, 9 September 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 162

national political spirit at that moment, it was pessimistic. We were too much on the happy side, and the people were not on the happy side. So, the No campaigners talked to them better than we did. … They caught people’s daily concerns more. It was not a matter of facts, it was a matter of feelings.

Alain Bergounioux of the PS explained that while most of their pro-TCE arguments were on institutional progress, the controversy in public opinion crystallized on the nature of Europe rather than the institutional aspect.431 According to him, this caused a gap between the concrete and social preoccupations of the people and the Yes campaign of the left, resulting in a less influential campaign. Similarly, Pierre Kanuty from the PS explained that the Yes campaigners called on ‘reason and political appreciation’, while the No campaign touched ‘people’s guts’.432 He added that their pro-TCE arguments sounded too distant and not socialist enough. On the other hand, the No campaigners argued that it was easy for them to link the socio-economic concerns with the TCE. Pointing to the same gap, Daniel Cirera – the international secretary of the PCF – explained that the Yes campaigners centered on the idea of peace, but that in the 2000s people were concerned about social policies instead.433 He added that people knew the contradiction between the French social policy and the European path as since the 1980s, French political elite frequently referred to European policies as the cause of social spending cuts. Alain Krivine, from the LCR and the leadership of the No Committee, explained the same phenomenon in the following way:434 Many people began to understand that there was a connection between Europe as it is now and their social situation in France. As you know this referendum has taken place in a moment where … there was big discontent with the social situation, for instance concerning the salaries and unemployment. There was also more and more privatization having a strong effect on the daily life of the population. I think if we have been able to interest people, it is not because we were able to give a big general speech about Europe, but because we tried to be very concrete, every time

431 Personal interview, 25 September 2008. 432 Personal interview, 8 September 2008. 433 Personal interview, 10 September 2008. 434 Personal interview, 25 September 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 163

we tried to show the links that existed between the national situation and the European decisions.

Similarly, UMP MP Nicolas Dupont-Aignan characterized the Yes campaign arguments as ‘either in the past or in technocracy’, explaining that their campaign would work better in the 1970s as the current concern was not peace anymore.435 Instead he stressed the importance of unemployment and relocation of industry today. Along the same lines, both the Yes and No campaigners of the PS highlighted the importance of the social concerns: Both Alain Bergounioux and Jacques Généreux suggested that between 1992 and 2005 Europe had taken a liberal path on key issues such as privatization of public services, and touched a sensitive spot in the French public opinion, in contrast with the expectations of the more left-wing section of the party.436 The No campaigners mentioned that the examples they used included Raffarin’s proposed pension reform, privatization in public services such as telecommunications or transportation, relocation of industry to Eastern Europe due to cheap labor or more advantageous tax systems, and the Bolkestein Directive. In France this directive became highly contentious, symbolizing the ‘Polish plumber’. It aroused the latent fears of potential influx of workers from the ten new Central and Eastern European member states.437 Francine Bavay, member of the executive committee of The Greens and a No campaigner, argued that the momentum of the No campaign was built specifically with the Bolkestein Directive debate, which helped people understand what was at stake.438 Even right-wing leaders such as De Villiers and Le Pen argued against ‘free-trade globalization’, rejecting relocation of industry, deregulation, and influx of immigrants.439 This combination, taken as a whole, tapped into the rising fear of

435 Personal interview, 24 September 2008. 436 Personal interviews with Alain Bergounioux on 25 September 2008, and with Jacques Généreux on 19 September 2008. 437 The fear was that the new directive would allow the incoming Central and Eastern European workers to be employed with lower working standards as a result of the rule of origin. Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 4. 438 Personal interview, 22 September 2008. 439 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 61. 164 globalization in France. The Yes campaign, faced with this No campaign framing, was unable to specify key issues in the same fashion. Overall, the No campaign managed to build stronger frames relative to the Yes campaign frames. The next sections will show that the credibility of the speakers and mobilizational structures further strengthened the No campaign frames.

Credibility of the Speakers The literature finds that credible sources increase the strength of campaign framing. I argue that in referendum campaigns, the impact of credibility can be assessed by looking into three components: Government popularity, scapegoating the EU, and disagreements within the parties. To start with government popularity, as I have explained above, the regular public opinion polls indicate that the French public was not satisfied with President Chirac or Prime Minister Raffarin. Specifically, the polls asked the public whether they trust them to address the major problems in France effectively, showing low levels of trust. This can be interpreted as a factor decreasing the strength of the Yes frames because they were not seen as credible. In cases where the governments were popular, there was less dissatisfaction that could compromise the government’s standing on the subject. The low trust in the ruling elite provides the No campaign with the opportunity to skillfully exploit this nerve among society. A commonly mentioned scene from the campaign was the debate between Chirac and the young people. Both the Yes and No campaigners mentioned Chirac’s detachment from the young people’s concerns such as unemployment and housing. Chirac, during this televised debate, openly admitted to not have understood the attitudes of the 18-29 year-olds.440

440 Christophe Piar and Jacques Gerstlé, "Le Cadrage du Référendum sur la Constitution Européenne: La Dynamique d’une Campagne à Rebondissements," in Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, ed. Annie Laurent and Nicolas Sauger (Paris: CEVIPOF, 2005), 50. 165

Frequently, other Yes campaigners called for Chirac and Raffarin to feature less prominently in the campaign.441 Second, the European level of governance was the traditional scapegoat for unpopular national measures in France.442 Particularly, as I discussed in detail, the French political elite justified the welfare state reforms which introduced market-friendly policies with the need to remain in the European Monetary System.443 This approach can be expected to decrease the speakers’ credibility in a Yes campaign as the positive frames conflict with the previous critical ones. This problem was mentioned within the context of the TCE referendum. Ricard- Nihoul explained that the use of the EU for a scapegoat for so long made it difficult for the Yes campaigners to suddenly turn around and ask the public to love it.444 Finally, disagreements within the campaigning parties reduce the credibility of the sources by sending conflicting signals to the voters. Among the Yes campaigners, the UDF was the most uniform, as 70-80% of the party members supported the TCE.445 Nonetheless, there were still prominent figures who expressed reservations, arguing against the Turkish membership in the EU. The UMP had an internal opposition group as some key party members led by Dupont-Aignan under the name souverainistes joined forces with the No camp.446 However the main division took place within the social democratic and green parties, the PS and The Greens. The PS held an internal referendum in December 2004, in which only 58.8% of party members voted in favor.447 The Greens also held an internal referendum in February 2005 and the decision to be in favor was

441 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 5. 442 Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," 3. 443 Schmidt, "Does Discourse Matter in the Politics of Welfare State Adjustment?," 187. 444 Personal interview, 24 September 2008. 445 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 7. 446 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 4. 447 Ibid.: 3. For a detailed analysis of the internal referendum and the split see Guillaume Duseigneur, "Lorsque l’Union Européenne Questionne l’Identité Partisane: Le Référendum Interne du PS sur le TCE," in Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, ed. Annie Laurent and Nicolas Sauger (Paris: CEVIPOF, 2005). 166 supported only by 52.72%.448 Given the disagreements, some top elected officials as well as activists joined the left-wing No campaign. Within the PS, well-known figures such as the previous Prime Minister Laurent Fabius advocated for a No vote. Other important figures such as Henri Emmanuelli or Jean-Luc Mélenchon even attended political meetings organized by the No campaigns of other parties.449 Généreux explained the split of the PS as a reflection of a broader spilt within the party since the 1970s.450 There have been two camps, first more centrist and closer to liberalization and deregulation, second more traditional and for more regulation. Strikingly, in the June 2004 EP election campaign the PS recommended amendments to the TCE on issues such as fiscal and social dumping. Généreux pointed to the clear contrast between the PS campaign materials for the 2004 EP elections and those for the 2005 TCE referendum. He explained that even though none of its recommendations were achieved, the PS campaigned for a Yes vote in the referendum. The contrast in two campaigns also creates a visible credibility problem, besides the split among the party. Bergounioux mentioned that the lack of party discipline and the partial participation in the No campaign was a major problem for the PS Yes campaign.451 The No campaigners mentioned the same split as a factor contributing to their campaign. Maxime Combes from the ATTAC explained that their main strategy was to target the center-left voters and show that they were not against Europe.452 He mentioned that the internal split of the PS greatly helped them in this regard. Therefore, the Yes campaign’s ineffective frames were also weakened by the credibility problems of the Yes campaigners due to low government popularity, frequent use of the EU as a scapegoat, and finally, disagreements within the parties.

448 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 4. 449 Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," 4. 450 Personal interview, 19 September 2008. 451 Personal interview, 25 September 2008. 452 Personal interview, 8 September 2008. 167

Mobilizational Structures Another factor that defines the success of a campaign is mobilizational structures. I divide this concept into financial resources and civil society contribution. First, in terms of financial resources, the Yes campaign had more resources paralleling the other three referendum cases. The official information campaign’s expenditures were not published; nonetheless, newspaper and monitoring reports mentioned figures around 10-20 million Euros.453 The political parties used both public and private sources for their campaigns. On the Yes side the UMP, the UDF, the PS, and The Greens; on the No side the PCF, the MPF, the RPF, and the FN received official state funding. The funds were distributed equally, each party could claim 800,000 Euros for posters/brochures/leaflets and meetings.454 Public means were allotted to these parties and each had space on the electoral communication boards to display their official campaign posters. The same parties were given air-time for campaign broadcasts.455 Civil society organizations did not receive any financial contributions from the state. Second, concerning civil society support, France had the strongest mobilization particularly among the left-wing No camp. In November, several organizations such as the New Republic, the Europa Nova, and the Robert Schuman Foundation organized a joint meeting in Paris to launch a platform for the Yes vote.456 In May, 100 business leaders signed an ‘Appeal in favor of a Yes to the constitutional referendum’. Similarly, 500 legal experts mounted an online petition in favor of the TCE. Nonetheless these Yes campaign civil society initiatives were not as intense as the left-wing No campaign mobilization. In October 2004, 200 representatives of left-wing groups – including political leaders, trade unionists, actors – published the appeal ‘Say No to the

453 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 13. Also see Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum." 454 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 14. 455 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 7. 456 Ricard-Nihoul and Larhant, "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum," 5. 168 constitutional treaty to build Europe’.457 The trade unions were divided on the subject and three of them joined the No camp.458 Therefore the main No campaign platform was formed by a coalition of left-wing political parties and civil society organizations; the PCF, the LCR, factions of the PS and The Greens, the Copernic Foundation, the ATTAC and the anti-TCE trade unions. Susan George, from the leadership of the ATTAC, explained that the local campaign groups were formed by different social forces, which first came to work together on the fight against the multilateral agreement on investment in 1997.459 Different social backgrounds included farmers, trade unions, development organizations, women’s organizations, ecologists, peace movements, and the ATTAC that was recently born. She explained that the ATTAC succeeded in bringing together all these forces primarily because it was not running for political office. She stressed that in 2005 the ATTAC local structure was extremely instrumental in pulling together these participants all over France. The 2005 left-wing No committee had remarkable implantation, having approximately 900-1,000 local committees in the entire country. The interviewees explained that these committees were formed by the local offices of political parties, trade unions and the ATTAC. They worked together to speak with people on the streets, organize big meetings, and distribute leaflets on a grand scale. Only in Toulouse, 22,000 issues of ATTAC brochures were disseminated. The campaigners mentioned that mainly the PCF and also the LCR backed the No campaign by adding significant financial resources and providing ad space which were provided only to parliamentary parties. Pierre Khalfa, a member of both the ATTAC and the Solidarity Unity Democracy, explained that primarily the ATTAC and the Copernic Foundation formed the No campaign arguments by their detailed analyses.460 Claude Debons of the CGT added that while the national committee decided on the logistics, the

457 L’appel des 200. Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 22. 458 The General Workers’ Confederation (CGT), the Workers’ Force (FO), and the Solidarity Unity Democracy (SUD) recommended a no vote. 459 Personal interview, 13 September 2008. 460 Personal interview, 22 September 2008. 169 leaflets and posters, the local level committees adapted the national-level decisions to their own reality.461 For instance Dominique Touraine of the PCF, who took part in the local campaign of a small town (Montlouis-sur-Loire – 12,000 inhabitants), explained that they not only used the PCF’s national campaign documents but also produced local brochures adapting the arguments for the local conditions.462 The Yes campaigners acknowledged the better local organization of the left-wing No campaign. Ricard-Nihoul mentioned that the Yes camp was much more institutional and less civil-society oriented, which gave them less opportunity to contact people.463 Patrick Farbiaz of The Greens specified their disconnection from social movements as a problem in the Yes campaign.464 Similarly, Pierre Kanuty of the PS mentioned that the local Yes campaign of the PS was not as professional as the ATTAC campaign.465 The right-wing No side did not have a strong local campaign. ‘Group for a Confederation of the States of Europe’ was opposed to the TCE on the basis of Gaullist principles, against supranational structures.466 Yet this campaign did not have similar support from the trade unions or strong civil society groups. Given the strength of the No campaign and the passionate debate around the subject even at the local level, the French campaign became the most intense among the four TCE referenda. A call center was set up to answer citizens’ questions and it recorded 2,500 calls a day and 110,000 calls in total (one call for every 380 voters).467 The media discussed the subject in great detail and a wide variety of material was published including non-traditional media sources such as internet blogs.468 Books on the treaty became best-sellers.469 According to a

461 Personal interview, 19 September 2008. He is also the co-initiator of the Appeal of 200 along with Yves Salesse. 462 Personal interview, 17 September 2008. 463 Personal interview, 24 September 2008. 464 Personal interview, 19 September 2008. 465 Personal interview, 8 September 2008. 466 Personal interview with Christophe Beaudouin, 15 September 2008. 467 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 20. 468 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 7. 170

French Survey Institute (SOFRES) survey, interest in the campaign raised from a 47% of the respondents in March, to a 60% in May.470 Similarly a French Institute of Public Opinion (IFOP) poll on the subject of conversation amongst French people showed that the referendum filled 26% of conversations in January, 48% in March, and 83% in May.471 The analysis of the campaign shows that the No campaign successfully countered initial positive public opinion on the subject. To recap, first, the No campaigners came into the debate earlier and placed their campaign themes on the agenda, forcing the Yes campaigners to take the defensive position. Second, the No campaign frames were stronger than the Yes campaign frames because they have successfully built on existing concerns in French society, and managed to link these problems to the referendum proposal. Third, credibility of the speakers and mobilizational structures have also strengthened the No campaign framing. The Yes campaign frames suffered from weak credibility due to government unpopularity, previous use of the EU as a scapegoat, and inner splits within the pro-TCE parties; whereas the No campaign framing was reinforced by its extraordinary civil society mobilization on the left-wing. In March 2005 the French cartoonist Million portrayed the same campaign assessment in two cartoons, shown in Figures 4.13 and 4.14.472

469 Corinne Deloy, "29th May 2005: A Devastated Landscape," Robert Schuman Foundation Report (2005). 470 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 7. 471 Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action," 3. 472 The second one portrayed the leader of the UDF François Bayrou hiding in the back as the leader of the PS François Hollande was getting beaten by the no campaign – by his own party members. Available at: http://www.ena.lu/ 171

Figure 4.13: Cartoon on the French 2005 TCE campaign

Figure 4.14: Cartoon on the internal split of the PS in the French 2005 TCE campaign

Analysis of the Vote In this section I use media content analyses and public opinion data to show the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. The assessment below confirms the previous campaign analysis; the No campaign framing was indeed stronger than the Yes campaign framing.

172

Shifts over Time As shown at the beginning in Figure 4.1, public opinion was positive in fall 2004, and the No intention gradually increased over time. Figure 4.15 illustrates the shifts in vote intentions over time, marking Chirac’s announcement of the referendum date:473

Source: IPSOS

Figure 4.15: Evolution of vote intentions in France (September 2004-May 2005)

Figures 4.16 and 4.17 show CSA data on voting intentions along with the main events of the campaign:474

473 Available at: http://www.ipsos.fr/CanalIpsos/articles/1545.asp 474 See Ricard-Nihoul, "The French 'No' Vote on 29 May 2005: Understanding and Action."; Gerstlé, "The Impact of Television on French Referendum Campaign in 2005." 173

Figure 4.16: Evolution of vote intentions in France and campaign events (September 2004-March 2005)

Source: CSA Figure 4.17: Evolution of vote intentions in France and campaign events (September 2004-May 2005)

174

As Figures 4.16 and 4.17 demonstrate, the No vote gained majority in the public opinion polls from March 2005 onwards. This date is particularly important as the active Yes campaign only started in March, indicating that the early start of No campaign was successful. In other words, the No vote was already rising when the Yes campaigners started. These graphs as well as most of the interviewees presented the Bolkestein Directive debate – a key No campaign theme – as crucial in raising the No vote intentions. Analysts suggested that only after Chirac tackled the directive in the European Council and it had been temporarily withdrawn that the Yes vote intention began to pick up.475 Furthermore, the intervention of previous Prime Minister Lionel Jospin in favor of the TCE in late April was interpreted as helping the Yes campaign.476 The No vote increased once again following the 16 May strikes against the removal of Whit Monday as a bank holiday. The graphs demonstrate that the No vote intentions have steadily risen throughout the campaign, which can be attributed to the success of the No campaign framing based on the following media content analysis and public opinion data.

Media Content Analysis Just like the other TCE referendum campaigns, the Yes campaign received better representation in the French media. However, as the assessment below will show, the No campaign was successful in priming its No campaign themes. Jacques Gerstlé conducted a detailed media content analysis of the six-month campaign, between 29 November and 29 May, based on the TV news coverage and surveys.477 Below I report the findings of this study, particularly on the time devoted to both camps and the topical coverage of the campaign.

475 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 4.; also see the analysis of the IPSOS available at: http://www.ipsos.fr/CanalIpsos/referendum/iv_tous_instituts.htm 476 Piar and Gerstlé, "Le Cadrage du Référendum sur la Constitution Européenne: La Dynamique d’une Campagne à Rebondissements," 51. 477 Gerstlé, "The Impact of Television on French Referendum Campaign in 2005." Also see Piar and Gerstlé, "Le Cadrage du Référendum sur la Constitution Européenne: La Dynamique d’une Campagne à Rebondissements." 175

During the six months, the time devoted to the Yes campaign reached 796 minutes, while it reached only 141 minutes for the No camp. Within the Yes camp the right-wing Yes campaign dominated the media coverage (650 minutes of the 796 minutes), whereas within the No camp the left-wing No campaign prevailed (87 minutes of the 141 minutes). This breakdown confirms the position of the UMP as the primary campaigner in the Yes camp, and of the united left- wing platform in the No camp. Table 4.2 demonstrates the topical coverage of each campaign regarding European issues:

Table 4.2: Topical coverage of the French Yes and No campaigns

176

This data considerably parallels the campaign frames outlined in the previous section. Therefore it proves that the campaign frames were picked up by the TV news in almost the exact way the campaigners intended. Another related question is whether the referendum became a common discussion topic for the French public. Table 4.3 shows the evolution of referendum debate in French people’s conversations:478 Date Subject Rank Percentage October 2004 Opposition of Fabius to the TCE 12th 29% November 2004 Turkey’s membership 7th 51% December 2004 PS internal referendum 10th 41% January 2005 Preparation for the referendum 12th 26% February 2005 Turkey’s membership 12th 40% Preparation for the referendum 26th 26% March 2005 The referendum 9th 48% April 2005 The referendum 4th 66% May 2005 Whit Monday 1st 84% The referendum 2nd 83% Table 4.3: Evolution of referendum debate in French people’s conversations

Table 4.3 demonstrates that the split within the PS and the upcoming decision to open accession negotiations with Turkey have captured public attention between October and February, followed by the referendum itself and the Whit Monday strikes. Attention to topics such as Turkey’s accession or strikes pointing at the declining welfare state was beneficial for the No campaign as they closely resembled their campaign themes. Concerning the topical coverage, Gerstlé points out that the news on the referendum campaign was characterized by the overwhelming presence of social issues. From mid-March onwards, with the key debate on the Bolkestein Directive, the European integration question was dramatized through the issue of enlargement. According to this analysis, the 14% drop in the Yes vote intentions over a 20-day period indicates that the social issues became decisive. A comparison of the TV news coverage of social issues in the 1992 Maastricht

478 IFOP data showing the percentage of those who recognized to have spoken these themes with their circle of friends and family, see Piar and Gerstlé, "Le Cadrage du Référendum sur la Constitution Européenne: La Dynamique d’une Campagne à Rebondissements," 55. 177 referendum with the 2005 TCE referendum presents a striking result: While the social matters took up only 250 minutes in 1992, this number reached 889 minutes in 2005 despite the fact that the unemployment rates were similar in these periods. The 2005 news coverage included unemployment, purchasing power, the 35-hour work week, the strong social mobilization of March 10th defending public services, the fears aroused by the Bolkestein Directive, the announcements of social redundancy plans and relocations, proposals for workforce reclassification in Romania or Mauritius, and social movements against the suppression of the Whit Monday bank holiday. In short, social issues overwhelmed the TV news. Therefore Gerstlé stresses that the social issues were primed successfully in the 2005 referendum campaign, as social questions under the media spotlight became particularly accessible concerns for the voters.

Public Opinion Data To what extent were these campaign messages echoed by the French public? Several polling agencies such as the CSA, the IPSOS, and the SOFRES conducted detailed post-referendum surveys to find out the reasons behind the Yes and No votes. To start with, the CSA post referendum survey asked the voters which issues were the most important to them when they were voting. Table 4.4 summarizes the answers:479

479 Available at: http://www.csa.eu/dataset/data2005/opi20050529d.pdf 178

Total Yes No (%) voters voters (%) (%) The social situation in France 41 23 55 The role of France in Europe 26 38 17 The place of Europe in the world 24 50 4 The content of the treaty 21 19 24 Turkey’s possible accession to the EU 14 6 20 The liberal or social direction of European politics 14 10 17 Globalization 11 10 12 Opinion on Jacques Chirac and the government 8 6 9 The enlargement of the EU to 25 countries 8 6 8 The position of political parties and politicians on the 5 4 6 TCE Don’t know 4 3 5 Table 4.4: Issues taken into consideration by the French voters

The fact that the Yes campaign frames are repeated mainly by the Yes voters, and that the No campaign frames are used mostly by the No voters confirm the relevance and necessity of frame analysis. The social situation in France prevails as the main concern, showing that this theme was indeed primed, paralleling the media content analysis above. More importantly, it was the most frequent response given by the No voters. This indicates that the most successful No campaign frame was the first frame blaming the TCE for not being social enough. For the Yes voters, the place of Europe in the world was the main concern, which could be related to the second Yes campaign frame arguing that the TCE institutional reforms would make Europe more powerful in the world. While this frame was not a strong frame based on the strength definition, it could be argued that this answer does not highlight the specific institutional details of the TCE but rather emphasizes a general support for Europe. More specifically, the motivations provided by the Yes and No voters are shown in Tables 4.5-4.6 below.480 It is important to note that the possibility to

480 Because the voters could select more than one option, the numbers represent the percentage of Yes/no voters who have chosen that statement The SOFRES data can be obtained from http://www.tns-sofres.com/; and the IPSOS data is available at: http://www.ipsos.fr/canalipsos/poll/8074.asp 179

pick more than one answer inevitably inflates the answers. To start with the Yes voters: IPSOS SOFRES The TCE increases the influence of Europe vis-à-vis the US 64 52 and China A constitution is necessary to ensure the functioning of the 44 21 institutions/of the EU of 25 The No vote can weaken the influence of France in Europe 43 42 The No vote can risk the continuation of European 34 44 construction The TCE is a step towards European political unification 28 26 The TCE is progress towards social Europe 26 The TCE is a progress in relation to existing treaties 25 19 European construction has been positive for France 19 Affinities with personalities supporting the ‘Yes’ vote 12 The other European countries rely on the French ‘Yes’ 11 Table 4.5: Reasons to vote Yes in France

The answers of the Yes voters show that there were a variety of reasons to support the TCE. Looking at both columns, the main factors behind the Yes vote were the influence of Europe in the world, the risk of weakening France’s role in Europe, the risk of blocking European integration, and ensuring the functioning of the EU institutions. Some components – particularly the last – in these four answers follow the second Yes frame that the TCE would institutionally enhance the EU. It is somewhat surprising as this frame is not a strong one based on the framing literature discussed above. However, there is some room for interpretation. Two of these answers reflect the fear of negative consequences instead of direct positive answers echoing the Yes frames. Moreover, these four answers also reflect the general and non-specific support for the continuation of European integration. The same IPSOS survey also found that among the Yes voters 97% was favorable (2% against) towards the European construction, whereas among the No voters only 40% was against (57% in favor) the European construction. This means that broad support for the European construction was indeed a key factor for the Yes voters. In contrast, this factor did not matter for the No voters as much. Looking at Table 4.5, the first Yes frame emphasizing the

180

EU’s role in maintaining peace or the third frame emphasizing Turkey’s possible privileged partnership were not reflected in the answers. The fourth and the fifth Yes frames used by the PS and The Greens which stated that the TCE would create a more social Europe is chosen only by 26% of the Yes voters (despite the possibility to pick more than one answer), showing its limited success. Social concerns and Turkey’s accession were contentious and thereby available issues in the society, which rendered these frames stronger. Yet, as the No motivations below will show, the counter frames that the TCE was too liberal and that it advanced Turkey’s accession were more powerful. The motivations of the No voters were: IPSOS SOFRES Discontent with the current economic and social situation in 52 France Annoyance with the current situation 40 Risk of making unemployment worse 46 The TCE/Europe is too liberal in economic policies 40 34 The TCE needs to be renegotiated 39 35 Turkey 35 18 The TCE is a threat to the independence/identity of France 32 19 Discontent with the political elite 31 European construction has been negative for France 27 Text too difficult 34 Affinities with personalities supporting the ‘No’ vote 12 Opposition to the government and to Jacques Chirac 24 Table 4.6: Reasons to vote No in France

The main issues highlighted by the No voters relate to the economic and social context in France, unemployment, and the liberal character of the economic policies in the TCE. These are very much in line with the first No campaign frame arguing that the TCE was not social enough. Moreover, the somewhat popular answer stressing the need to renegotiate the treaty demonstrates a rejection of the TCE itself, instead of a broad opposition to the Union. The secondary motivations concern Turkey’s membership and the protection of French sovereignty, reflecting the sixth and seventh No frames. On the other hand, the second, third, fourth and fifth No campaign frames are not represented in the answers. As such,

181

Table 4.6 confirms the framing analysis above by showing that among the No campaign frames, the first, sixth and seventh frames were indeed the stronger ones given the fact that they were based on recent contention in society. 24% of the No voters specify their opposition to Chirac and the government as one of their motivations. This is not a high number especially because the respondents were given the chance to select more than one answer. Similarly, the CSA October 2005 poll asked the French public whether they think Jacques Chirac caused the No result in the referendum. Only 23% answered positively, while 58% answered negatively.481 These figures confirm that the opposition to the government does not define the referendum result per se, instead it should be analyzed in relation to the campaign. There are more survey results that illustrate the centrality of socioeconomic concerns in the No vote. The CSA poll in April 2005 asked the French public how much the social situation would matter in their vote on the TCE, 59% said a lot or sufficiently, and 33% said little or none.482 More importantly, SOFRES data from May 2005 showed that both social concerns and opinion on Turkey’s EU membership conditioned the French public’s voting intentions.483 Table 4.7 cross-tabulates these two concerns with the voting intentions, proving that the first one was more central in the No vote. Voting intentions Yes No The TCE would guarantee social rights 82% 18% The TCE would not guarantee social rights 19% 80% Favorable to Turkey’s accession 63% 36% Opposed to Turkey’s accession 44% 55% Table 4.7: French public’s evaluation of various campaign topics – the Yes/No breakdown

The survey results also indicate that the Yes vote was mainly from the right-wing, while the No vote primarily came from the left-wing. The IPSOS data

481Available at: http://www.csa-fr.com/dataset/data2005/opi20051007b.htm 482 Available at: http://www.csa.eu/dataset/data2005/opi20050317b.htm 483 Brouard and Sauger, "Comprendre la Victoire du "Non" : Proximité Partisane, Conjoncture et Attitude à l’égard de l’Europe," 132-34. 182 in Table 4.8 shows the following breakdown of the Yes/No vote by party proximity:484 Yes No voters voters (%) (%) Far left 6 94 PCF 2 98 PS 44 56 The Greens 40 60 LEFT 37 63 UMP 80 20 UDF 76 24 MPF 25 75 FN/MNR 7 93 RIGHT 73 27 No Party 31 69

Table 4.8: The breakdown of the French Yes/No vote by party proximity

The primacy of the left-wing in the No vote once again signals that the first No campaign frame was the most effective. Table 4.8 reflects not only the centrality of social concerns, but also the problems the center-left had in persuading their followers. The PS and The Greens were split and thereby not credible in their campaigns. Furthermore, the CSA May 2005 poll cross-tabulated vote intentions with trade union proximity, which shows that without exception the followers of all trade unions were divided on the subject:485 General Workers’ Confederation (CGT) (26% Yes, 74% No), French Democratic Confederation of Labour (CFDT) (54% Yes, 46% No), Workers' Force (FO) (36% Yes, 64% No), French Confederation of Christian Workers (CFTC) (63% Yes, 37% No), Solidarity Unity Democracy (SUD) (35% Yes, 65% No), General Confederation of Executives (CGC) (65% Yes, 35% No), Peasant Confederation (29% Yes, 71% No).

484 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 8. 485 Available at: http://www.csa-fr.com/dataset/data2005/opi20050529e.pdf 183

Once again, the party share of the No vote confirms that it was mainly from the left-wing, as can be seen in the pie chart in Figure 4.18:486

Source: IPSOS Figure 4.18: Party share of the French No vote

While the majority of the No vote came from the left-wing, there was still a considerable right-wing portion. In an IPSOS exit-poll, Turkish accession was cited as the first reason for voting No by the UMP and the FN/MNR supporters.487 This shows once more that the seventh No frame on potential Turkish accession was effective as well. The public opinion data points to the success of mainly the first (anti- social TCE) and the seventh (Turkey’s membership) No campaign frames. Detailed statistical studies of the French case have also emphasized the socioeconomic anxiety, particularly concerning the possible negative socio-economic consequences of the TCE.488 However, the important point is that this general

486 Marthaler, "The French Referendum on Ratification of the Constitutional Treaty ": 9. 487 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 58. 488 See Brouard and Sauger, "Comprendre la Victoire du "Non" : Proximité Partisane, Conjoncture et Attitude à l’égard de l’Europe." ; Bruno Cautrès, "Les Clivages Socio-politiques sur l’intégration Européenne et le Vote du 29 Mai 2005," in Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, ed. Annie Laurent and Nicolas Sauger (Paris: CEVIPOF, 2005). 184 socio-economic anxiety already existed in fall 2004 when the public was in favor of the treaty. There was only an episodic dip in the public opinion towards the European integration. The CSA post-referendum survey found that 66% of the public wanted France to ask for a new Constitution, which corresponded to 67% of the Yes voters and 66% of the No voters.489 Similarly, the IPSOS post- referendum poll indicated that 72% were in favor of continued European integration.490 My argument is that the No campaign succeeded in linking the existing discontent to the referendum proposal. The referendum vote was a result of both domestic and European factors. According to the CSA post-referendum poll, 60% of the French public thought about European construction while 35% thought about national problems when they were voting. Among the No voters, 42% came to this choice thinking of Europe and 52% said that they thought of national problems first. On the other hand, 82% of the Yes voters were motivated by European construction rather than national problems. Similarly, the SOFRES post-referendum survey reports that the majority of the Yes voters focused their decision on the European dimension (52%), while the No voters were more motivated by France’s economic and social situation (47%).491 These percentages are somewhat in the middle and show that the vote was the result of a mix, caused by factors in both domestic and European domains. Overall the public opinion data confirms the expectations of the frame analysis presented above. The No campaign framing was indeed more effective than the Yes campaign framing. The anti-TCE frames succeeded in tapping into the socio-economic and immigration-related concerns by emphasizing the liberal economic content of the treaty as well as Turkey’s potential membership. Only the second Yes campaign frame proved more successful than expected. Yet, as discussed, because these survey responses do not refer to the specific institutional content of the TCE, they could also be read as reflecting broad support for the European integration.

489 Available at: http://www.csa.eu/dataset/data2005/opi20050529d.pdf 490 Available at: http://www.ipsos.fr/canalipsos/poll/8074.asp#07 491 Gerstlé, "The Impact of Television on French Referendum Campaign in 2005." 185

Diffusion The French voted on 29 May 2005 and became the second country to hold a referendum, three months after the Spanish campaign. I argue that the later a country held its referendum relative to other states, the more the previous referenda results and campaigns in the other states influenced its campaign dynamics and public. France is a crucial case in this regard because not only did the Spanish case fail to have an impact on the French one but France also influenced all the other cases – Spain and the Netherlands to a small extent and Luxembourg to a large extent. The case chapter on Luxembourg will explain this impact in great detail. This section will ask: Why was the French not influenced by the Spanish referendum but was able to influence Luxembourg to such an extent? I propose two explanations. First, I argue that cross-case influences are conditional, and depend on the existence of transmission belts between the states such as shared language/culture, common media sources, and collaborative networks/transnational linkages. The more that these channels are open, the more the later cases can be influenced by previous ones. Below I will demonstrate that France did not share substantial diffusion channels with the Spanish or the Dutch cases, with the exception of limited collaborative networks/transnational linkages. In contrast, it was connected to Luxembourg with remarkable diffusion channels. Second, I argue that campaign intensity mattered. As the French campaign was the most intense campaign of all, it created a strong momentum and put significant amount of pressure on the existing diffusion channels. To start with the first channel, shared language and culture, France did not speak a common language with Spain or the Netherlands. In contrast, France and Luxembourg had a high level of lingual and cultural connection. The importance of this factor was mentioned in the interviews as a facilitator of cooperation with Luxembourg. Khalfa from the ATTAC explained that they had invited people from other countries only if they spoke French.492 Similarly Yves Salesse, the co-President of Copernic Foundation, mentioned that he was invited

492 Personal interview, 22 September 2008. 186 to Luxembourg but not to the Netherlands due to the language problem.493 Well- known campaigners such as Henri Emmanuelli, José Bové, Raoul-Marc Jennar, and Francine Bavay participated in the Luxembourg campaign. Importantly, both French and Luxembourgish campaigners acknowledged that after the French referendum, the unused No campaign materials (leaflets and posters in French) were delivered to Luxembourg to be used in their No campaign. Second, regarding the common media channels, French television channels and newspapers were available in Luxembourg on a daily basis. However, the Spanish or the Dutch did not share any media channels with France. This was an important difference, and the Luxembourg case chapter will show how the common media sources played an important role in connecting the campaigns in France and Luxembourg. The third and the final channel, collaborative networks and transnational linkages, was not only limited to France-Luxembourg. Within this channel, I investigate the institutional networks and personal connections by looking at the European Parliament groups, the European anti-globalization network, ad hoc European networks, and existence of mobile communities. Here, all four cases shared the institutional networks, but once again it was only France and Luxembourg that had personal connections. I discuss each in turn. In terms of the institutional networks, France was connected with all the other cases through the EP groups as well as the European anti-globalization network. French political parties are members of the EP groups. All campaigners from political parties explained that there were discussions with their EP party groups which included similar Spanish, Dutch and Luxembourgish political parties. As was the case in all other countries, the campaigners mentioned occasional meetings in which they shared arguments with their counterparts but also pointed out that the campaigns were run separately. Also, the heads of states visited one another during the campaigns. German Chancellor Schröder and Spanish Prime Minister Zapatero spoke twice before the French parliament, and

493 Personal interview, 17 September 2008. 187

Luxembourg’s Prime Minister Juncker also appeared in the debate.494 Nevertheless, just as in the other cases, the No campaigners stated stronger collaboration on a day-to-day basis. Especially the far left campaigners explained that they had permanent discussions with their counterparts. On the center-left, the campaigners from the PS and The Greens mentioned the EP group pressure on their parties to adopt pro-TCE positions. Concerning the far right, it was frequently mentioned that the right-wing No campaigners were less organized at the European level. Within the institutional networks, an important component is the European anti-globalization network. In all four countries, the ATTAC and the European Social Forum meetings (specifically the 2003 ESF in Paris, and the 2004 ESF in London) were commonly mentioned by the far left political parties and the civil society organizations as significant platforms where the TCE was debated. The campaigners in the other three countries specified the ATTAC France as a reference point, referring particularly to their early (2003) theoretical analysis of the TCE. Similarly, the book by Yves Salesse, one of the key left- wing campaigners in France, was mentioned as a resource. The French campaigners confirmed that the ATTAC France provided the first and main left- wing analysis of the TCE. They stated that at the November 2004 ESF meeting they distributed their analysis and the ‘Appeal of 200’, to all other European left- wing political and social forces.495 The ATTAC France has even financially contributed to the ATTAC Luxembourg’s campaign to a small extent.496 The French left-wing No campaigners explained their motivation to act European and to build a European opposition in order to show that it was not only a French resistance, ‘not a Franco-French affair’.497 Overall, the anti-globalization network in Europe has spread the arguments which originated in the Paris ESF and the ATTAC meetings and thereby formed a left-wing master frame, advocating an

494 Richard and Pabst, "Evaluation of the French Referendum on the EU Constitution, May 2005," 17. 495 Personal interview with Maxime Combes, 8 September 2008. 496 Personal interview with Pierre Khalfa, 22 September 2008. 497 Personal interviews with Susan George on 13 September 2008, and with Francine Bavay on 22 September 2008. 188 alternative ‘social’ EU. In turn, for the other referendum cases it was important to adapt these frames to the local reality, which I discuss in the case chapters. Apart from these institutional networks, the other aspect within the third channel is the personal connections. Only France and Luxembourg shared a mobile community. Due to Luxembourg’s peculiar geography and the institutional setup, which I will discuss in detail in the Luxembourg case chapter, these two countries were connected with mobile communities composed of students and commuters. For instance, student left-wing activists from Luxembourg mentioned that they participated in both campaigns and used the ATTAC Campus – a sub-group of the ATTAC – as a platform. This was confirmed by the French campaigners. Despite being the second in line, France made a bigger impact than Spain on the subsequent cases, minimally on the Netherlands, and largely on Luxembourg. The shared diffusion channels have played a significant role in connecting the two campaigns. However, as the Spanish case has shown, campaign intensity also matters. The quiet campaign in Spain did not generate the expected momentum, whereas the highly intense French campaign did. The difference is visible in the absence of Spanish influence in the existing collaborative networks/transnational linkages channels among all four cases. The ATTAC France’s analysis of the TCE was even influential on Spain, a case which came before itself, due to their early start. The French media content analysis or the public opinion data does not show a related change after the Spanish vote; there is neither an intensive coverage of the Spanish campaign nor a shift in the voting intentions. In contrast, the French influence is visible in the data on the Dutch and Luxembourgish campaigns. The Dutch referendum study by Aarts and Van der Kolk show that only the French referendum – not the Spanish – was widely known by the Dutch electorate.498 As such, France became the momentum case with an intense campaign and strong diffusion channels.

498 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 27. 189

The Momentum Campaign Ivaldi describes the 2005 French vote as a retrospective evaluation of the EU’s model of social and economic governance, since macroeconomic social issues and fears of globalization were central to the referendum campaign.499 In 1992 President Mitterrand argued for a European social model, which he secured by including a social chapter into the Maastricht Treaty. In May 2005, evaluating the post-Maastricht European social system, the French left-wing voted negatively.500 The existing studies parallel the finding that social fears were decisive in the No result, besides highlighting other factors such as government unpopularity and scepticism towards Europe. What the literature has ignored so far is the following: Despite the existence of all the fears, dissatisfaction and scepticism concerning both domestic and European domains, the fall 2004 polls showed positive attitude towards the TCE. I argue that the main reason for losing this support was the relative No campaign strength. It was the key factor which connected the existing contention with the referendum proposal in the French public’s mind. The No campaigners successfully placed their themes on the agenda and linked the present contentious issues to the TCE. The Yes campaigners were forced to take the defensive ground and failed to build effective pro-TCE arguments. Furthermore, the French referendum has built momentum and became the most influential case among the four because of its highly intense campaign and strong diffusion channels particularly with Luxembourg. The negative result of the referendum presented a sudden but episodic dip in the public opinion. The post-referendum polls showed that the majority of the French public was in favor of continued European integration and even a new European constitution. What happened was a sudden crystallization of negative opinion towards the TCE as a result of effective anti-TCE framing that capitalized on the existing social anxieties.

499 Ivaldi, "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second- order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia," 59. 500 Ibid.: 64. 190

PERSONAL INTERVIEWS

Political parties:

Yes campaigners: Olivier Ubéda (UMP, European Affairs Delegate) Alain Bergounioux (PS, Secretary General of the PS Scientific Council) Pierre Kanuty (PS, International and European Affairs Political Assistant) Patrick Farbiaz (The Greens, International Secretary) Isabelle Sicart (UDF)

No campaigners: Nicolas Dupont-Aignan (UMP, MP) Jacques Myard (UMP, MP) Jacques Généreux (PS Member and Economist) Francine Bavay (The Greens Executive Committee Member, and the Vice-President of the Regional Council of Ile-de-France) Daniel Cirera (PCF, International Secretary) Dominique Touraine (PCF) Alain Krivine (LCR, also part of the national leadership of the No Committee) Catherine Salagnac (National Front)

Civil society:

Yves Salesse (Co-President of Copernic Foundation, Conseil d’Etat Member, Co-initiator of the Appeal of 200) Claude Debons (General Workers’ Confederation CGT, Co-initiator of the Appeal of 200) Pierre Khalfa (ATTAC, and Solidarity Unity Democracy) Susan George (Part of leadership of the ATTAC) Aurélie Trouvé (ATTAC) Maxime Combes (ATTAC) Christophe Beaudouin (Secretary General of the ‘Group for a Confederation of the States of Europe’, campaigned with the MPF) Raoul-Marc Jennar (Member of the No Committee) Gaëtane Ricard-Nihoul (Secretary General of Notre Europe)

191

CHAPTER 5 THE NETHERLANDS: THE TURMOIL CASE

Only three days after the French rejection, on 1 June 2005 the Dutch became the second country to reject the European Constitution. 61.5% of the Dutch voters rejected the treaty, while 38.5% approved it.501 The Dutch No vote was not only the highest among the four cases but also the turnout rate was 63.3%, demonstrating unexpectedly high interest in the vote. The result shocked Europe for a second time within a matter of days. The high negative vote was puzzling for two reasons. First, just as in the other cases, the Dutch public opinion was in favor of the TCE at the outset. Eurobarometer surveys demonstrated that 73% of the Dutch population supported the general idea of a constitution for the EU in fall 2004.502 Furthermore, the November 2004 special Eurobarometer showed that 63% of the Dutch population supported the TCE specifically, while those who were against were limited to 11%.503 A national public opinion agency, Marketresponse, conducted a similar survey on behalf of the Government Information Service and the Ministry of Foreign Affairs, finding that 61% of the population was still in favor of the TCE in January 2005.504 Second, the political party preferences paralleled the other referendum cases. The mainstream political parties were in favor of the TCE: The Christian Democratic Appeal (CDA), the Labor Party (PvdA), the GreenLeft (GL), the conservative liberal People's Party for Freedom and Democracy (VVD), and the social liberal Democrats 66 (D66) were all pro-TCE. The opponent parties were from the extremes: The far left Socialist Party (SP), the small Orthodox protestant parties. Christian Union (CU) and the Reformed Political Party (SGP), the (now defunct) far right List Pim Fortuyn (LPF), and finally the far right MP Geert Wilders who had split from the VVD in 2004. If both the

501 The Electoral Council (Kiesraad) at: www.kiesraad.nl 502 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 503 ———, "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution." 504 Dekker and Wennekers, "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen." 192 initial positive public opinion and the political party preferences were similar, why was the early positive attitude not sustained in the Netherlands, why did the Dutch reject the TCE? Furthermore, did diffusion play an important role? Did the Dutch vote simply follow the previous cases and did it influence Luxembourg? The Dutch and French campaign dynamics are surprisingly similar. I argue that the Dutch No campaigners came into the debate much earlier than the Yes campaigners and their strong No campaign succeeded in linking the existing contentious issues to the TCE. The literature stressed the dissatisfaction with the domestic elites and certain European policies as the causes of the negative vote. However, just as in the French case, the literature ignored a key fact. As I will show below, these issues relating to both the domestic and EU domains were equally contentious when the public supported the TCE in fall 2004. In this period, prior to the beginning of the No campaign's activities, the two remained unlinked in the public’s mind. Thus I argue that the main reason behind the Dutch rejection was the relative strength of the No campaign. While the No campaign strategically tapped into the latent fears of society, the Yes campaign was dramatically unable to counter these effective anti-TCE frames. The result was a case of turmoil, where public anxieties were suddenly activated, translating into the highest negative vote rate among the four cases. The Yes campaign, just as in France, failed to sustain the initial positive public opinion. Second, because the Dutch case only shared limited diffusion channels with the other cases, it was only minimally affected by the French case, and in turn did not have a substantial impact on Luxembourg. Despite being the third referendum in line, diffusion was not substantial in the Dutch case. To demonstrate my argument, I will first present the shifts in the vote intentions over time to show the evolution of public opinion. Second, I will explore Dutch public opinion to discern the salient contentious issues of the period, relating to both the domestic and EU levels of governance. Third, based on extensive interview data, I will provide an analysis of the long campaign period examining the Yes and No campaign strategies. In this campaign analysis, I will look into the strength of the campaign frames, credibility of the speakers,

193 and mobilizational structures available to both sides. I will discuss how the strong No campaign linked contentious issues to the referendum proposal with effective anti-TCE frames, while the Yes campaigners failed to build strong arguments to counter them. Fourth I will analyze the referendum vote, based on media content analyses and public opinion data, to show the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public in turn. This section will confirm the campaign frame analysis. Finally, I will focus on the limited diffusion channels the Netherlands shared with the other three cases, and explore the extent to which diffusion was at work.

Evolution of the Public Opinion Kees Aarts and Henk van der Kolk combined all public opinion polls which reported the evolution of the Yes and No vote intentions over time.505 In Figure 5.1 the lines represent the percentage of No voters among those who were decided. In other words, the graph excludes the undecided voters, and thereby inflates the level of No vote intention in the beginning, when a large part of the public was still undecided.

505 Kees Aarts and Henk van der Kolk, eds., Nederlanders en Europa: Het Referendum over de Europese Grondwet (Amsterdam: Utigeverij Bert Bakker,2005), 148. 194

Figure 5.1: Evolution of No vote intentions in the Netherlands

As can be seen, public opinion was in favor of the TCE in fall 2004. Except for the data collected by Maurice de Hond,506 all the lines above show a rough increase in No vote intentions until mid-May, followed by ups and downs for the rest of the campaign.507 I will explain these shifts in greater detail below.

Discontent with Domestic and EU Governance The three years leading up to the referendum had been a striking period for the Netherlands. Dutch politics faced a period of political unrest and turmoil from 2002 onwards due to economic downturn, unpopular government reforms of the welfare state, and contention over national identity and immigration. As a result, the pubic was dissatisfied with the domestic level of governance. At the EU level,

506 De Hond data is less reliable than the data from other sources because he is using a self- selective panel where everyone can participate, based on reported voting behavior (around each election) and based on some other reported characteristics (gender, place of residence, etc.) The other bureaus are using less self-selected panels and are more explicit in explaining their methods of computing the percentages. 507 For a detailed analysis of these fluctuations, see Kleinnijenhuis, Takens, and van Atteveldt, "Toen Europa de Dagbladen Ging Vullen." 195 the Dutch have been one of the most enthusiastic followers of the European integration project. Nevertheless, just as in the other cases, the public had certain fears. Recently the political elite have also begun stressing the need to protect Dutch interests in certain policy domains. Below I explain the contention relating to both domains, which was already present in fall 2004 when the polls showed a positive attitude towards the TCE. The contentious issues provided campaign material for the No side, but importantly they remained unlinked to the referendum proposal before the No campaign successfully tapped into them.

Contention concerning domestic politics The first government of Jan Peter Balkenende was a center-right coalition, formed in 2002 between the CDA, the LPF and the VVD. The far right populist LPF was a new party, which gained a remarkable number of seats in 2002, a factor that I will discuss in detail below. This government was very unstable and fell within the same year after an internal disagreement between the two LPF ministers.508 The second Balkenende government, a coalition of the CDA, the VVD and the D66, came into power in 2003. The relatively stable 1990s was an era of ‘purple’ coalitions of parties from the left and right.509 Nonetheless, the general satisfaction with the Dutch government and public administration began to fall in 2000. In 2000, 77% of the public was satisfied with the government, while the figures dropped to 59% in 2002 and 48% in 2004.510 Figure 5.2 shows the clear dip in public opinion towards the government between 2002 and 2004.

508 Theo Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006 (The Hague: SCP -The Netherlands Institute for Social Research 2008), 29. 509 Ibid., 136. 510 Ibid., 132. 196

Figure 5.2: Satisfaction levels with the Dutch government (1995-2006)

What were the main reasons for this dip? As mentioned at the outset, from 2002 onwards, Dutch social and political life had been characterized by unrest and turmoil primarily because of an economic downturn and welfare state reform on the one hand, and controversy on national identity and immigration on the other. I will discuss each in turn. To start with the economic and welfare-related concerns, it must be acknowledged that the Netherlands has a strong welfare system providing generous social protection. In the 1960s the Dutch became a leader in welfare spending but in the mid-1970s it entered a long crisis.511 Becker and Cuperus

511 The WAO (Labor Disability Act), an ambitious law introduced in 1967, gave disabled workers the right to a replacement income, providing security and universal coverage regardless of cause. But in the 1980s the Dutch ended up spending 26% of GDP on social security transfers, more than any other country. The ratio of welfare dependents to workers increased tremendously, unemployment quadrupled, and less than 50% of working-age adults actually worked. See Lechner, The Netherlands: Globalization and National Identity, 175-77.; Evelyne Huber and John 197 characterize the first thirty years after the Second World War as spent on developing the welfare state and the next thirty dedicated to reform, adaptation, modernization and economizing: Thirty fat years were followed by thirty lean ones.512 Public spending declined and policies shifted from protecting labor to promoting work, from social support to social inclusion, from guaranteeing universal entitlements to selective targeting of deserving groups.513 Moreover, the economic growth of the 1990s largely reversed itself in 2001, and had adverse effects on income, labor markets, employment, the accessibility of housing and the affordability of homes.514 Unemployment steadily increased between 2002 and 2005, from 4% to 7%.515 As a result, the second Balkenende government – the government that held the referendum – embarked upon a major reform in 2003.516 This reform was a continuation of the old politics of reform, but this time with an extensive agenda with substantial interventions in social security with a tough program of cutbacks in housing and pensions.517 Despite supporting previous rounds of reform, the social democrats (PvdA) did not support this move, questioning the spirit of the proposal, as social security was framed as a burden rather than a keystone of

Stephens, Development and Crisis of the Welfare State: Parties and Policies in Global Markets (Chicago: University of Chicago Press, 2001). 512 Frans Becker and Rene Cuperus, "Welfare State Blues: Thirty Years of Reform Policy in the Netherlands," in Public Discourse and Welfare State Reform: The Social Democratic Experience, ed. Vivien Schmidt (Amsterdam: Mets & Schilt, 2005), 135. 513 Lechner, The Netherlands: Globalization and National Identity, 197. 514 Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006, 21. 515 Ibid., 49. These figures are published by Statistics Netherlands. The OECD statistics show an increase from 2% to 4% instead. Available at: http://stats.oecd.org/index.aspx?queryid=21760 516 Its mottos were ‘more responsibility for individual citizens, a greater sense of belonging together’ and ‘reform of a welfare state that has become bogged down in bureaucracy’. Ibid., 30. Also see Wim van Oorschot, "The Dutch Welfare System: From Collective Solidarity towards Individual Responsiblity," in The Handbook of European Welfare Systems, ed. Klaus Schubert, Simon Hegelich, and Ursula Bazant (Abingdon: Routledge, 2009). 517 Income protection was restricted to groups who genuinely had no alternative. The Work and Social Assistance Act (WWB) was designed to incorporate more benefit claimants back into work, and other arrangements such as subsidized jobs and housing benefits were subjected to thorough scrutiny. The first three cabinets of Balkenende, between 2002 and 2006, took steps to prepare the rental market for liberalization as an incentive to encourage housing which matched income and thus to foster movement up the housing ladder. Moreover, the previous early retirement (VUT) and pre-pension schemes were replaced with a limited life course arrangement, funded by individual employees themselves. Becker and Cuperus, "Welfare State Blues: Thirty Years of Reform Policy in the Netherlands," 139.; Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006, 30. 198 society.518 These reforms were also highly unpopular among the public. In October 2004, reacting to the early retirement reform, hundreds of thousands protested on the streets.519 This was the greatest mass demonstration since the Second World War.520 Survey data suggests that since the 1990s, support for more equality had grown across the political spectrum, and by 2004 two-thirds of the Dutch thought income differences were too large.521 Views on government expenditure illustrated a growing preference for increased government spending and declining support for spending cuts between 2004 and 2006.522 Another survey before the 2006 general elections found that a large majority thought the Netherlands had become less socially sensitive under the second Balkenende government.523 The director of the Social and Cultural Planning Office, Paul Schnabel stated that ‘the public wants to go down the Scandinavian route, and the government down the United States route’.524 Minorities became another salient issue in Dutch politics at this time. The proportion of non-Western immigrants had increased from 7.3% in 1995 to 10.6% by 2007, with two-thirds of these immigrants originating from Turkey, Morocco, and Suriname.525 The European Social survey measures the ‘ethnic distance’, which is the degree of resistance to social interaction with members of ethnic groups in various domains such as the residential neighborhood, work, and personal relationships.526 The ethnic distance in the Netherlands, along with France, is close to the EU average, but higher than the Spanish and Luxembourgish figures.527

518 Lechner, The Netherlands: Globalization and National Identity, 196. 519 Ibid., 200. 520 Becker and Cuperus, "Welfare State Blues: Thirty Years of Reform Policy in the Netherlands," 141. 521 Lechner, The Netherlands: Globalization and National Identity, 201. 522 Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006, 116. 523 Lechner, The Netherlands: Globalization and National Identity, 204. 524 Becker and Cuperus, "Welfare State Blues: Thirty Years of Reform Policy in the Netherlands," 141. 525 Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006, 13. 526 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 74. 527 Ibid., 75. 199

While the government pursued a ‘multicultural society’ policy in the 1980s, in the 1990s the discourse changed to the need for ‘integration’ into the nation.528 In particular, after the terrorist attacks of September 11, the place of Islam in Dutch society became the subject of heated debate.529 The increasing immigration into the Netherlands led to a new kind of self-questioning: Were Muslims ‘Islamizing’ the Dutch culture?530 The Dutch became anxious to maintain their identity as one of the most liberal societies on earth.531 The anxiety was activated by incidents in which imams denounced homosexuals and by other countries’ criticisms of Dutch liberal policies. Indeed, two shocking political assassinations showed that immigration had become a highly contested topic. The first took place in 2002, as the rise and assassination of the charismatic populist politician Pim Fortuyn dragged the country into a crisis. Openly homosexual, a Marxist sociology professor, and a member of the PvdA, Fortuyn became a right-wing politician in 2001.532 He particularly criticized Islam and its impact on Dutch society, saying that ‘as a result of our disinterest in our own identity and in the essence of our society, our original culture is in danger of perishing’.533 He argued that the Netherlands should assert its own core values as the separation of church and state, and the full equality of men and women regardless of sexual orientation. He emphasized the need for newcomers to adopt these core values, framing Islam as a major threat to Dutch tolerance.534 He asked: Where else could a candidate for a huge political movement such as his own be openly homosexual?535 Fortuyn declared Islam a ‘backward culture’,

528 Lechner, The Netherlands: Globalization and National Identity, 14. Also see Philip Muus, "The Netherlands: A Pragmatic Approach to Economic Needs and Humanitarian Considerations," in Controlling Immigration: A Global Perspective, ed. Wayne Cornelius, et al. (Stanford: Stanford University Press, 2004). 529 Lechner, The Netherlands: Globalization and National Identity, 15. 530 Ibid. 531 Ibid., 21. 532 Even though political opponents considered him a racist, he distanced himself from other far right-wing leaders such as Jörg Haider or Jean-Marie Le Pen. 533 Quoted in Lechner, The Netherlands: Globalization and National Identity, 67. 534 Ibid., 75. 535 Ian Buruma, Murder in Amsterdam: The Death of Theo van Gogh and the Limits of Tolerance (New York: The Penguin Press, 2006), 57. 200 called the Netherlands a ‘full country’, and advocated closing the borders to asylum seekers.536 After his success in the local elections of Rotterdam, a city with a particularly high number of immigrants, he ran in the national elections with his own party List Pim Fortuyn (LPF). Only nine days before the national elections, on 2 May 2002, an environmentalist and animal-rights activist named Volkert van der Graaf killed Fortuyn. This provoked an extraordinary outburst of shock, grief and anger in the society, a situation which Balkenende described as ‘un- Dutch’.537 In turn, on 15 May, his LPF party gained 26 seats out of 150 (17%), becoming the second-largest party in the Dutch Parliament.538 In a Dutch TV poll in November 2004, respondents named Pim Fortuyn the greatest figure in Dutch history, above William the Silent, Rembrandt and Erasmus.539 Pollsters interpreted the Dutch result as a search for heroes to confirm shared norms and express pride in local traditions as a response to globalization.540 Fortuyn’s LPF did not continue its successful path, eventually dissolving itself in 2007. Nevertheless, MP Geert Wilders followed Fortuyn’s agenda, established his own party, the Party of Freedom (PVV), and won nine seats in the 2006 general elections. The 2004 killing of the film-maker Theo van Gogh, aggravated instability in the Netherlands. That same year, he made the controversial short movie Submission, on violence against women in Islam. Mohammed Bouyeri, a Dutch citizen, assassinated him on 2 November 2004, in Amsterdam. The letter he attached to van Gogh’s body made it clear that his primary target was Ayaan Hirsi Ali, the former VVD MP and prominent critic of Islam in the Netherlands who had written the screenplay. Dutch intellectuals and public officials argued that the killings exposed the country’s ‘confusion’, its ‘identity crisis’, the ‘fading liberal dream’, a ‘cultural

536 Quoted in Lechner, The Netherlands: Globalization and National Identity, 74. 537 Quoted in Buruma, Murder in Amsterdam: The Death of Theo van Gogh and the Limits of Tolerance, 38. 538 For a detailed discussion on the rise of the LPF, see Eric Belanger and Kees Aarts, "Explaining the Rise of the LPF: Issues, Discontent, and the 2002 Dutch Election," Acta Politica 41 (2006). 539 Buruma, Murder in Amsterdam: The Death of Theo van Gogh and the Limits of Tolerance, 45. 540 Lechner, The Netherlands: Globalization and National Identity, 72. 201 suicide’, and that while the real September 11 events unmasked Dutch ‘multicultural utopia’, these killings were the Netherlands’ own September 11.541 Then leader of the VVD, Jozua van Aartsen, said ‘these people don’t wish to change our society, they want to destroy it’. Moreover, LPF leader Matt Herben said ‘society is being threatened by extremists who spit on our culture; they don’t even speak our language and walk around in funny dresses’. These statements were complemented by violent attacks against mosques, Islamic schools, and churches.542 These sentiments were also reflected in the media, with the peak in December 2002 after Pim Fortuyn’s death.543 It is within this context the first and second Balkenende cabinets’ policies emphasized the importance of fundamental Dutch norms and values. Balkenende openly stated that Dutch tolerance had gone too far, and the Dutch now had to take their culture and values seriously.544 The cabinets transformed minority policy in response to public pressures, aiming to curb the immigration rate and assimilate the Netherlands’ existing non-Western ethnic minority groups.545 This analysis shows that contentious domestic themes of the period evolved around economic downturn, decreasing welfare, and immigration. Eurobarometer data from fall 2004 confirm this: When asked about the two most important problems facing their country, the Dutch selected the economic situation first and crime second, followed by the health care system, pensions, and unemployment.546 These key words can easily come together in an argument about defending the country’s interest against the dangers of globalization. The wall around the welfare state could complement the wall around the country in a coordinated effort to stop unwelcome aliens and welfare claimants.547 As a result,

541 Quoted in Ibid., 78. 542 Buruma, Murder in Amsterdam: The Death of Theo van Gogh and the Limits of Tolerance, 7. 543 Lonneke van Noije, The Democratic Deficit Closer to Home: Agenda-building Relations between Parliament and the Press, and the Impact of European Integration, in the United Kingdom, the Netherlands and France. (The Netherlands 2007), 166. 544 Quoted in Lechner, The Netherlands: Globalization and National Identity, 163. 545 They implemented laws that asked non-Western immigrants to take and pay for an integration course and to pass a test as a condition for permanent residence. Roes, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006, 32. 546 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 547 Lechner, The Netherlands: Globalization and National Identity, 202. 202 although public opinion favored the TCE in fall 2004, the No campaign’s effective framing later managed to attach these worries to the TCE and drive down public support for it.

Contention concerning the EU The Netherlands, both in terms of the broad public opinion and the political elite, have been one of the most enthusiastic followers of the European integration project. Figures 5.3 and 5.4 show the trends in the Eurobarometer surveys in comparison to the EU average:548

Figure 5.3: Dutch support for EU membership (1984-2004)

548 Eurobarometer, "Standard Eurobarometer 61: Public Opinion in the EU, Spring 2004." 203

Figure 5.4: Dutch public’s perceived benefit from EU membership (1984-2004)

Figure 5.5 shows the same Eurobarometer data for a broader period:549

Source: Eurobarometer Figure 5.5: Dutch support for EU membership (1973-2005)

549 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 11. 204

These figures make it clear that the Dutch public opinion towards the EU has been clearly above the EU average. Nonetheless, as in every European country, the Dutch also had certain concerns concerning the European integration project. The Eurobarometer fall 2004 survey asked citizens of every member state about their fears concerning the building of Europe.550 While the set of concerns highlighted by all four countries is similar, the Netherlands has the highest rank among the four – also among the EU25 – in stating the loss of social benefits. The Netherlands also ranks particularly high when it comes to the loss of power for smaller member states and difficulties for Dutch farmers. These concerns parallel the ones in the domestic context especially regarding the welfare state, but combining it with the fear of losing control in the European project as a small member state. A crucial addition to this list is the Euro, as the transition to the common currency became considerably unpopular in the Netherlands. The Eurobarometer special report on EU citizens’ initial reactions to Euro confirms the Dutch negative opinion on the subject.551 The Dutch perception was that the changeover to the Euro was imposed on them, without having any say in the matter.552 The public also thought that it reduced buying power, having a widespread impression that the prices were largely rounded upwards.553 The report singles the Netherlands out as the only member state where revulsion at the changeover to the Euro remains deeply rooted.554 Therefore, the Euro should be considered as one of the EU-related concerns in Dutch society.

550 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 551 ———, "Qualitative Study on EU Citizens and the Euro in the Months Following its Introduction," European Commission: Brussels (2002). For a careful analysis of the origins of this negative opinion see E Engelen, "How to Solve the Riddle of Belated Euro Contestation in the Netherlands?," WRR Scientific Council for Government Policy Web Publications, no. 26 (2007). 552 Eurobarometer, "Qualitative Study on EU Citizens and the Euro in the Months Following its Introduction," 20. 553 Ibid.: 21. 554 Ibid.: 26. 205

Political Party Attitudes towards Europe Among the political elite, the automatic pro-EU attitude has been recently replaced by a discourse of national interest, stressing the need for careful assessment of the costs and benefits of further integration.555 The literature points to two key issues leading to this change: concerns over the size of the net national contribution to the EU budget, and the implications of EU enlargement for the Dutch. Despite being a net recipient of EU funding earlier, the Netherlands has now become a net contributor to the budget since the 1990s. Table 5.1 compares the Dutch benefits and contribution between 2000 and 2005:556

Total benefit from the EU Total contribution to the EU (Euro million) (Euro million) 2000 2,241.2 3,749.1 2001 1,676.4 3,771.3 2002 1,585.5 3,506.3 2003 1,991.6 3,635.9 2004 2,115.8 3,891.8 2005 2,093.5 4,462.7 Table 5.1: The Netherlands’ benefit and contribution in the EU budget (2000-2005)

The budget contribution has been an issue for the Dutch political elite, since the Finance Minister Gerrit Zalm asked for budgetary reform in the 1990s. This ‘served to crystallize a reaffirmed sense of a distinctive Dutch national interest which would have to be more vigorously defended within the Community institutions’.557 VVD parliamentary leader Frits Bolkestein similarly painted a highly pragmatic attitude towards the Union and raised the need for a cultural break in the conception of Dutch foreign policy, allowing the national interest to

555 Robert Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," in Euroscepticism: Party Politics, National Identity and European Integration, ed. Robert Harmsen and Menno Spiering (Amsterdam: Rodopi, 2004). 556 Detailed information can be reached at the EU’s official website on financial programming and budget. Available at: http://ec.Europa.eu/budget/budget_glance/where_from_en.htm 557 Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," 100. For a detailed analysis on the evolution of the budget debate see Pieter de Wilde, "Reasserting the Nation State: The Trajectory of Euroscepticism in the Netherlands 1992-2005," RECON Online Working Paper 1 (2009). 206 occupy its central place.558 The second problematic issue was EU enlargement, which became prominent in both the 2002 and 2003 parliamentary election campaigns. Even if European issues did not have a significant impact on either election results, the subject became politicized.559 The political elite’s concerns on the enlargement were mainly based on the pre-enlargement reform of the EU budget. Who are the main Eurosceptic actors in the Netherlands? Europe does not emerge as a new cleavage dividing the Dutch parties, and it never played a serious role in electoral competition.560 No party in the Netherlands qualifies for hard Euroscepticism advocating withdrawal; however there are soft Eurosceptic parties criticizing certain EU policies and defending national interests: the LPF on the right and the SP on the left.561 Pim Fortuyn did not use Europe as a significant mobilizing theme but his political message carried an articulated Eurosceptic discourse.562 In his book Soulless Europe: Against a Europe of Technocrats, Bureaucrats, Subsidies and Inevitable Fraud, he argued that the EU was serving the interests of the political and bureaucratic elite and not of citizens.563 But Fortuyn’s stance towards Europe was rather populist and nationalist, and not overtly xenophobic as other extreme right-wing political parties in Europe.564 On the left, the Socialist Party is an important actor, with a strong grassroots

558 He later became a European Commissioner and the architect of the contentious Bolkestein Directive. Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," 107. 559 Ibid., 109. 560 Harmen Binnema and Ben Crum, "Resistance to Europe as a Carrier of Mass-Elite Incongruence: The Case of The Netherlands," in Les Résistances à l'Europe: Cultures Nationals, Ideologies et Strategies d'Acteurs, ed. Justine Lacroix and Ramona Coman (Bruxelles: Editions de l'Université de Bruxelles, 2007). 561 Paul Taggart and Aleks Szczerbiak, "Contemporary Euroscepticism in the Party Systems of the European Union Candidate States of Central and Eastern Europe," European Journal of Political Research 43, no. 1 (2004). 562 Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," 116. 563 Given his hard-line stance on immigration issues, he also advocated for the renunciation of the Schengen Accord, among other proposals such as renegotiating the budget, abolition of structural funds, holding referenda on admission of new members. Pim Fortuyn, Zielloos Europa: Tegen een Europa van Technocraten, Bureaucratie, Subsidies en Overmijdelijke Fraude (Utrecht: A.W.Bruna, 1997).; Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," 118. See also Pim Fortuyn, De Puinhopen van Acht Jaar Paars: Een Genadeloze Analyse van de Collectieve Sector en Aanbevelingen voor een Krachtig Herstelprogramma (Rotterdam: Karakter, 2002). 564 This point was also continuously repeated in my personal interviews with the Dutch campaigners. 207 implantation.565 It is active in the anti-globalization movement, and consistently campaigned against the further transfer of power to ‘untransparent, fraudulent, bureaucratic Brussels’.566 The SP claims that the EU is too ‘neoliberal’, where states race to the bottom decreasing public services and favoring business interests. Overall, the loss of social benefits and power at the EU level, the Euro, the budget contribution and enlargement can be identified as politicized and contentious issues concerning Europe. Nevertheless, the recent increase in the critical approach should be understood as nuancing rather than reversing the basic commitment to the integration process.567 This is matched by the Netherlands Bureau for Economic Policy Analysis (CPB) and the Netherlands Institute for Social Research (SCP) report which shows a pattern of decreasing Dutch support for EU enlargement after 2002.568 Based on Eurobarometer data from 2001-2005 however, these reports show that there is no basis for a turnaround in Dutch thinking on Europe, and apart from a fluctuation (10% decrease in support for European unification between the end of 2003 and early 2004, restored back in fall 2004), it is a pro-EU country.569 Therefore these contentious issues should be seen as politicized EU-related issues, rather than symbols of strong Euroscepticism. In the campaign, we see that these issues became salient. The No campaigners successfully used these themes both directly and indirectly as examples of their broader arguments. In other words, they either linked these issues to the TCE directly, or they used them as examples of many problematic steps the EU has taken, presenting the TCE as the latest one in this series. Importantly, Dutch public opinion towards Europe was restored to previous

565 Harmsen, "Euroscepticism in the Netherlands: Stirrings of Dissent," 120. 566 Even though neither LPF nor SP used Europe as a major mobilizing theme, their Eurosceptic stance did not become an impediment to their electoral success, breaking with conventional wisdom. Ibid., 121. 567 Ibid., 101. 568 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 22. 569 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 21. 208 positive figures in the aftermath of the referendum. Post-referendum surveys show Dutch support for European membership as high as 82%.570 Yet, there was a sharp decrease in the Netherlands in support for ‘a’ European constitution between the end of 2004 and spring 2005.571 In the next section I will explain the reasons behind this dip, by exploring the campaign dynamics and showing how the No campaigners linked both domestic and EU-related contention to the TCE.

The Long Campaign In this section I provide an analysis of the Yes and No campaign strategies and framing. The focus will be on the agenda-setting and priming/framing stages, based on personal interviews with the campaigners, campaign materials, television clips, and strategy papers of some parties.572 I will show that the No campaign was very strong, just as in France. They started earlier and placed their campaign themes successfully on the agenda, forcing the Yes campaigners to take the defensive position. They strategically linked the existing contention to the referendum proposal by using effective anti-TCE frames. The Yes campaign was reluctant, failed drastically to respond to this argumentation and suffered from credibility problems. Consequently, the Dutch No campaign successfully countered the early positive opinion on the subject. Below I first provide a brief overview of the campaign and the campaigners. The Dutch referendum was consultative. However the Prime Minister promised in 2003 summer that he would respect the outcome if a referendum was to be held.573 This was the first time the Netherlands held a national referendum

570 Eurobarometer, "The European Constitution: Post-referendum Survey in the Netherlands." Also studies of the CPB and the SCP suggest that the percentage of citizens opposed to further enlargement has stabilized again after the referendum dip, while the percentage of citizens who believe that European integration is developing ‘too quickly’ is also declining. See WRR Scientific Council for Government Policy, Rediscovering Europe in the Netherlands (Amsterdam: Amsterdam University Press, 2007), 61. 571 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 7. 572 A full list of the campaigners I interviewed is presented at the end of this chapter. 573 Arjen Nijeboer, "Peoples’ Vengeances: The Dutch Referendum," European Constitutional Law Review 1 (2005): 399. The GL, the SP, and the D66 also made a promise without any conditions, whereas the PvdA and the CDA did so only conditionally. The PvdA demanded a minimum of 209 in a long time, since 1797, and the referendum required a new bill to serve as a legal basis. The PvdA, the GreenLeft, and the D66 introduced this bill in May 2003.574 The referendum was thus not an initiative of the second Balkenende government, a coalition of the CDA, the VVD and the D66, which was in power since 2003. This reluctance was a key factor in understanding the campaign dynamics, which I will discuss below. The government explicitly mentioned that it was against the referendum but agreed to accept the outcome.575 Just as in the other referendum cases, the government ran an information campaign. The Yes campaigners were the CDA, the PvdA, the GL, the VVD, and the D66. There was no uniform Yes campaign despite an initiative to form one composed of multiple parties. Lousewies van der Laan, then MP of the D66, tried bringing together the Yes forces under the campaign named Foundation for a Better Europe. However, the director of this foundation, Michiel van Hulten, explained how they were frustrated that the parties did not want to come together primarily because they did not want to share the same platform as the unpopular government.576 This tendency, the unwillingness to campaign together with the other mainstream parties, was mentioned in most of my interviews. The lack of a broad platform was accompanied by the lack of a clear common message. Hilde Laffeber, member of the Yes campaign team from the Ministry of Foreign Affairs, mentioned that the lack of coordination resulted in a differentiated Yes campaign, without a uniform message.577 On the No side, the campaigners were the far left SP, the small Orthodox protestant parties CU and the SGP, the far right LPF, and the far right MP Geert Wilders. As with the other cases, there was also a left-wing civil society platform, ConstitutionNo, against the TCE. The No campaign was not united either but the campaign messages shared a core message, as I will show below.

30% of turnout, and the CDA followed, asking also a minimum of 60% against as an additional condition. 574 Ibid.: 396. This bill, envisioning the referendum, came into force finally in January 2005. 575 Ibid. 576 Personal interview, 30 January 2009. 577 Personal interview, 16 December 2008. 210

Agenda-setting Studies find that if the campaigners start early, they have an advantage to set the tone of the debate. The No side started their campaign earlier than the Yes side in the Netherlands. This not only gave them the upper hand in agenda-setting but also it forced the Yes campaigners to take the defensive position and defend the TCE on the issues specified by the No campaigners. My interviews with the Yes campaigners show that indeed, one of the major difficulties they faced was the early start of the No campaign. Delphine Pronk, the head of the EU communications unit at the Ministry of Foreign Affairs, stated: ‘We started later and we had to explain why they should not vote No, and not why they should vote Yes, so we were in a double No-campaign instead’.578 Gerben Jan Gerbrandy, then secretary of the D66 parliamentary group, noted the same problem: ‘It was the No campaign that was setting the agenda, we had to explain people why they should not vote against it, we were running behind the No campaign, instead of the other way around’.579 These remarks confirm that the Yes campaign started with a handicap because of their timing. Despite the earlier parliamentary meetings, the pro political parties started their real campaign only in May. Anker argues that the No camp benefited from a quick start also because the slow start of the Yes campaigners left the impression that the Yes camp took victory for granted and refused to reach out to the voters.580 It was only in the last few days that government ministers who had been accused of launching a late, lukewarm campaign began distributing tracts in the streets, going to schools and debating in various places with the population.581 The reasons behind this late start are various. First, as in the other cases, the positive public opinion towards the TCE in early polls led to optimistic

578 Personal interview, 19 January 2009. 579 Personal interview, 6 November 2008. 580 Hans Anker, "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment " in Election Time 2005: The European Yearbook of Political Campaigning 2005, ed. Sepp Hartinger (European Association of Political Consultants and Hartinger Consulting Communications, 2006). 581 Corinne Deloy, "The Dutch Reject the European Constitution," Robert Schuman Foundation Report (2005). See also Robert Harmsen, "The Dutch Referendum on the Ratification of the European Constitutional Treaty," European Parties Elections and Referendums Network, Referendum Briefing Paper 13 (2005). 211 expectations and decreased the urgency to campaign for a Yes. However there was a second reason in the Dutch case. There was a responsibility crisis concerning who would run the campaign for the Yes side. Crum argues that the ministers lacked enthusiasm to campaign, did not attach sufficient attention to the issue and disowned the referendum as a concern for parliament since it was the parliament who had initiated it.582 On the other hand, the Yes camp opposition parties were inclined to leave the initiative to the government, thinking that they had little to gain by taking the lead. This responsibility problem on the side of the Yes campaigners was also evident in my interviews. These factors altogether resulted in a vacuum, which was successfully filled in by the No campaigners.

Priming and Framing As I have argued before, there are three factors that matter in the priming and framing stage: strength of the campaign frames, credibility of the speakers, and mobilizational structures. The No campaigners clearly had more success in priming and framing. The media content analysis in the next section will demonstrate this in detail. Their early start and strong frames enabled them to link the existing contentious issues to the TCE in the public’s mind. The Yes campaign was not successful due to weak frames and weak credibility.

Content of the campaign frames Based on the strength definition of the framing literature, the No campaign frames were stronger. I will explain the content of the Yes and No campaign messages followed by a discussion of their strength. Below I list the Yes and No campaign frames. I identified the campaign frames primarily based on the written campaign materials.583 Next, I categorized them according to their subjects such as peace, economy, democracy, institutions, and other salient issues like welfare state, immigration, accession of Turkey, or national identity.

582 Ben Crum, "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?," WRR Scientific Council for Government Policy Web Publications, no. 25 (2007): 19. 583 Where available, one to two page leaflets listing the campaign arguments provided the best source. If not, I used interview data and secondary sources to complement the available materials. 212

The main Yes campaign frames can be categorized as: 1. Approving the TCE is approving Europe. European integration has guaranteed peace in Europe in the last 60 years. Furthermore, the TCE consolidates European values such as democracy, gender equality and respect for human rights. 2. The TCE would institutionally enhance the integration project, and make it more democratic, efficient, and powerful in the world vis-à-vis the US and China. It would simplify former treaties, clarify the competences of the Union and the member states based on the principle of subsidiarity, make an enlarged EU more effective in decision making, give both the European Parliament and the national parliaments more power, give the Union better representation and visibility in the international arena with the new European Minister of Foreign Affairs. 3. The TCE would help in formulating common environmental and immigration policies, and a common fight against terrorism, drug trafficking, illegal immigration. 4. The TCE would place social rights at the European level, within the Charter of Fundamental Rights. Moreover, it would introduce the right of initiative for citizens; if they collect one million signatures in a significant number of member states, they may ask the Commission to submit a proposal on a particular subject. 5. The TCE would lead to a more prosperous Netherlands. The Netherlands benefited vastly from exporting to the EU members in the past. 6. The TCE would not replace the Dutch constitution. There is no European ‘super-state’. Among the Yes campaign, none of the political parties wanted to be the main campaigner because of the responsibility problem I explained above. The government information campaign also refrained from running a strong Yes campaign to remain neutral. All the Yes campaigners used the first four Yes

213 frames. While variations of the fifth and sixth Yes frames were used by most campaigners, they were made explicit only in the CDA campaign materials. The CDA’s slogan was ‘Europa verdient ja’ (Europe deserves a Yes). The GreenLeft campaigned with the slogan ‘Verbeter Europa begin met een ja’ (A better Europe starts with a Yes). The PvdA used ‘Nederland sterk en sociaal in Europa’ (The Netherlands strong and social in Europe). Similarly, the slogan of the Foundation for a Better Europe – the D66 initiative for a united campaign – was ‘Een sterker Nederland in een beter Europa’ (Stronger Netherlands in a better Europe). On the other hand, the No campaign frames can be categorized as: 1. The TCE would cause loss of veto rights and thereby sovereignty in a European ‘super-state’. It would establish a European federal state with a single constitution, where the Netherlands would become a powerless province. Veto rights are especially important for small countries. A decrease in veto rights means giving up power to block unfavorable decisions, in a bureaucratic and undemocratic Brussels. Furthermore the decision to adopt the TCE, similar to the decision to switch to Euro, is a symbol of a ‘super-state’ and one of many problematic and costly steps taken by the EU. 2. The TCE is not social enough. The TCE contains anti-social and market-friendly economic policies, and thereby would make it mandatory for all member states to base their regulation of public services such as healthcare and education on free market rules. Also, connecting this frame to the first one: Brussels would have more control over the Dutch social policy. 3. The TCE is not democratic enough. Neither the process nor the result was democratic. 4. The TCE is increasing defense spending and thereby militarizing Europe and creating a very expensive European army. 5. The TCE would further the EU enlargement process, advancing the accession of Turkey (an Islamic country) and increasing immigration in the Netherlands. This would pose a threat to Dutch society and

214

workers. Also, the Netherlands would pay even more to the EU budget to subsidize poorer new members in an enlarged Union. Furthermore, Turkish accession was also connected to the first frame: As a large country, Turkey would lessen the power of small countries in the Union. The main No campaigner was the SP from the far left. It campaigned with the first four frames but built its core campaign message around the first frame. As such, the first frame connected the SP campaign framing to one another. It used the slogan ‘Weet waar je ja tegen zegt’ (Know what you are going to say Yes to). Running a separate campaign, the left-wing civil society group ConstitutionNo advocated for the same agenda as the SP except for the ‘super-state’ argument. On the far right, the LPF and Geert Wilders combined the first frame with the last frame. With the slogan ‘Nederland moet blijven’ (The Netherlands must remain/withstand), Wilders signaled that with the TCE the Dutch would no longer be in control of their country. Following Pim Fortuyn’s previous line of framing, he also stated ‘we have been too tolerant with intolerance’, referring directly to the political events that have shaken the Netherlands over the last three years.584 Finally, the small Orthodox protestant parties the CU and the SGP, despite having different reasons, also adopted the first ‘super-state’ frame, criticizing the word ‘constitution’. They had other reservations on the issue of reference to God in the preamble, and also on the meaning of Turkish accession for the borders of Europe but these were not fully developed campaign frames. They used variations of the slogan ‘Europe ok, this constitution No’.585

Relative strength of the campaign frames As was pointed out earlier, research on framing effects shows that frames are stronger if they can invoke available, accessible and applicable concerns. If the frames are concrete, vivid, image-provoking, emotionally-compelling, and contain negative information they tend to be more effective.

584 Quoted in Deloy, "The Dutch Reject the European Constitution." 585 Personal interview with Esme Wiegman, 28 October 2008. 215

The first two Yes frames, just as in the other cases, were weak as they contained abstract and technical content. The first frame stressing the importance of the Union in maintaining peace did not involve an applicable concern in the absence of an imminent threat to European peace. The second frame presenting the TCE as an institutional step towards a better Europe did not involve available or accessible concerns either, given the voters’ lack of familiarity with the technical content of the text. As such the first two Yes frames were neither emotionally compelling nor based on contentious issues. The third Yes frame was somewhat stronger as it highlighted the need for Europe to tackle common problems as climate change and organized crime. While these issues were available concerns to the public, they were not presented as solutions to immediate problems within Dutch society. The fourth, fifth and sixth Yes frames were based on recent social contention. The fourth Yes frame emphasized the social progress brought by the TCE and therefore related to the welfare state related concerns. The fifth one pointed to the economic benefits gained through Europe, which was indirectly connected to the contentious national contribution to the EU budget. The last one was used directly to counter the No campaign’s core argument that a Yes vote would leave the Netherlands powerless in a European ‘super-state’. While the last three frames invoked available, accessible and applicable concerns in the society, they were not presented in a way that carried any emotional content. They were developed to counter the No campaign frames on the subject, which were considerably stronger, as will be seen below. Moreover, as a key difference with the other cases, the Dutch Yes campaigners decided to acknowledge voters’ doubts. Therefore, these Yes campaign frames were presented in a hesitant manner, as I will discuss in detail below. For instance, the GreenLeft slogan, ‘A better Europe starts with a Yes’, purposefully emphasized that this was only a beginning.586 The GL campaign material openly stated that they were not fully satisfied with the Constitution but that it was a good start to improving European integration. Even when they did

586 Personal interview with Bas Eickhout, 5 November 2008. 216 not explicitly acknowledge the doubts of voters, the Yes campaign frames signaled a message of compromise. Balkenende mentioned that the Netherlands might sacrifice some voting power under the new arrangements, but that they would make the Union more effective while still constraining the power of big countries.587 Also, the leader of the PvdA, Wouter Bos, said that the TCE was a step toward a more social Europe. The head of the Democratic Europe Association similarly argued that on balance it would help to tame the market by emphasizing positive cooperation over the blind forces of competition.588 These remarks clearly included compromise messages. Thus the responsibility crisis in the leadership of the Yes campaign coupled with the unenthusiastic and balanced nature of the Yes messages led to weak frames despite having some strong components in them. The No campaign did not have such problems. According to the relevant criteria from the framing literature, all No campaign frames were strong except for the third one. The third No frame, blaming the TCE for not being democratic, was abstract and non-controversial for the Dutch society. The rest of the No frames clicked well with the recent contention with the society. The first No frame was the strongest since it was the key theme bringing the rest of the frames together. The ‘super-state’ frame, complemented with issues such as decreasing welfare, Turkish accession, increasing immigration, and controversial EU policies such as the adoption of the Euro and budget contribution, effectively tapped into the existing contention. A former economics minister from the LPF, Heinsbroek, said that the Netherlands already had given up too much sovereignty, and that sacrificing more was a step too far.589 The editor of the most influential conservative magazine claimed that the Dutch welfare system was a vital interest, potentially at risk if the Netherlands could not control access, and recommended pulling the emergency break at the referendum.590 The Yes campaign sought to counter these frames as I have shown

587 Quoted in Lechner, The Netherlands: Globalization and National Identity, 262. 588 Quoted in Ibid. 589 Quoted in Ibid., 265. 590 Quoted in Ibid. 217 above, yet the No campaign frames were more effective because they carried negative information, concretely built on the recent problems, and invoked existing fears of globalization concerning the loss of social standards and Dutch values. While the Yes campaigners were decidedly acknowledging the doubts of the voters, the SP was using a counter strategy with its slogan. Renske Leijten, then leader of the SP youth organization, explained that the SP slogan, ‘Know what you are going to say Yes to’, was aimed at creating doubt among the voters, instead of putting forward a clear negative agenda.591 Their professional communication organization intended to make the voters doubt what this long and technical document was really about and thereby persuade them to say No. The combination of the cautious Yes framing with the sceptical doubt-boosting No framing resulted in an even stronger No campaign. The campaign materials in Figures 5.6-5.14 provide an interesting demonstration of these patterns. The Yes campaign was not very successful with its materials. The government sent every household a newspaper six to eight weeks before the referendum, with a nice picture on the cover, showing a happy family on the beach. However, behind the cover there was only the exact text of the TCE without any explanation. One week before the referendum, the government sent every household a short leaflet, which explained what was really at stake. This text was approved by the Referendum Commission but it caused debate due to a factual problem. It mentioned that by saying Yes the Netherlands would accept the superiority of EU laws over national laws. This legal doctrine, known as the supremacy of EU law, was actually not introduced by the TCE and has been long established. Atzo Nicolaï, then minister of EU Affairs, mentioned this as one of the mistakes, an unfortunate statement, as people did not know that the EU laws were already superior to national laws.592 According to CDA MP Jan Jacob van Dijk also, this reinforced the feeling among Dutch citizens that they

591 Personal interview, 11 April 2008. 592 Personal interview, 29 October 2008. 218 would lose control over their own legislation.593 The government’s campaign material was also criticized in the literature. Lechner portrays the situation as: ‘In its cheerful, nationally distributed leaflet, complete with pictures of typical European scenes like a sleeping baby and a boy running on a beach, the government laid out its case; the Constitution would help to preserve the peace, prosperity, and security of the past sixty years’.594 The Yes campaign posters were not very striking either, mostly portraying the abstract idea of Europe. Only the GreenLeft poster, shown in Figure 5.7, sought to be more concrete by demonstrating air pollution in Europe. In contrast, the SP material presented their messages in a clear and strategic manner. An SP campaign booklet of 20 pages specified 10 reasons to vote against the TCE, each reason explained and illustrated by excerpts from the TCE itself. Their main poster, shown in Figure 5.11, included a map of Europe without the Netherlands, emphasizing the loss of sovereignty/‘super-state’ frame. Almost all of my interviewees recalled this map and brought it up during the interviews, as it was considered very striking. The top-left corner showed the slogan, ‘Know what you are going to say Yes to’, playing on the doubts. Moreover most of their campaign materials, such as the one in Figure 5.11, started with the phrase ‘Euro-Referendum’. This, SP MP van Bommel stated, was also a strategic choice of wording.595 The term ‘Euro’ in the poster was actually referring to a ‘European’ referendum, however it was used strategically to build on the dissatisfaction with the currency ‘Euro’ in the society. A leaflet, shown in Figure 5.12, presented a similar picture and was used both by the ATTAC Netherlands and the SP. The ConstitutionNo group used a variety of posters such as the one in Figure 5.13, which protested against animal cruelty in Europe, showing Spanish bull-fighting.

593 Personal interview, 29 October 2008. 594 Lechner, The Netherlands: Globalization and National Identity, 261. 595 Personal interviews; first 8 April 2008, second 31 October 2008. 219

Figure 5.6: Documents used by Figure 5.7: The GL’s poster the government

Figure 5.8: Poster of the Foundation for a Better Europe

Figure 5.9: The CDA’s poster Figure 5.10: The PvdA’s poster

220

Figure 5.12: The ATTAC’s poster Figure 5.11: The SP’s poster

Figure 5.13: The ConstitutionNo’s poster Figure 5.14: Posters of the SP and the PvdA

221

In general, both the Yes and No campaigners agreed with this campaign assessment. Hilde Laffeber, Ministry of Foreign Affairs Yes campaign team member, explained that their strategy was to emphasize that it was normal to have doubts.596 Nevertheless, she stated that this turned out to be a mixed message, and played into the hands of the No camp. Similarly, the GreenLeft also decidedly campaigned in an unenthusiastic way, as there was a 50-50% split within the party regarding their position towards the TCE. GL campaigner Bas Eickhout pointed out that the decision of the GL to be evenhanded in their campaign was reflected in their campaign material as well.597 As I mentioned already, their slogan was ‘A better Europe starts with a Yes’, emphasizing that this was a beginning. He explained how their campaign material presented balanced thinking, while the SP map bluntly showed that the Netherlands would disappear: ‘Then you are asked to fight an uphill battle. They say it is a disaster, we say it is not that bad. It was very difficult’. Overall, confronted with a No campaign that actively tried to link doubts to a No vote, this strategy did not succeed for the Yes camp. The messages that played on ‘when in doubt, vote No’ resonated strongly among the undecided voters.598 The Yes campaigners also acknowledged that they sounded overly technical and broad, unable to specify key issues in the way the No campaigners did. Jan Jacob van Dijk, CDA MP, admitted that the Yes campaign spent too much time explaining what was in the treaty, and said less about more political points of view that specified the reasons why people needed to be in favor.599 Van Hulten, director of the Foundation for a Better Europe, has also explicitly mentioned this as a problem. Referring to his generation (39 years old), he argued: ‘We have always lived in peace and stability. We need to find a new argument for Europe’.600

596 Personal interview, 16 December 2008. 597 Personal interview, 5 November 2008. 598 Anker, "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment ". 599 Personal interview, 29 October 2008. 600 Personal interview, 30 January 2009. 222

Furthermore, there are numerous references to the ineffective argumentation of the Yes side in the literature on the Dutch campaign. Crum emphasized the Yes campaign’s inability to communicate a clear, coherent and convincing message on the merits of the TCE.601 Nijeboer also criticized the Yes arguments as not very well considered and technocratic; especially facing a No camp involving the SP, who began much earlier, spent a lot of energy on their key message and slogans, and were passionate in their reasoning.602 Anker also highlighted that in January 2005, while the No campaign was strongly arguing that the TCE was a step towards the European ‘super-state’, the Yes campaign did not have a unifying message, was on the defensive and lacked the ability to control its own message.603 He described the European ‘super-state’/loss of sovereignty frame as the ‘home base’ for the No camp, a button they cleverly and relentlessly pushed. Additionally three themes, the Euro, enlargement, and Turkey eclipsed the debate.604 The Yes campaign, in turn, had to redefine the political battlefield altogether to win. In contrast with the Yes camp, the No campaigners were strategic in their choice of campaign frames. Hans van Heijningen, the secretary general of the SP, mentioned that referendum required a different type of strategy than regular elections, referring to their realization that radical left messages would not carry them to 51% in the Netherlands.605 He said that they brought the ‘super-state’ argument in strategically, as the other arguments had clear left-wing political content. Responding to their opponents who blamed the SP for being populist, he suggested that they combined socialist messages with people’s concerns. He added: ‘If populism means communicating with the masses, you can call us populist’. Harry van Bommel, MP of the SP, also acknowledged that the first No

601 Crum, "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?," 20. 602 Nijeboer, "Peoples’ Vengeances: The Dutch Referendum," 400. 603 Anker, "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment ". 604 Ibid. 605 Personal interview, 11 April 2008. 223

frame (to a greater extent) and the second No frame worked very well, as people could relate to them very much.606 The No campaign used various concrete examples to demonstrate these arguments, in order to link these statements specifically to people’s daily lives. For instance, an example concerned window cleaners who use ladders to reach windows. The No campaign criticized an EU regulation, which sought to replace ladders with electronic elevators to prevent accidents. The SP framed this as a regulation favoring big companies, claiming that an independent window cleaner could not afford the electronic machinery and would be out of business.607 Similarly, there were other examples concerning the privatization of the public sector in the housing, transportation and health domains. The Bolkestein Directive was another simple example, yet it was not as contentious as it was in France.608 The Euro was frequently used as a concrete example of problematic decisions taken by the Union. Jan Marijnissen, then leader of the SP, stated that everyone knew that there had been a price increase with the switch to the Euro, and that the people who drafted the TCE were also responsible for the Euro.609 Moreover, distinctive features of the Dutch national identity – gay marriage, soft drugs policy, euthanasia – were presented as under threat by European intervention.610 On the right-wing, Mat Herben, then chairman of the LPF, used federalist remarks made by German and Spanish leaders before their constituencies, stating that the TCE was the birth certificate of the United States of Europe, to prove that a European ‘super-state’ was being established.611 On the other side, almost all of the Yes campaigners accepted that they had tremendous difficulty in linking the broad abstract institutional content of the treaty to concrete daily issues. To counter the No campaign, some Yes

606 Personal interviews; first 8 April 2008, second 31 October 2008. 607 Personal interview, 11 April 2008. 608 For a detailed discussion of the evolution of the political debate on Bolkestein Directive see J Pelkmans and S van Kessel, "Encapsulating Services in the 'Polder': The Bolkestein Directive in Dutch Politics," WRR Scientific Council for Government Policy Web Publications, no. 31 (2007). 609 Quoted in Aarts and van der Kolk, eds., Nederlanders en Europa: Het Referendum over de Europese Grondwet. 610 Harmsen, "The Dutch Referendum on the Ratification of the European Constitutional Treaty." 611 Personal interview, 7 November 2008. 224

campaigners – key figures – resorted to drastic statements warning the public of a potential disaster if they voted against the text. These speeches were recalled and brought up in almost all interviews. Some are:612 CDA Justice Minister Piet-Hein Donner warned against the possibility of war and a possible Balkanization for Europe if the No would win the referendum. VVD-MEP Jules Maaten canceled a TV spot showing images of Auschwitz, the genocide in Srebrenica and the terrorist bombings in Madrid, which aimed to persuade voters to vote in favor. The video was released and withdrawn within a matter of hours. D66 Economic Affairs Minister Laurens-Jan Brinkhorst remarked that if the Dutch people were to reject the constitution, the ‘lights would go off in the Netherlands’, ‘our country would be locked up’ and ‘the Netherlands would become the Switzerland of Europe’. CDA Minister of Foreign Affairs Ben Bot recommended voters to abstain in case of doubt, while former Prime Minister Ruud Lubbers (CDA) repeated this position one day before the vote. Executive Director of the Dutch Central Bank Henk Brouwer admitted that, during the creation of the Euro, the Dutch guilder had been undervalued by five to ten percent vis-à-vis the German mark. The last remark essentially confirmed the concerns of the public on the Euro, as there was already dissatisfaction on the subject that the prices went up. This statement conflicted with the official state discourse used in the last three years. The other comments on the other hand, gave the impression that the government was not interested in the real opinion of the voters and that the consequences would be disastrous. The public saw an elite that lost control over politics and resorted to threats.613 Wim van de Donk, the chairman of the

612 Anker, "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment ". Also, Nijeboer, "Peoples’ Vengeances: The Dutch Referendum."; Harmsen, "The Dutch Referendum on the Ratification of the European Constitutional Treaty." 613 Anker, "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment ". 225

Scientific Council for Government Policy (WRR), also characterized the Yes campaign’s strategy as clumsy and not-convincing, as the political elite were not able to reinvent or develop a new narrative for Europe.614 These scare tactics backfired, in line with the framing literature which suggests that competitive environments place a premium on strong frames, and that weak frames can backfire.615 As a result, both planned and unplanned moves by the Yes campaigners helped the No campaign because either they were not forceful enough or they backfired. When asked about the meaning of the initial high support for the TCE, van Hulten, director of the Foundation for a Better Europe, said: ‘During these early polls nobody knew anything. People were not thinking yet. The No camp started raising issues, they tapped into the latent Eurosceptic sentiments. I am saying latent because they had to be activated.’ CDA MP Jan Jacob van Dijk has also mentioned that the No camp effectively took aspects of the treaty in which there were some negative feelings or sensibility in the Dutch society.616 Hilde Laffeber, member of the Ministry of Foreign Affairs (MFA) Yes campaign team, similarly acknowledged that the No campaigners had better arguments, tapping into the general sense of discontent in the society, and that the Yes camp could not give a message of change.617 On the No side, SP MP van Bommel mentioned that the Yes campaigners advocated for the TCE by referring to the past, to the benefits Europe had brought to the Netherlands, and not by referring to the future.618 He explained that, in a discussion about the future of Europe, while the SP was concrete about what it would mean to change Europe, the Yes campaigners only referred to changing institutions in Europe, which was not important for people.

614 Personal interview, 23 October 2008. 615 LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro," 191. 616 Personal interview, 29 October 2008. 617 Personal interview, 16 December 2008. 618 Personal interviews; first 8 April 2008, second 31 October 2008. 226

Overall, the No campaign frames were stronger than the Yes campaign frames. Below I show how the credibility of the speakers and mobilizational structures further strengthened the No campaign frames.

Credibility of the Speakers Credible sources are proven to strengthen campaign frames. As I argued previously, in referendum campaigns, the impact of credibility can be assessed by looking into three components: Government popularity, scapegoating the EU, and disagreements within the parties. Starting with government popularity, polls showed that the second Balkenende government was far from being popular. A poll that was published in mid-May 2005 revealed that only 19% was satisfied with the government, which had started its term with a promise of re-establishing contact with citizens.619 This dissatisfaction can be seen as a factor decreasing the strength of the Yes frames because they were not seen as credible. The No campaigns can easily play to this sentiment among the public. Second, the habit of using the EU as a scapegoat for unpopular measures was another problem for the Yes campaign’s credibility. A great majority of the campaigners mentioned this as a problem. Atzo Nicolaï, then minister of EU Affairs, said that the main problem was starting the campaign 20 years too late: ‘What we did for 20 years was actually negative, to blame Brussels, as a general line, if something did not work out nicely…This is the negative side, and the positive side is a general feeling after the war, that Europe is good. In between there was not so much.’620 Thus it was not easy to switch from this negative discourse to a positive one. Similarly, the Scientific Council for Government Policy WRR’s report on the referendum suggests that the European integration project had been largely invisible to Dutch citizens before the referendum, apart from sensitive issues such as enlargement, the Bolkestein Directive, the introduction of the Euro, and the

619 Deloy, "The Dutch Reject the European Constitution." 620 Personal interview, 29 October 2008. 227

negotiations with Turkey, which all were placed within a system where all positive news was presented as national successes, and negative news as coming from Brussels.621 Therefore Dutch politicians were not used to presenting persuasive framing on Europe, which also met the concerns of the citizens. The report explicitly states that ‘the governing parties and most of the Cabinet members not only voiced their positions far too late in the debate, but also had no idea of how to talk about Europe in the media’.622 Third, disagreements within the campaigning parties decreased the credibility of the sources by sending conflicting signals to the voters. An interesting similarity between the French and Dutch cases (and the Luxembourgish case to a smaller extent) is the hesitation of the mainstream left- wing parties – social democrats or green parties – towards the TCE, due to inner discussions. The PvdA and the GL in the Netherlands had internal debates. Even if these parties did not have their factions joining the No campaign as in France, these discussions led to a strategic hesitation and an unenthusiastic campaign. Not so surprisingly, these parties have particularly low levels of persuasion in terms of their voters’ behavior in the referendum. As I will show below, a large percentage of their voters voted negatively, just as was the case in France. I have already mentioned the GL’s decision to make their hesitation clear in the campaign as a strategy. The PvdA had a similar problem. The international secretary of the PvdA, Marije Laffeber, said that campaigning with the government would have been inconsistent as the PvdA had campaigned against the government in the EP elections in 2004, and against the welfare reforms in October 2004.623 Acknowledging the internal debate within the party, Marije Laffeber explained that the party leadership was nevertheless involved in the negotiations for this treaty and considered that this was a good step forward, as opposed to a final stage. However, she added that her party decided not to invest too much time or resources in this campaign, as it was considered to be the government’s responsibility. These divided stances were highly visible in the Yes campaign.

621 WRR Scientific Council for Government Policy, Rediscovering Europe in the Netherlands, 27. 622 Ibid., 84. 623 Personal interview, 22 October 2008. 228

Mobilizational Structures The last factor that influences campaign success is mobilizational structures. I divide this concept into financial resources and civil society contribution. Regarding the financial resources, the independent Referendum Commission allocated 400,000 Euros to each of the Yes and No campaigners, while giving 200,000 Euros to the neutral organizations.624 Political parties could apply to the Referendum Commission for funding. Most parties had their own sources too but all except the LPF and GL applied and received up to 40,000 Euros from this Commission.625 However, seeing the success of No campaign in the polls, the Yes campaign decided to spend an additional 3.5 million Euros in mid-May,626spending 6 million Euros total in the end.627 The No camp parties protested against this extension. The active campaigners were mainly the government and the political parties. The Yes campaign was done mainly by the government whereas the No campaign was mainly carried out by political parties.628 Paralleling the other cases, the Yes camp had better resources. Yet due to hesitation and responsibility issues, the Yes side did not make a major effort with important resources. Among the No campaigners, it was mainly the SP which devoted considerable amounts of its own resources to the campaign and organized local activities using its grassroots implantation. Thus, apart from the SP campaign, the other No campaigns remained limited and low profile. Second, concerning the contribution of civil society, the Dutch mobilization was not comparable to that of France. These organizations did not play a significant role in getting citizens involved in the Dutch referendum campaign.629 While civil society supported the TCE widely, including all the business organizations and the trade unions, none of them launched a substantial campaign. Therefore the civil society engagement was limited, and had low

624 Nijeboer, "Peoples’ Vengeances: The Dutch Referendum." 625 Arjen Nijeboer, "The First Dutch Referendum: A Pre-Ballot Assessment," Notre Europe, no. 18 (2005): 8. 626 Nijeboer, "Peoples’ Vengeances: The Dutch Referendum." 627 Personal interview with Hilde Laffeber, 16 December 2008. 628 Personal interview with Hilde Laffeber, 16 December 2008. 629 WRR Scientific Council for Government Policy, Rediscovering Europe in the Netherlands, 66. 229

political visibility.630 Van Hulten, director of the Foundation for a Better Europe, mentioned that although they coordinated with the trade unions and employer associations, this was ‘all lackluster’.631 For instance the Federation Dutch Labor Movement (FNV), the main trade union, declared that the TCE was a step forward, but did not propose a vote instruction for its members.632 The No campaign lacked support from the broad civil society and the trade unions. Nonetheless, the SP has strong grassroots implantation, is known for its extra-parliamentary activities, and thus adapted quickly to this type of campaign.633 The SP took the initiative to form the ConstitutionNo group, the left- wing civil society campaign against the TCE. Apart from this initiative however, this group was independent from the SP throughout the campaign. It was formed by a small number of people and had a small budget. The ATTAC Netherlands is a very small, loose network of left-wing activists. Just as in Spain, members of the ATTAC Netherlands acknowledged their limited mobilization capacity compared to that of the ATTAC France, as they lacked its local organizational structure.634 Therefore, at the local level too, it was mainly the SP who backed up the No campaign institutionally with its local network and financial resources. On the right, the Young Fortuynists, youth organization of the LPF had a minor campaign but it lacked resources and was not visible. While the Yes campaigners had access to better resources and had wider civil society support, they did not use their mobilizational structures effectively because of their responsibility crisis and strategic hesitation. In contrast, the SP resolutely used its local implantation and resources to support the No campaign. As a result of the lively debate, the interest in the Dutch campaign increased over

630 Crum, "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?." 631 Personal interview, 30 January 2009. 632 Crum, "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?," 19. 633 Ibid.: 20. 634 Personal interviews with Willem Boss, 8 April 2008; and with Erik Wesselius, 4 November 2008. 230 time. While 32% said they discussed the topic with family, friends or colleagues in April, this number rose to 68% by the end of the campaign.635 In sum, the campaign analysis demonstrates that the No campaign successfully countered the initial positive public opinion on the subject. The No campaigners not only started earlier and defined the terms of the debate but also strategically framed their No arguments around people’s concerns. The Yes campaigners had to take the defensive position and had tremendous difficulty in countering these criticisms with effective Yes frames. Their frames were hesitant, involved messages of compromise, and suffered from weak credibility. While they had better access to resources, they did not use these in a dedicated campaign. The No campaign, in turn, was supported decisively by the SP financial and institutional structure. Figures 5.15 and 5.16 present cartoons portraying the inadequacy of the Yes campaign.636 Figure 5.15 depicts the problematic Yes campaign material since the government distributed the full text to every household. Figure 5.16 specifically deals with the campaign frames; as opposed to other European leaders appealing to their public with statements as ‘I have a dream’, ‘I have a plan’, ‘I have a vision’, the Dutch leaders are presented as only saying ‘I have a folder’.

Figure 5.15: Cartoon depicting the Yes campaign material in the Dutch 2005 TCE referendum

635 Aarts and van der Kolk, eds., Nederlanders en Europa: Het Referendum over de Europese Grondwet. The figures represent the valid percentages. 636 Published in Dutch newspapers Trouw and Algemeen Dagblad in May 2005, taken from Ibid. 231

Figure 5.16: Cartoon depicting the Yes campaign style in the Dutch 2005 TCE referendum

Analysis of the Vote Below I analyze the referendum vote based on media content analyses and public opinion data, to observe the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. The assessment in this section confirms the preceding campaign analysis; the No campaign framing was indeed stronger than the Yes campaign framing.

Shifts over Time As Figure 5.1 showed at the outset, public opinion was in favor of the TCE in fall 2004 but No vote intentions increased steadily until mid-May, followed by fluctuations for the rest of the campaign.637 Mid-May onwards was the most intense phase of the campaign as the Yes campaigners actively campaigned only during this period. The No campaign, particularly the SP, had been working since December 2004. The polls still showed a lead for the Yes side in March 2005, but as the debate unfolded, the No camp took the lead April onwards.638

637 For a detailed analysis of these fluctuations, see Kleinnijenhuis, Takens, and van Atteveldt, "Toen Europa de Dagbladen Ging Vullen." 638 Nijeboer, "Peoples’ Vengeances: The Dutch Referendum," 401. 232

Media Content Analysis There have been several in-depth studies of the role of the news media in the Dutch referendum campaign, by Andreas Schuck, Claes de Vreese, Jan Kleinnijenhuis, Janet Takens and Wouter van Atteveldt.639 Below I present findings of these studies on the time devoted to both camps and the topical coverage of the campaign. The referendum topic was highly visible in the news with a positive tone towards the Constitution. The Yes campaigners received more coverage than the No campaigners, however they were evaluated as performing poorly and more negatively than the No campaigners.640 Takens suggests that although the tone of the news about the No campaigners was slightly more negative, this negative exposure was overshadowed by the enormous amount of negative attention for the Yes campaigners.641 Therefore, while the Dutch Yes campaign received more coverage in the Dutch media, paralleling the other TCE referendum campaigns, their coverage was significantly more negative relative to the other cases. This shows that the Yes campaign was indeed perceived poorly, especially due to the drastic statements I mentioned above. The news media devoted a lot of attention to the topic in the final weeks leading up to the referendum, reaching a level that European issues had never seen before. The media content analyses track the main events that captured attention. The Bolkestein Directive, which was a major theme in the French campaign, became also visible in the Dutch case in March and April 2005. It symbolized the ‘Polish plumber’ and the influx of cheap labor from the new EU members.642 Nonetheless, the first phase of the campaign was mainly on what to expect if the referendum failed.643 Schuck argues that this pattern reached a

639 See Schuck and de Vreese, "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign."; and Kleinnijenhuis, Takens, and van Atteveldt, "Toen Europa de Dagbladen Ging Vullen." 640 Schuck, "Referendum Campaign Dynamics: News Media, Campaign Effects and Direct Democracy", 14. 641 Takens, "A Negative Campaign? A Study of the News Coverage of the European Constitution and its Effects". 642 Kleinnijenhuis, Takens, and van Atteveldt, "Toen Europa de Dagbladen Ging Vullen." 643 Schuck, "Referendum Campaign Dynamics: News Media, Campaign Effects and Direct Democracy", 12. 233

climax with the contentious remark of Justice Minister Donner on 18 April, suggesting a war could have been a possible consequence.644 Similarly, the comment of the Executive Director of the Dutch Central Bank Henk Brouwer on 30 April received attention. Geert Wilders started actively campaigning against Turkish integration and was the center of public attention April onwards. This was happening at the same time that the Yes campaign suffered a drawback where the VVD was about to broadcast a TV spot linking the failure of referendum to the Holocaust, receiving wide criticism from the public.645 In the last few weeks, the referendum became highly visible in the media with many TV and public debates. These studies also explicitly confirm that the No campaigners started early and the Yes campaigners came in very late. Looking at the topical focus of the news coverage, the majority of reporting focused on either the campaign itself, the decision-making process of voters, or dealt with the possible consequences of an approval or a rejection.646 Takens specifies the topical distribution of attention in the news coverage in Table 5.2:647 Issue Distribution of attention for issues (%) General Before After 18 April 2005 18 April 2005 European Constitution 37.4 32.3 42.1 Content of the European Constitution 24.5 32.3 16.9 Referendum Campaign 16.3 12.1 20.3 Euro 8.7 5.4 11.9 Enlargement with Turkey 5.8 11 0.9 European Integration 4.8 3.3 6.3 Enlargement (Turkey not included) 2.5 3.4 1.6 Table 5.2: Topical distribution of the Dutch news coverage

644 Ibid. 645 Ibid., 13. 646 Ibid., 14. 647 This study is based on automatic content analysis of the newspaper coverage for the period between June 1st 2004 and June 1st 2005, and also manual content analysis of more specific newspaper coverage between January 25th and June 1st 2005. Takens, "A Negative Campaign? A Study of the News Coverage of the European Constitution and its Effects". 234

The media content data is not well-suited for analyzing the extent to which individual campaign frames were covered in the media, because the topical distribution is based on broad categories. Nonetheless, it shows that the No campaign themes were indeed placed on the agenda and linked to the TCE. Moreover, using both media content and public opinion data, Takens argues that the amount of attention for the TCE had a larger effect on the opponents of the Treaty than the supporters.648 In other words, more respondents judged the content of the European Constitution as a reason to vote negatively rather than positively. This once again confirms that the No campaigners used the content of the text more effectively in demonstrating their frames. The Yes campaigners highlighted the institutional elements of progress brought in by the TCE but these frames were not strong due to their technical content, as I will further demonstrate below.

Public Opinion Data In this section I will explore the public opinion data, to answer: To what extent do the campaign frames match the reasons given by the voters explaining their voting behavior? Below I present results from a dataset designed by Aarts and van der Kolk, which contains pre-measurement (22 April-31 May 2005) and post- measurement (2-12 June 2005) survey data among the same group of citizens.649 This dataset provides two ways of analyzing why the Dutch rejected the TCE. One set of questions ask the respondents to evaluate specific statements, which highly resemble the campaign frames I have identified in the previous sections.650 This presents the best test for the persuasiveness of the campaign frames, as respondents are asked to evaluate each statement individually. In the second type of questions, the respondents name the main reason for their voting

648 Ibid. She takes 18 April as the start of the actual campaign, the day the Minister of Justice Donner warned for the negative consequences of a rejection. 649 This data was originally collected by GfK Benelux on behalf of Kees Aarts and Henk van der Kolk, both of the University of Twente. The study was made possible by a grant from the Ministry of Home Affairs and Kingdom Relations. The original collectors of the data do not bear any responsibility for the analyses or the interpretations published here. 650 The results I present here are from the post-referendum survey as the questions asked in this survey are more detailed. The pre-referendum survey results are also similar. 235

behavior but here the answers are limited to one argument. Furthermore, the valid response rates to the first method is much higher, ranking between 65% and 82%, whereas the valid responses to the second method is lower, between 23% and 44%. I will analyze the two sets of questions in turn.651 Table 5.3 shows various statements and the respondents are asked whether these statements could be the consequences of European unification. Although ‘European unification’ is not exactly synonymous with ‘European Constitution’, it can be considered a close proxy because the TCE represents a step towards increasing European integration, and more importantly, it specifies its direction for the future. Furthermore, these statements were specifically brought up during the campaign, although the ones on welfare and international organized crime are reversed. The first five statements test the No campaign frames, whereas the last three statements analyze the Yes campaign frames. Certainly Probably Probably Certainly (%) (%) Not (%) Not (%) Smaller member states will lose influence 39.7 44 13.1 3.2 Wealthier countries will be obligated to 34.4 44.9 18.6 2.2 pay more Employment will move to other countries 47.4 44.7 6.9 1 where production is cheaper Welfare will increase 4.4 30.3 54.7 10.6 Our national identity as well as our 14.3 28.3 35 22.4 national culture will disappear Europe will try harder to achieve 20.3 57.3 19.7 2.7 worldwide peace and stability International organized crime will 17.4 33.5 42.2 6.9 increase Environment will be better preserved 8.5 45.3 40.4 5.9 Table 5.3: Dutch public’s opinions on the consequences of European integration

The answers suggest that the first No campaign frame concerning the loss of sovereignty, the second No campaign frame on relocation of industry and decreasing welfare as a result of the anti-social EU policies, and the last No campaign frame touching upon the Dutch budgetary contribution were indeed strong. These answers can be interpreted as an expression of fear of globalization,

651 All the figures represent percentages of valid answers. 236

where right-wing and left-wing voters uncomfortably meet. While the left-wing voters might be concerned about the social dumping and the race to the bottom in terms of social standards, the right-wing voters might consider foreign workers as a threat because of increasing immigration and identity-related reasons. The Dutch did not think that they would lose their national identity and national culture, although the balanced distribution of the percentages shows that the public was confused about this matter. On the other hand, the resonance of the Yes campaign frames is less clear. The public believed that the EU would contribute to international peace and stability but they were considerably less persuaded by its contribution to environmental protection and the fight against organized crime. These responses demonstrate that the third Yes frame, that the EU would help fighting common threats, only had limited success. This problem with the environmental agenda could partly be because of the decidedly hesitant and balanced GreenLeft campaign, as I explained above. The dataset also asked the respondents to locate the positions of political parties towards the TCE; interestingly, a great majority of the population perceived the GL as opposed to the treaty.652 The following questions in Table 5.4 test the campaign frames further. The perception of the Euro, the loss of sovereignty and the religious roots of the EU can be seen to test the No frames. In turn, the last question measures the opinion on a common Minister of Foreign Affairs, which was used as a Yes campaign frame.

652 However, these responses are from the pre-referendum survey. The correct answers increase over time, and the perception of the party positions become more accurate for each party. For detailed data on the evolution in the responses, see Kees Aarts and Henk van der Kolk, "Op Weg Naar 1 Juni," in Nederlanders en Europa: Het Referendum over de Europese Grondwet, ed. Kees Aarts and Henk van der Kolk (Amsterdam: Utigeverij Bert Bakker, 2005), 167. 237

Fully Agree Agree nor Disagree Fully agree (%) disagree (%) disagree (%) (%) (%) Transition from the Guilder to 37.3 32.9 17.5 9.8 2.5 the Euro caused serious damage to the Netherlands The introduction of the Euro is 3.6 21.8 21.1 40.3 13.2 beneficial to Dutch economy Prices in the Netherlands have 60.1 33 4.1 2.5 0.3 risen because of the introduction of the Euro Euro introduction made foreign 55.9 39.9 3.2 0.5 0.5 payments easier Right to exercise the power of 68.3 31.7 veto should be maintained European Constitution should 48 52 acknowledge Judeo-Christian roots Minister of Foreign Affairs is 57.4 42.6 necessary Table 5.4: Dutch public’s evaluation of various steps taken by the EU

Even though the public believed that the Euro made foreign payments easier, they clearly deemed it a costly step, in parallel with the No campaign framing. The success of the first No frame – loss of sovereignty – is visible once again, as the majority of the public expressed their approval to maintain veto powers. The religious concerns occasionally used by small Orthodox protestant parties, the CU and the SGP, were not very decisive. Similarly the majority supports a common Minister of Foreign Affairs, but not at a very high level, indicating that the second Yes frame presenting the institutional advances of the treaty was not particularly strong. Aarts and van der Kolk break down the same data (on the ‘Euro’) in Table 5.5 to show the No voter percentage within each category:

238

Agree % No Disagree % No (%) within (%) within this this category category (%) (%) Transition from the Guilder to the 69 73 12 29 Euro caused serious damage to the Netherlands The introduction of the Euro is 26 39 53 75 beneficial to Dutch economy Prices in the Netherlands have 93 63 2 21 risen because of the introduction of the Euro Euro introduction made foreign 96 60 1 75 payments easier Table 5.5: Dutch public’s evaluation of the Euro – the Yes/No breakdown

The results clearly show that dissatisfaction with the Euro became closely linked with the vote on the TCE. Those who believe that the Euro was a costly step are mostly the No voters, meaning that this particular dissatisfaction correlated with the No vote intention. Studying the reasons behind the Dutch contention with the Euro thoroughly, Engelen notes that the currency was not only technically unrelated to the TCE, but also separated from it in time.653 He stresses the effectiveness of campaign framing by the SP MPs such as Marijnissen and van Bommel as political entrepreneurs. This framing, combined with the remarks of Henk Brouwer on the undervaluation of Guilder during the transition to Euro, led to a sensational story during the campaign. His faux pas was transformed into political capital by the SP MPs.654 In the surveys there were also direct questions on the TCE’s consequences on various domains, as can be seen in Table 5.6:

653 Engelen, "How to Solve the Riddle of Belated Euro Contestation in the Netherlands?." 654 Ibid. 239

Positive No Negative (%) Difference (%) (%) TCE’s Consequences for Collaboration (Improve/Get 42.3 42.8 14.9 Worse) TCE’s Economic Consequences (Increase/Decrease) 33.4 44.9 21.7 TCE’s Democratic Consequences (More/Less) 27 42.1 30.9 TCE’s Consequences for Social Security (Improve/Get 3.5 31.2 65.2 Worse) TCE’s Consequences for Turkey (Speed up/Slow down) 23.1 59 17.9 Table 5.6: Dutch public’s opinions on the consequences of the TCE

Here the first three statements mirror the Yes campaign frames. The TCE is considered a positive step for European cooperation among member countries despite the high percentage that believed it would make no difference. The opinions on the economic and democratic consequences are more balanced, indicating that these frames were not particularly strong. The last two statements provide a test for the second and last No campaign frames. The second No frame on losing social standards was particularly strong as a great majority believed that the TCE would worsen social security in the Netherlands. The opinions on the last No frame – Turkish accession – are less straightforward as a majority said the TCE would not make any difference. Answering a further question on their opinion on the Turkish accession, 64.7% said that under certain conditions Turkey could join, while 35.3% thought it should not be allowed to join. The second set of questions is more direct, asking the respondents their main reason for voting in favor or against the TCE. Thus the answers are limited to one concern, and as I mentioned above, the valid answer rate is lower with this type of questions.

240

Reasons to Vote Yes Before After the Vote the (%) Vote (%) Deepening 27.1 18.6 unification/collaboration Good to have a Constitution 25.5 23.8 Stronger Europe 9.9 11.9 Constitution is unavoidable 5.2 10.8 In order to have a say in Europe 4.7 4.5 Both pro’s and con’s/Difficult to 3.3 2.2 decide Authority politicians/Political 2.5 3.5 party Democracy 2.2 2.4 Economic growth 2.2 4.3 Distribution of power between 1.4 1.7 small and large nations Power of the Netherlands 1.4 (-) Voting against is disadvantageous 1.4 (-) for the Netherlands Collaboration/Unification 1.1 3.2 Free and peaceful Europe 1.1 0.6 Decisiveness (-) 2.4 Defense policies 0.5 2.4 Equality (-) 1.1 Table 5.7: Reasons to vote Yes in the Netherlands

Table 5.7 shows that the Yes voters provided a wide variety of reasons to support the TCE. Given the differentiated Yes campaign frames, this distribution is not surprising. The high ranking reasons stressed a further step in European integration, indicating broad and non-specific support for the Union. The institutional content of the treaty ranked lower for the voters. Some voters explicitly mentioned that they could not decide whether the positive or negative sides gained the upper hand, which is coherent with the hesitant nature of the Yes campaign that signaled a message of compromise. A small percentage emphasized the disadvantageous consequences for the Netherlands, which could be interpreted as being in line with the dramatic remarks by the ministers

241 mentioned above. Interestingly, an important percentage of voters approved the treaty because they found the TCE unavoidable. The fourth Yes frame emphasizing social progress, the fifth Yes frame pointing at the economic benefits gained through Europe, and the last Yes frame indicating that there was no European ‘super-state’ were not effective despite being strong frames. This confirms the framing analysis above: There were stronger No campaign frames on the subject, which successfully countered these Yes frames, as can be seen in Table 5.8.

242

Reasons to Vote No Before After the Vote the Vote (%) (%) Power of the Netherlands 10.3 3.6 Loss of sovereignty 9.3 7.4 Not enough information/ 8.8 17.5 Consequences unknown/ Constitution is unclear Bad to have a Constitution 5.5 14.5 Distribution of power between 5.0 0.4 small and large nations Identity 4.8 5.3 No advantages 3.6 (-) Political distrust/cynicism 3.6 4.2 Deepening 3.3 1.0 unification/collaboration Velocity 3.3 7.6 unification/collaboration Reject the present government 2.6 2.4 No improvement 2.4 (-) Enlargement/ More countries 2.1 2.7 We pay too much 1.9 4.2 Loss of Dutch culture 1.4 0.1 Religion (as content) 1.4 1.6 We have to improve the 1.4 3.4 Netherlands first before paying attention to Europe Non-democratic decision 1.2 1.0 making in Brussels Big countries do what they want 1.2 3.0 Admission of Turkey 1.2 0.3 Euro 1.2 1.3 No need for another 0.7 3.0 Constitution The quality of the campaign 1.2 2.1 Social welfare 0.7 1.8 Table 5.8: Reasons to vote No in the Netherlands

243

The motivations specified by the No voters are also various, however a few key themes emerge. The power of the Netherlands, the loss of sovereignty, the distribution of power between small and large nations – all reasons with higher percentages – echo with the main No campaign frame that the EU was becoming a ‘super-state’ in which the Netherlands would not have sufficient power. Also, the statement that it would be bad to have a constitution could be evaluated as being in line with this broad concern, as the No campaign presented the constitution as a major feature of the ‘super-state’. These suggest that the Dutch feared losing control in the EU, which can also be seen in reasons that stressed identity, loss of Dutch culture, deepening and speed of the integration project and enlargement, as well as in the statement that big countries do what they want. Another high-ranking reason suggests that the voters did not have enough information on the unclear text, and had doubts on its consequences. This could be seen as partly a result of the No campaign strategy to create doubt and warn the voters that they should not vote Yes for something they did not understand thoroughly. Specific issues such as the budgetary contribution to the EU, the Euro, the admission of Turkey and social welfare carry small percentages. However, the fact that they were presented as reasons is still important as these issues were not directly related to the TCE before the No campaign linked them to the referendum proposal. Some voters specifically referred to political cynicism and the rejection of the present government, which are not related to the campaign frames, but show distrust in the Dutch political elite. Another survey by Marketresponse at the end of May 2005 asked the respondents about their voting intention in three imaginary referenda on the introduction of the Euro, the 2004 enlargement, and the decision to start negotiations with Turkey. The majority of the respondents who intended to vote No in the TCE referendum said that they would have rejected all three other referenda proposals.655 On the other hand, among the people who intended to vote

655 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 31. 244

Yes on the TCE, only 10-15% stated that they would vote against these three proposals. This presents clear evidence that these issues were primed, as they differentiate the No voters from Yes voters. Thomassen similarly reports a small conversation from an interview conducted on the street during the campaign.656 The respondent says that he would certainly vote No, and the two reasons he provides are the Euro and Turkish enlargement. When the interviewer reminds him that the TCE is not about these issues, he responds that his opinion on these themes was never asked before. Therefore, even if these issues were irrelevant, they were connected to the TCE during the campaign. Aarts and van der Kolk mention both the Euro and Turkey as central issues in the debate, and refer to the Thomas theorem in their explanation, saying that ‘if men define things as real, they are real in their consequences’.657 Overall, the strategic use of the ‘super-state’ frame by the No campaigners, complemented with concrete issues such as decreasing welfare, EU enlargement, the Euro, and the accession of Turkey worked well and connected all these themes to one another, as I described in detail above. The SCP has conducted a focus group study and found that the content of the Constitution remained a black hole for almost everyone in late 2004-early 2005.658 Whereas citizens were moderately positive towards the TCE in early 2005, this switched over the next few months into a predominantly negative attitude. People frequently mentioned that they did not understand the information provided by the government and dismissed the available leaflet as ‘uninteresting, unattractive and heavy-going’. The most important negative issues were the expensive Euro, the threat of competition from cheap labor, and the position of the Netherlands as a net contributor. Similarly, people had the feeling of ‘losing out’ because of Brussels, and that their employment and income position were under threat. The

656 Jacques Thomassen, "Nederlanders en Europa. Een Bekoelde Liefde?," in Nederlanders en Europa: Het Referendum over de Europese Grondwet, ed. Kees Aarts and Henk van der Kolk (Amsterdam: Utigeverij Bert Bakker, 2005), 64. 657 Aarts and van der Kolk, eds., Nederlanders en Europa: Het Referendum over de Europese Grondwet. 658 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 37. 245

report also finds that as the date of the referendum approached, the issues sharpened and became more tangible, causing attitudes to turn more negative, as opposed to the initial general positive approach towards the Union. Finally, the Yes campaigners’ lack of success in persuading their followers is reflected in the data showing the breakdown of the No voters according to their political party preferences. The social democrats were the least successful in keeping their voters in line with their party’s position, which could be a result of their hesitant campaign strategy. 63% of the PvdA voters, 53% of the GL voters, 49% of the CDA voters, 48% of the VVD voters, and 37% of the D66 voters refused to follow their party’s advice and voted negatively.659 The No campaigners, the SP and the other smaller parties, had more than 85% of their voters on their side – with the exception of the CU at 75%. The themes that emerge from the surveys as reasons behind the No vote are closely connected to the issues deemed problematic by the Dutch, which I discussed in the previous sections. I argue that this is a result of the strategic campaigning choices made by the No campaigners, who managed to set the agenda and link contentious themes to the TCE vote. The post-referendum study by Aarts and van der Kolk also asked the Dutch public to pick the most important social issues in the Netherlands; the answers, in descending percentages, are: Crime, healthcare, unemployment, safety, integration, intolerance, fading of norms and values, violence, lack of respect, economy, poverty, the expensive Euro, and the asylum seeker policy. It is no coincidence that these themes in turn emerged as reasons for the No vote. However, in the absence of No campaign framing which linked these issues to the TCE, the public opinion remained positive until the first months of 2005. Just as in France, the Dutch negative vote was a result of both domestic and European factors. According to de Hond polls, clear majorities of both the Yes (62%) and No (58%) voters were primarily motivated by their feelings

659 Aarts and van der Kolk, "Understanding the Dutch 'No': The Euro, the East, and the Elite." 246 towards European integration.660 The weight given to domestic political considerations were slightly different, 11% of the No voters gave this as their primary motivation as opposed to 4% of the Yes voters. The public opinion data, as a whole, clearly demonstrates that the vote was the result of a mix, caused by factors in both domestic and European spheres.

Diffusion The Dutch vote came only three days after the French rejection, on 1 June 2005. I maintain that the later a country held its referendum relative to other states, the more the previous referenda results and campaigns in the other states influenced its campaign dynamics and public. The Netherlands is an interesting case because it was not influenced by the Spanish and only minimally influenced by the French. It did not have an important impact on the Luxembourgish campaign either. Then the key question is: Why was this case not substantially influenced by the diffusion processes, despite being the third referendum in line? I propose two explanations. First, I argue that cross-case influences are conditional and dependent on the existence of transmission belts between the states such as shared language/culture, common media sources, and collaborative networks/transnational linkages. The more open these channels are, the more the later cases can be influenced by previous ones. Because the Dutch case shared only limited diffusion channels with the other cases, it was only minimally affected by the French case, and in turn, did not have a substantial impact on Luxembourg. Second, I argue that campaign intensity matters. As the French campaign was the most intense campaign of all, it was the French and not the Spanish campaign which put pressure on the existing diffusion channels and had minimal impact on the Dutch. Below I show that while the French case influenced the Dutch campaign to a certain extent, it did not have a significant impact on the vote itself.

660 Cited in Harmsen, "The Dutch Referendum on the Ratification of the European Constitutional Treaty." 247

To begin with the first channel, shared language and culture, almost all of the interviewees mentioned the importance of language as a facilitating factor. This came up mostly in reference to the difficulty in connecting with the French campaigners. Most of them were aware that the Luxembourg campaigners and the public followed the French debate more than the Dutch debate. Willem Bos, president of the ConstitutionNo group, acknowledged that the connection between the French and Luxembourgish campaigns was more intense than their connection to the Dutch one.661 He added that the Luxembourgish campaigners mainly invited French speakers from France. A member of the ConstitutionNo group Erik Wesselius, who visited Luxembourg during their campaign, also confirmed this pattern: ‘In Luxembourg there were people from France going and leafleting at doors. It is easier, they share the language. It would make no sense to get a Frenchmen here to campaign, there is too much cultural difference. Luxembourg is much closer to France both in terms of language and culture’.662 Van Hulten, director of the Foundation for a Better Europe, similarly mentioned that they had some connection with the UK, which did not hold a referendum but wanted to know about the Dutch experience.663 He added, ‘in the EU, they usually discuss with us, it is also the language which makes it easier’. Second, on the common media channels, the Netherlands and France do not share any common TV channels or newspapers, the way Luxembourg and France do. However, the French campaign was covered significantly in the Dutch media, starting in March with the attention to the fierce French debate on the Bolkestein Directive.664 Thus the news coverage during the campaign kept a close eye on the French mood.665 Aarts and van der Kolk have used several survey questions about the French referendum. The first one, in the pre-referendum survey, asked whether France was going to have a referendum, 96.6% of those

661 Personal interview, 8 April 2008. 662 Personal interview, 4 November 2008. 663 Personal interview, 30 January 2009. 664 Kleinnijenhuis, Takens, and van Atteveldt, "Toen Europa de Dagbladen Ging Vullen." See also Pelkmans and van Kessel, "Encapsulating Services in the 'Polder': The Bolkestein Directive in Dutch Politics." 665 Dekker and van der Horst, Europe's Neighbours: European Neighbourhood Policy and Public Opinion on the European Union, 49. 248 who answered knew the correct answer.666 In the post-referendum survey they asked the outcome of the French referendum, 99.6% of those who answered knew that the French rejected the TCE.667 Alternatively, they asked for the outcome of the Spanish referendum and of those who answered the question, 88.5% knew that the Spanish supported the TCE. However, 51.6% did not know and did not answer. The percentage who those did not know/answer in the French case was as low as 0.6%. This showed that knowledge about the French referendum was higher, in part because the media’s focus in the last phase of the campaign was almost exclusively on France and the Netherlands.668 It is important to note that here, the variation between the knowledge on the Spanish and French cases is not dependent on common media channels. Instead, it is essentially a function of the campaign intensity which gets reflected through media coverage. With its intense campaign, the French referendum – the momentum case – received better media coverage. The third and the final channel, collaborative networks and transnational linkages, were shared by all four cases. Here I explore the institutional networks and personal connections by looking at the European Parliament groups, the European anti-globalization network, ad hoc European networks, and existence of mobile communities. I discuss each in turn. Both the Yes and No campaigners pointed to EP groups and parties, to which their parties are members, as common platforms to share ideas with other similar European political parties. This forms a regular networking opportunity, where most of the campaigners said they discussed their experiences. However, there is significant variation on what type of experiences and advice were gained. Some of them mentioned the influence of these EP groups while the party was deciding on its position towards the TCE. Some others mentioned how Irish and Danish parties warned them not to have a referendum, based on their own experience. Therefore, while these networks formed a platform for political

666 27.9% did not know and so did not answer. 667 Only 0.6 % did not know and did not answer. 668 Cited in Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 29. 249

parties, the extent to which they were influential in shaping the campaigns is not clear. Some campaigners emphasized national differences as an impediment to full coordination of the campaigns. Hilde Laffeber, member of the Ministry of Foreign Affairs Yes campaign team, noted that as a strategy, they did not want to invite speakers from Germany, France or the UK, as they tend to overrule the Netherlands in the EU. She said this could have backfired.669 Nevertheless, concerning the institutional linkages, the far left and far right took extreme positions. The SP mentioned close contacts with the French Communist Party in setting the agenda. Van Bommel, MP of the SP, also mentioned that he went to Ireland several times to help them in their campaign against the Lisbon Treaty.670 The other extreme was the Dutch far right-wing parties, the LPF and the local party, Livable Rotterdam, which stressed that they were not xenophobic like Front National, or other radical right-wing parties in Europe. They rejected any links to these groups. Within the institutional networks, the European anti-globalization network is crucial. Just as in the other cases, these institutional linkages were more important in the analysis of the left-wing No campaigners. The strong anti- globalization network in Europe provides a platform for these actors to come together, discuss agendas, analyze new developments, and form positions. Willem Bos, president of the ConstitutionNo group, said that in preparation for the campaign, they first read on the topic; they knew the discussion in France as they were reading their work.671 Erik Wesselius, member of the ConstitutionNo group, also mentioned the ATTAC network, and the European Social Forum (ESF).672 He specifically referred to the ESF in Paris in November 2003, where the TCE was debated intensely. He said that ‘the No in France, and the whole debate in France was really very important for the other countries’. He added the legitimization impact of the French debate and results for the Netherlands,

669 Personal interview, 16 December 2008. 670 Personal interviews; first 8 April 2008, second 31 October 2008. 671 Personal interview, 8 April 2008. 672 Personal interview, 4 November 2008. 250

explaining how it lent credence to their arguments, strengthened their case, and contributed to their visibility. Consequently, this European anti-globalization network has spread a left- wing master frame, advocating an alternative ‘social’ EU, via the ESF and the ATTAC meetings. An important factor determining the success of borrowed frames is their adaptation to local reality. Despite mentioning close contacts with the French Communist Party, the SP secretary general van Heijningen strongly emphasized the importance of the work that had to be done nationally.673 He explained the strategic choice of their first argument, the loss of sovereignty in a European ‘super-state’, in order to complement the far left critique of the TCE prevalent in Europe. The SP campaigners frequently mentioned how the French had more political knowledge concerning privatization arguments, and that in the Netherlands, the Bolkestein Directive did not become as contentious as it was in France. Nevertheless, the ConstitutionNo civil society platform emphasized how they only used the left-wing arguments and did not use the SP argument on sovereignty. This once again shows the careful and strategic planning by the SP, as it is precisely their first argument, the European ‘super-state’, that worked as the master frame and was the most effective in the Dutch context. Alain Krivine, from the Revolutionary Communist League of France, explicitly criticized the SP for ‘making too many concessions’ to gain the popular vote.674 Lastly, besides the institutional networks I discussed, personal connections can become an important diffusion channel within the collaborative networks and transnational linkages. Yet this was not the case for the Netherlands. It did not have a mobile student or worker community which could have carried the debate across borders, as was the case between France and Luxembourg. The interviews are useful to discern the impact of the French campaign on the Dutch campaigners, especially on the left-wing civil society No campaigners. However, it is very difficult to evaluate the effect of the French case on the Dutch vote itself. The fact that they were only three days apart also poses a challenge in

673 Personal interview, 11 April 2008. 674 Personal interview, 25 September 2008. 251

terms of research. In the Luxembourg case, because their referendum was one month later, this aspect becomes visible in the Luxembourgish focus group data. After the Dutch referendum, Aarts and van der Kolk asked whether the respondents agreed that the French rejection made the outcome of the Dutch referendum less important. Of the ones who answered, while 35% fully disagreed, 49% disagreed; followed by 15% agreeing and 1% fully agreeing. This demonstrates that a great majority of the Dutch considered their referendum as an important venue to express their views, regardless of the French vote.

The Turmoil Campaign The French and Dutch rejections, only three days apart from each other, shocked Europe. ‘We have a serious problem but we should continue our work’, declared the President of the European Commission Barroso.675 The Dutch case was the most striking of all as it carried the highest No vote percentage. There was even a more puzzling fact, which was largely ignored by the literature: Just as in France, the contention with the domestic and European levels of governance was very much present in the society in fall 2004 when the polls showed positive attitude towards the TCE. I argue that the strong Dutch No campaign countered this initial support for the treaty. Although the Dutch case came third in line, this was a case of minimal diffusion due to the limited diffusion channels it shared with the other cases. A careful analysis of the campaign dynamics indicates that the result was far from shocking. It was the aftermath of a politically turbulent period for the Dutch, and the nerve which was already sensitive since 2002 was struck again by the campaign on the TCE. Similar to the situation in 2002, the 2005 anti-TCE frames (re)activated concerns of losing control in a rapidly changing Union, focusing specifically on the dangers of globalization such as decreasing welfare and increasing immigration. The EU was painted as a distant entity, unable to see people’s real concerns, as was the Dutch mainstream political elite. A local right-

675 Quoted in Deloy, "The Dutch Reject the European Constitution." 252

wing campaigner from the Livable Rotterdam party,676 Marco Pastors, specified the real problems Rotterdam faces as increasing pollution, crime and immigration, and called for an EU that addresses real concerns, not an EU which legislates on the ‘curve of bananas or dwarf-throwing’.677 While the TCE did not embody the widespread socio-economic anxieties of the Dutch, it came to symbolize them.678 Not only did the No campaign successfully set the agenda around contentious themes but the Yes campaigners were dramatically unsuccessful in reversing this mobilization. They suffered from a responsibility crisis and did not have effective frames to counter the No campaign framing. In an interview, Balkenende explained that misguided arguments on Turkey’s membership or the ‘super-state’ put the whole debate ‘on the wrong leg’ and that the vote had ‘nothing to do with the popularity of the cabinet’.679 Academic experts have also tried to clear up the ‘misconceptions’ from the campaign, such as that EU criminal law would not override Dutch tolerance of soft drugs.680 Paralleling the French case, the result was a sudden but episodic dip in public opinion. Post-referendum surveys found that Dutch public opinion towards the Union was restored to the previous positive levels. The French and Dutch cases demonstrate that specific negative arguments are more persuasive than arguments which idealize Europe. Negative facts often win out over vague benefits and ideals.681 A small scene from the campaign period summarizes the core dynamic in the Dutch campaign:682 During an evening lecture/debate organized by a regional newspaper at a Catholic church in Eindhoven in May 2005, SP activists were handing out leaflets at the door (possibly of a European map without the Netherlands). Inside, former Tilburg University philosopher

676 This party was also against the TCE, however it has campaigned only at the local level. Livable Rotterdam is the party of Pim Fortuyn, with which he has won the local elections in Rotterdam for the first time in 2002. 677 Personal interview, 8 April 2008. 678 Crum, "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?," 24. 679 Quoted in Lechner, The Netherlands: Globalization and National Identity, 263. 680 Ibid. 681 Dekker and Wennekers, "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen." cited in Dekker and van der Horst, Europe's Neighbours: European Neighbourhood Policy and Public Opinion on the European Union, 71. 682 Lechner, The Netherlands: Globalization and National Identity, 267. 253

Gido Berns sketched a grand vision of the new Europe made possible by the TCE. He was citing well-known European thinkers such as Kant and Habermas, to a somewhat befuddled audience, arguing that Europeans were ready to become hybrid citizens, both national and transnational. He was explaining that the sacrifice of sovereignty would help Europe show the way toward Kant’s eternal peace. While this is indeed the most important contribution of the European integration to European society, it did not resonate in a country where the public feared losing control of their state in Europe.

254

PERSONAL INTERVIEWS

Political parties:

Yes campaigners: Atzo Nicolaï (VVD, MP, then Minister of European Affairs) Jan Jacob van Dijk (CDA, MP) Marije Laffeber (PvdA, International Secretary) Bas Eickhout (GreenLeft, Delegate in the European Greens Party) Gerben-Jan Gerbrandy (D66, then Secretary of the D66 Parliamentary Group, now International Secretary) Michiel van Hulten (Director of Stichting Beter Europa) Hilde Laffeber (MFA, Member of the Yes Campaign Team) Delphine Pronk, (MFA, Head of the EU Communications Unit) Marco Pastors (Political Leader of the local party Liveable Rotterdam, Member of the City Council for Rotterdam) Otto Ter Haar (The Greens, International Secretary)

No campaigners: Harry van Bommel (SP, MP) Renske Leijten (SP, then leader of the ROOD, SP’s youth organization, now MP) Hans van Heijningen (SP, Secretary General) Esme Wiegman (CU, MP) Mat Herben (LPF, then Chairman of the LPF) Alexander van Hattem (Young Fortuynists, Youth Organization of the now defunct LPF)

Civil society:

Willem Bos (President of the ConstitutionNo, and ATTAC Netherlands) Erik Wesselius (Member of ConstitutionNo) Wim van de Donk (President of the WRR, Scientific Council for Government Policy) Monica Sie (President of the Scientific Council of the PvdA) Patrick van Schie (President of the Scientific Council of the VVD)

255

CHAPTER 6 LUXEMBOURG: THE DIFFUSION CASE

The Luxembourg referendum came last, a month after the French and Dutch rejections, on 10 July 2005. The referendum approved the European Constitutional Treaty by 56.52%, and the turnout was very high – above 90% as a result of compulsory voting. Luxembourg paralleled the previous cases on two dimensions: Early positive public opinion and the political party preferences. The public was initially positive towards the Constitution. Eight months before the referendum, in November 2004, the Luxembourg Institute for Social Research (TNS-ILRES) surveys showed that 64% was in favor of the TCE, while only 17% was against.683 Similarly, the fall 2004 standard Eurobarometer found that 77% supported the general idea of a constitution for the EU.684 The November 2004 special Eurobarometer further demonstrated that 57% of the Luxembourgish society supported the TCE specifically, while only 12% was against.685 Regarding the political parties, following the three previous cases, Luxembourg’s political mainstream formed the Yes campaign; the Christian Social People's Party (CSV), the Democratic Party (DP), the Luxembourg Socialist Workers' Party (LSAP), and The Greens campaigned for the TCE. Only the extremes, the far right and the far left joined the No camp; the Alternative Democratic Reform Party (ADR),686 The Left and the Communist Party (KPL) campaigned against the TCE. Just like in Spain, the Luxembourg vote did not attract much scholarly attention since it yielded a positive result. However, it provides an interesting puzzle for two reasons. First, the result was surprising as the No vote reached 43.48% in this highly pro-EU country with a highly popular government. In fact, satisfaction with both levels of governance was the highest among the four. This relatively high No vote percentage should not be ignored by categorizing Luxembourg simply as a positive case. Luxembourg was not able to sustain its

683 Available at: http://www.tns-ilres.com/cms/Home/Nos-secteurs/Politique---Opinion 684 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004," 29. 685 ———, "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution." 686 The party was then called Action Committee for Democracy and Pensions Justice. 256

positive attitude as successfully as Spain did. Second, this case was the fourth and last referendum on the topic and thereby provides a great opportunity to study diffusion dynamics. While France influenced Luxembourg to a great extent, the other two cases did not. Thus this chapter has two key questions. First, why did the No vote in Luxembourg reach 43.48% despite the early positive opinion on the subject? Second, why was this case influenced by the French referendum but not by the Spanish and the Dutch? I argue that the Luxembourgish No campaign was not as strong as the French or Dutch No campaigns, yet not as weak as the Spanish either. The No campaign suffered from lack of resources, credibility and partly from a lack of fit between their frames and contention in the society. They had only limited success in linking certain contentious issues to the referendum proposal in the public’s mind. Similar to the French and Dutch cases, the Yes side started as a weak campaign and had difficulty in countering the No campaign frames. However, they changed strategies following the French and Dutch rejections and used stronger pro-TCE framing. As a result they controlled the rising negative voting intentions. This explanation based on campaign strength is not complete as the impact of the independent variable was conditioned by diffusion in the following way. The second answer to the puzzle is rooted in the diffusion process as Luxembourg was the last campaign. Due to its peculiar geographic, demographic, and institutional set up, Luxembourg shared significant transmission belts with France. As a result, the Luxembourgish No campaign was considerably influenced by the strong French example. The main No campaigner, the left-wing committee, received explicit campaign support from the strong French left-wing No campaign, both in terms of borrowed campaign frames and logistics. Therefore the existence of the Luxembourgish No campaign depended on the French example, but at the same time these borrowed frames fit the local reality only partly, resulting in a semi-strong No campaign framing. In turn, strikingly, the heavy exposure to the French campaign caused some Luxembourgers to confuse French problems with their own problems, which could be seen as

257

remedying the adaptation problem to some extent by strengthening the borrowed frames. On the other side, the Yes campaign’s shift to a stronger campaign, following the double rejection, is also partly caused by diffusion. As such, diffusion has an input in both the increase of the negative vote and its containment in the final stage. Among the four cases, it was only Luxembourg where diffusion significantly conditioned the outcome. It amplified the effectiveness of the campaigns directly. The structure of the chapter will be as follows. First, I will present the evolution of the public opinion on the subject throughout the campaign. Second, I will discern the salient contentious issues of the period by exploring the public’s discontent with both domestic and EU levels of governance. Spotting these concerns will help understand the initial public opinion towards the referendum proposal, and draw attention to the societal problems that formed potential No campaign material. Third, I will conduct an analysis of the long campaign based on extensive interview data. I will examine the Yes and No campaign strategies and particularly focus on the strength of the campaign frames, credibility of the speakers, and mobilizational structures available to both sides. These components will explain why the No campaign only had limited success and how the Yes campaign used better pro-TCE framing with new strategies after the double- rejection of the TCE. Fourth, I will analyze the referendum vote, using public opinion and focus group data, to explain the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. The data will confirm the preceding campaign frame analysis. Last, I will discuss why the Luxembourg case was the only case where diffusion had a significant influence and why it was particularly affected by the French campaign. In this final section I will explore the diffusion channels between Luxembourg and the other cases and show how diffusion is critical to understanding the campaign dynamics.

258

Evolution of the Public Opinion Figure 6.1 shows the evolution of vote intentions over time:687

Source: TNS-ILRES

Figure 6.1: Evolution of vote intentions in Luxembourg

The No vote starts increasing mid-march onwards but this increase is contained in the last phase of the campaign. These shifts are closely related to campaign strategies of the Yes and No sides, as I will demonstrate below.

Discontent with Domestic and EU Governance In this section I explore the Luxembourgish public opinion towards the domestic and EU levels of governance in the period leading up to the referendum. Luxembourg was highly satisfied with its government and the public attitude was also remarkably positive towards the European integration. As in the other cases, this non-specific positive attitude can be assumed to contribute to the initial

687 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés. 259

favorable opinion on the TCE. Yet at both levels there were certain problematic issues that potentially constituted No campaign material.

Contention concerning domestic politics The Grand Duchy of Luxembourg held legislative elections in June 2004, which resulted in the re-election of the center-right Christian Social People's Party and the Prime Minister Jean-Claude Juncker. Since the Second World War, with the exception of 1974-1979, every Luxembourgish Prime Minister has been a member of the CSV. Thus the CSV has been in power almost continuously since 1945 choosing alternately the socialists or the liberals as coalition partners. After the 2004 election the CSV and the social democrat LSAP formed a coalition. The main contentious issues of the period were unemployment, education, and immigration. The TNS-ILRES monitor the public opinion on government performance regularly. The September 2004 survey inquiring the level of satisfaction with the government found that 71% of the public was satisfied whereas only 22% was not satisfied.688 Furthermore, the same survey asked the respondents about the degree of confidence in the government. On a scale from 1 to 10, the average response was 6.3. The TNS-ILRES surveys also ask the public to specify the most important problems in Luxembourg. The September 2004 survey responses show the following concerns in decreasing order:689 unemployment, inflation, environmental problems, poverty, human rights violations, crime/terrorism, and corruption. Eurobarometer data from fall 2004 underlines similar concerns: When asked about the two most important problems facing their country, the Luxembourgers selected the unemployment first and the educational system second, followed by rising prices/inflation, housing, immigration, economic situation, crime, health care system, public transport and terrorism.690

688 Available at: http://www.tns-ilres.com/cms/Home/News/Publications/Archive/Mammerent 689 Available at: http://www.tns-ilres.com/cms/Home/News/Publications/Archive/Voice-of-the- People--Probleme-in-Luxemburg-- 690 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 260

Because Luxembourg is a small country with peculiar characteristics, the problems mentioned in the surveys need to be assessed within its particular context. I will first discuss the economic concerns, and then move on to the related issue of immigration. Luxembourg’s economy has been stable and powerful, historically featuring solid growth, low inflation, and low unemployment.691 However, the unemployment rate rose to 3.8% in 2003 and 5% in 2004, while it was strictly below 3% in the years before.692 From 2001 on, the Luxembourg economy followed the global slowdown in economic activity, reflected as a significant deterioration in the labor market and heavy rise in unemployment.693 The high ranking of unemployment in the survey answers should be read as a reflection of this economic downturn. As opposed to the situation in France and the Netherlands, the Luxembourgish economic concerns do not prioritize pension security and social protection. The welfare state of Luxembourg has been particularly strong, and among the EU 15 Luxembourg has the highest social welfare spending per head in absolute terms.694 While many European countries had a crisis in financing their social security systems, Luxembourg avoided this problem due to the high increase in the number of workers paying contributions over the past years, which led to increasing revenue.695 The influx of labor, particularly cross-border labor, rejuvenated the population; therefore the effects of an ageing population on

691 The country continues to enjoy an extraordinarily high standard of living and its GDP per capita (PPP) ranks third in the world, after Liechtenstein and Qatar, with $81,200. CIA, "The World Factbook,"(2010), https://www.cia.gov/library/publications/the-world- factbook/geos/lu.html. 692 See OECD unemployment statistics, available at: http://stats.oecd.org/index.aspx?queryid=21760 693 STATEC National Statistical Institute of Luxembourg, "Economic and Social Portrait of Luxembourg," (Information and Press Service of the Luxembourg Government and Ministry of Foreign Affairs, 2003), 27. 694 It exceeds the level in Denmark, which is in second place, by 13.6% and compared with the EU average, the gap is 46.8%. However when the percentage share of GDP is considered, Luxembourg is among the countries where this is lowest due to its economic performance. Ibid., 183. Also see Smith, France in Crisis: Welfare, Inequality and Globalization since 1980. 695 For a detailed description of the Luxembourg welfare system see International Monetary Fund, "IMF Country Report: Luxembourg," (2006).; Nicole Kerschen, "The Welfare System of Luxembourg: From Past Dependency to the European Approach," in The Handbook of European Welfare Systems, ed. Klaus Schubert, Simon Hegelich, and Ursula Bazant (Abingdon: Routledge, 2009). 261

pensions and healthcare spending are hardly felt.696 Thus the major economic concern for Luxembourgers has been unemployment. The unemployment problem is linked to the particular type of economy Luxembourg developed in the past decades. The 1970s saw a major transition in economy with the decline of the iron and steel industries and the growing banking industry. The recognition of Luxembourg as an international financial center was solidified through banking secrecy laws that prevent the disclosure of information about account holders.697 The political stability, good communication, easy access to other European financial centers, skilled multilingual staff, and banking secrecy have contributed to the growth of the financial sector.698 As a result, today, the service sector accounts for more than three-quarters of total employment. Precisely this shift from an industrial to a service-oriented economy necessitated higher language, literacy and technical skills. Yet the education system has been unsuccessful in fostering such skills especially among the lower classes, leading to unemployment among Luxembourg nationals.699 Therefore, the top two problems highlighted by the Eurobarometer survey results above, unemployment and educational system, could be seen as two sides of the same coin. For many years, there has not been enough native labor to meet labor requirements.700 The demand for labor could only be met by using foreign labor – both immigrants and

696 Since 2004, Luxembourg has faced a structural deterioration of public finances due to the fast growth in public expenditures in relation to GDP. Accordingly, in 2006, the government decided to reduce public expenditures particularly in the field of social security. Just like other European countries, incremental liberalization steps have also been taken over the last decade in telecommunications, energy distribution and rail transport within the context of globalization. However in 2004-2005, the major economic concern for Luxembourgers has been unemployment. See Kerschen, "The Welfare System of Luxembourg: From Past Dependency to the European Approach," 323.; STATEC National Statistical Institute of Luxembourg, "Economic and Social Portrait of Luxembourg," 48. 697 Kathryn Anne Davis, Language Planning in Multilingual Contexts: Political Communities and Schools in Luxembourg (Amsterdam and Philadelphia: John Bejamins Publishing Company, 1994), 5. For a detailed analysis of banking secrecy in Luxembourg see Jérôme Lasserre Capdeville, Le Secret Bancaire: Étude de Droit Comparé: France, Suisse, Luxembourg (Aix-en- Provence: Presses Universitaires d’Aix-Marseille, 2006). 698 United States Department of State, "Country Reports,"(2010), http://www.state.gov/p/eur/ci/lu/. 699 Davis, Language Planning in Multilingual Contexts: Political Communities and Schools in Luxembourg, xix, 179. 700 STATEC National Statistical Institute of Luxembourg, "Economic and Social Portrait of Luxembourg," 72. 262

cross-border workers:701 The number of jobs occupied by Luxembourg nationals remained virtually constant from 1980 (100,000), while the number of cross- border workers rose from 3,700 in 1961 to 17,000 in 1985, followed by a drastic rise to over 100,000 in 2001. In 2002, only around 35% of wage earners employed in Luxembourg were natives of the country. In turn, a large number of Luxembourgers are employed in the ‘protected’ public sector, (general government, railways) and the para-public sector (energy, water and national health):702

Figure 6.2: Breakdown of Luxembourg workforce according to nationality

As Figure 6.2 demonstrates, almost 40% of the Luxembourgish working population is employed in the public administration, 90% of which is ‘reserved’ for Luxembourg nationals. This segmentation of the labor market became important when the 2005 referendum campaign brought the risk of privatization of public services to the forefront. This claim particularly threatened the citizens of Luxembourg, which I will discuss below. Overall, its labor market patterns make Luxembourg a very unusual case.

701 Ibid., 38. 702 Ibid., 40. 263

Concerning immigration, another problem mentioned in the survey answers, Luxembourg is once again in a peculiar situation because of its labor market. The population of the country is just below half a million. According to the 2000 census the ethnic breakdown is as follows:703 Luxembourger 63.1%, Portuguese 13.3%, French 4.5%, Italian 4.3%, German 2.3%, citizens of other EU countries 7.3%, and citizens of other countries 5.2%. This indicates that over one- third of the population is non-Luxembourger. Figure 6.3 shows the percentage share of foreigners in the EU member states in 2001:704

Source: Eurostat

Figure 6.3: Percentage share of foreigners in the EU member states

Apart from the permanent foreign population, Luxembourg has a significant amount of cross-border movement during the day due to its special geography and institutional architecture. It shares borders with France, Germany, and Belgium. Especially given its small size, these three countries can be reached within less than half an hour. Furthermore, because many of the EU institutions are located in Luxembourg, in 2009 around 9,000 international civil servants worked at these

703 CIA, "The World Factbook." 704 The graph does not show the full bar representing the EU citizens residing in Luxembourg, the bar normally extends beyond 30%. Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 34. 264

European institutions.705 Altogether, in 2009 more than 145,000 cross-border employees came into Luxembourg on a daily basis from France, Germany and Belgium.706 This number was around 118,385 in 2005, which shows that cross- border movement is rapidly increasing.707 The European Social survey measures the ‘ethnic distance’, which is the degree of resistance to social interaction with members of ethnic groups in various domains such as the residential neighborhood, work, and personal relationships.708 The ethnic distance in Luxembourg is lower than the EU average.709 However more detailed studies show some scepticism towards immigration and foreigners in Luxembourg. A March 2005 study by TNS-ILRES summarized the attitude of Luxembourg citizens towards the presence of foreigners in different domains.710 The results can be seen in Table 6.1: Economic Cultural Social Labor Crime General development richness Life market rate evaluation (%) (%) (%) (%) (%) (%) Good 64 64 54 49 10 53 Bad 12 12 18 33 63 17 Not good- 19 21 26 17 23 27 Not bad Table 6.1: Luxembourgers’ attitude towards foreigners in various domains

While citizens of Luxembourg value the foreign presence in general and in terms of economic development and the cultural/social domain, they are less enthusiastic when it comes to the labor market. As I explained above, the shift to a service-oriented economy necessitated the employment of multi-lingual and skilled foreigners in Luxembourg. The modest scepticism towards foreigners concerning the labor market could be seen as a result of this heavy reliance on

705 The National Portal for the Grand-Duchy of Luxembourg, (2010), http://www.luxembourg.public.lu/fr/societe/population/societe-multiculturelle/index.html. 706 Ibid. 707 The figures reflect the situation in 2004 and 2005. European Employment Services (EURES), "Les Cahiers Transfrontaliers d’EURES," 4(2005), http://www.eureslux.org/eures.taf?Idnav=70&Idnav=fr. 708 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 74. 709 Ibid., 75. 710 Available at: http://www.tns-ilres.com/cms/Home/News/Publications/Archive/La-perception-des-etrangers 265

foreigners in the labor force.711 On the other hand, the foreigners are perceived negatively regarding crime. The results therefore indicate that immigration is considered problematic in relation to two domains; crime and fear of losing jobs. The same survey further asked the respondents which groups of foreigners were responsible for these positive or negative effects:712 Citizens of Cross- Citizens of Citizens of EU 15 (%) border new EU non-EU workers (%) members (%) countries (%) Responsible 45 42 25 11 for positive effects Responsible 10 17 31 50 for negative effects Table 6.2: Luxembourgers’ attitude towards different groups of foreigners

Table 6.2 reveals which particular groups are seen as being more problematic – an important question in a country with such a complex immigration profile. The answers demonstrate that the attitude is significantly more favorable towards the citizens of the EU 15 and more sceptical towards the citizens of non-EU countries. Immigration is also seen as a threat to cultural cohesion.713 It is only after the economic change of the 1970s that Luxembourg launched an official government language policy protecting the Luxembourgish language Lëtzebuergesch.714 Linguistically, the diversity in the country is once again surprising. Luxembourgish is only one of three official languages along with French and German. While Luxembourgers use Lëtzebuergesch as their sole means of oral communication among each other, the high number of French

711 At the start of 2002, only cross-border workers (excluding immigrants) occupied over 37% of paid jobs in Luxembourg, leaving Luxembourg nationals at 35% as opposed to 70% in 1970. STATEC National Statistical Institute of Luxembourg, "Economic and Social Portrait of Luxembourg," 38. 712 This time the question is posed to all Luxembourg residents, not limiting it to Luxembourg citizens. 713 Davis, Language Planning in Multilingual Contexts: Political Communities and Schools in Luxembourg, 10. 714 The use of Lëtzebuergesch became intertwined with national emblematism. Gerald Stell, Luxembourgish Standardization: Context, Ideology and Comparability with the Case of West Frisian (Leuven: Peeters Louvain-La-Neuve, 2006), 26. 266 speakers led to a situation where Luxembourgers can expect to speak foreign languages regularly in their daily lives.715 Surveys show that French is the language best known in Luxembourg.716 In practice, however, there is no friction as there is little overlap between the realms of use allocated to each of them.717 To summarize, Luxembourgers were concerned about the economic situation and immigration at the domestic level of governance. Yet these problems should be interpreted in light of Luxembourg’s peculiar demographic characteristics. Because Luxembourg is exceptionally small and hosts a remarkable number of foreigners in its labor force, the main economic anxiety of Luxembourgers centers on unemployment. Similarly, the concerns over immigration are not of the xenophobic type pervasive in most Western European countries. Immigration has a different connotation in Luxembourg because first, the majority of non-Luxembourgers originate from EU member states as opposed to non-EU states, and second, there is a large group of cross-border employees who commute on a daily basis. Given these characteristics, immigration reinforces the fear of losing jobs and generates a mild fear of losing cultural cohesion. The fear of immigration is a mix of economic and cultural factors but the economic dimension is the primary one. Furthermore, because most of the Luxembourgish population is confined to public office jobs while foreigners dominate the private sector, there is sensitivity towards protecting the public sector. As such, Luxembourgers did not share the French and Dutch profile wherein the public was primarily concerned about the welfare state and immigration. Because of their country’s unusual demographic and institutional conditions, Luxembourgers’ anxiety rested upon unemployment and immigration, though these terms were defined differently.

715 Jean-Paul Hoffmann, "Lëtzebuergesch in Luxembourg Today," in Luxembourg and Lëtzebuergesch: Language and Communication at the Crossroads of Europe, ed. Gerald Newton (Oxford: Oxford University Press, 1996), 103. 716 Stell, Luxembourgish Standardization: Context, Ideology and Comparability with the Case of West Frisian, 29. 717 Also, there are no active politics around the language issue. Ibid., 26. 267

Contention concerning the EU Both the Luxembourgish public and the political elite have historically been highly pro-EU, making it by far the most Europhile country among the four referendum cases. Figures 6.4 and 6.5 show the trends in the Eurobarometer surveys:718

Figure 6.4: Luxembourgish support for EU membership (1984-2004)

718 Eurobarometer, "Standard Eurobarometer 61: Public Opinion in the EU, Spring 2004." 268

Figure 6.5: Luxembourgish public’s perceived benefit from EU membership (1984-2004)

When looked at a broader period from the Eurobarometer data, Figure 6.6 shows that the Luxembourgish public opinion has always been positive:719

Source: Eurobarometer Figure 6.6: Luxembourgish support for EU membership (1973-2005)

719 Dekker and Ederveen, eds., European Times: Public Opinion on Europe & Working Hours, Compared and Explained, 11. 269

Figures 6.4, 6.5, and 6.6 show that Luxembourgish opinion towards the Union is much higher than the EU average. As any other small state, Luxembourg has to take into account its military weakness and economic dependence on external markets.720 As a result, the Luxembourgish political elite have been very eager to further European integration. Luxembourg indeed owes much of its prosperity to the EU. It profited greatly from having access to the EU market as well as from having a weight disproportionate to its size due to the equal treatment of member states in most institutional arrangements.721 It also hosts a large number of EU institutions, which with their massive international staff have contributed considerably to Luxembourg’s economic development over the last half century.722 Moreover, Luxembourg is one of the biggest net recipients from the EU budget – per head of the population – in the last decade.723 Table 6.3 summarizes Luxembourg’s benefits and contributions between 2000 and 2005:724

Total benefit from the EU Total contribution to the EU (Euro million) (Euro million) 2000 885.2 161.4 2001 889.7 236.5 2002 954.1 173.7 2003 1,095.8 192.1 2004 1,132.0 218.7 2005 1,105.8 211.2 Table 6.3: Luxembourg’s benefit and contribution in the EU budget (2000-2005)

As can be seen in Table 6.3, the benefits received from the EU have almost always been four times more than the national contribution.

720 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 11. 721 Ibid. 722 These include the Commission, the Court of Justice, the general secretariat of the Parliament, the European Investment Bank, and the European Court of Auditors Ibid. 723 Ederveen and Dekker, Destination Europe: Immigration and Integration in the European Union, 11. 724 Detailed information can be reached at the EU’s official website on financial programming and budget. Available at: http://ec.Europa.eu/budget/budget_glance/where_from_en.htm 270

This positive picture does not suggest that Luxembourgers did not have any concerns regarding the European integration project. The Eurobarometer fall 2004 survey asked citizens of every member state about their fears concerning the building of Europe.725 While the sets of concerns highlighted by all four referendum countries are similar, Luxembourg has the highest rank among the four – also among the EU 25 – in stating the declining use of their language as a cause of concern. Luxembourg also ranks high when it comes to the fear of organized crime, the transfer of jobs to other member countries which have lower production costs, the loss of national identity/culture and the loss of power for smaller member states. These answers parallel the domestic concerns highlighting fear of losing jobs and cultural cohesion due to the peculiar conditions of the country. Moreover, Eurobarometer surveys show that in 2004, 23% of the population associated the EU with unemployment.726 Importantly, the trade union of the Luxembourgish public/civil servants is critical of the EU because of its plans to open public sector employment to EU citizens.727 The Eurobarometer surveys also showed that the EU enlargement was unpopular among the public for mainly economic reasons.

Political Party Attitudes towards Europe When looking at the political elite, paralleling the pattern in other countries, Euroscepticism in Luxembourg is confined to the extremes of the political spectrum. On the far right, we can find the Alternative Democratic Reform Party (ADR), and on the far left, The Left is classified as soft Eurosceptics in the literature.728 An in-depth study, using data from 1996, shows that with 5.5% Luxembourg holds the lowest percentage of hard Euroscepticism among the EU

725 Eurobarometer, "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." 726 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 12. 727 Personal interview with Ben Fayot, MP and the President of the Parliamentary Group of the LSAP, 14 November 2008. 728 Taggart and Szczerbiak, "The Party Politics of Euroscepticism in EU Member and Candidate States." 271

15 countries.729 But it also demonstrates that Luxembourg’s MPs and MEPs are largely pro-EU except in one domain: Enlargement. The scepticism towards the inclusion of new members does not form an outright opposition but rather a soft Euroscepticism. An important reason behind this attitude was the small size of the country. The enlargement from the EU 15 to the EU 25 meant new institutional arrangements involving the loss of an assured seat in the Commission.730 Apart from enlargement, another EU-related concern surfaced in the early 2000s: Banking secrecy. The banking secrecy has been an important asset for Luxembourg’s economic development. In 2003 the EU’s attempts to further European tax harmonization became a challenge for Luxembourg’s standing as a financial centre.731 In March 2009, Luxembourg, among other European countries such as Austria and Switzerland, agreed to loosen its banking secrecy laws to allow for more cross-border cooperation, after its fight not to do so.732 Yet, the issue of banking secrecy came up in the 2005 referendum campaign, as I will discuss below. Similarly, during the 2003 European Convention that drafted the TCE, many professional organizations and unions raised concerns about maintaining control over national policy to protect the Luxembourgish economic and social model.733 The Chamber of Commerce (CLC), the Federation of Luxembourg Industrialists (FEDIL), and the Association of Banks and Bankers of Luxembourg

729 Richard Katz, "Euroscepticism in Parliament: A Comparative Analysis of the European and National Parliaments," in Opposing Europe? The Comparative Party Politics of Euroscepticism Volume 2, ed. Aleks Szczerbiak and Paul Taggart (New York: Oxford University Press, 2008), 158-59. 730 This was particularly problematic for small countries as representation in bodies other than the Commission such as the Council and the Parliament is already population-weighted. Ibid. 731 The Council of the EU reached an agreement on the taxation of savings, aiming to share information on as wide a basis as possible. Nevertheless, Luxembourg was exempted from this regulation such that it would not automatically share information until the date on which ‘equivalent measures’ have been introduced by certain third countries and all dependent or associated territories of the EU. STATEC National Statistical Institute of Luxembourg, "Economic and Social Portrait of Luxembourg," 119. 732 Michele Sinner, "Swiss, Luxembourg, Austria Defend Bank Secrecy," Reuters(2009), http://www.reuters.com/article/idUSL865351220090308.; EurActiv, "Austria, Luxembourg and Switzerland to Change Bank Secrecy Rules "(2009), http://www.euractiv.com/en/financial- services/austria-luxembourg-switzerland-change-bank-secrecy-rules/article-180329. 733 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 15. 272

(ABBL) supported the draft constitution provided that it did not advance harmonization of social and fiscal policies at the EU level, trying to protect the favorable employer and tax laws of the county. Furthermore, the president of the Confederation of Independent Trade Unions of Luxembourg (OGBL, the trade union associated with the social democrat LSAP) Jean-Claude Reding warned against creating competition for workers, particularly in relation to different social legislation in force.734 Overall, while Luxembourg has been the most pro-EU country among the four referendum countries, there were issue areas that were deemed problematic both by the public and the political elite primarily because of the specific characteristics of the country. The public fears relating to the EU level of governance parallel the ones in the domestic sphere; fear of losing jobs and cultural cohesion as well as losing influence in the EU as a small member. Based on similar concerns, the political elite was mildly sceptical towards the enlargement process and the EU’s attempts to minimize banking secrecy. As can be seen in this analysis, Luxembourg is far from being a Eurosceptic country but the No vote did reach 43% in the 2005 referendum. Luxembourg was clearly not able to maintain its positive attitude as successfully as Spain did, yet the No vote formed only an episodic dip rather than permanent opposition, just as in France and the Netherlands. The fall 2005 Eurobarometer surveys found that 82% of Luxembourgers were in favor of their membership in European integration.735 The next sections analyzing the campaign and diffusion effects will show the reasons behind this dip. While the No campaigners achieved limited success in linking the existing concerns with the referendum proposal, the French referendum also contributed to the result.

734 Quoted in Ibid., 17. 735 Eurobarometer, "Standard Eurobarometer 64: Public Opinion in the EU, Autumn 2005," European Commission: Brussels (2005). 273

The Long Campaign This section will provide an in-depth analysis of the campaign strategies and framing of the Yes and No campaigners. I will focus on the agenda-setting and priming/framing stages in turn, based on personal interviews with the campaigners, campaign materials, television clips, and strategy papers of some parties.736 The analysis will show that the initial phase of the campaign highly resembled the French and Dutch campaigns. The No campaign started early and set their campaign themes on the agenda, while the Yes campaign had to take the defensive position and failed to build strong frames. However, the No campaign only had limited success due to its partially effective frames, problems of credibility and lack of resources. Interestingly, the Luxembourgish campaign had a turning point. Following the French and Dutch rejections, the Yes campaign changed its strategy and switched to stronger frames. Luxembourg was not as successful as Spain in sustaining the early positive public opinion on the subject. I argue that the No campaigners succeeded partly in linking the existing contention to the referendum proposal, yet the Yes campaign contained the rising negative public opinion by shifting its strategy in the last phase of the campaign. I first provide a brief overview of the campaign and the campaigners. The government decided to hold a referendum on the TCE in June 2003 but the date of the referendum was only announced in November 2004.737 Paralleling the previous three cases, the government ran an information campaign. The Yes campaign was formed by the CSV, the DP, the LSAP, and The Greens. Unlike the Yes campaigns in the other referendum cases, these parties formed a united front. On the other hand, the No campaign was run by the far left and the far right. On the left, The Left and the KPL joined the No campaign but they did not have any representation in the parliament at the time. Thus the main left-wing No campaigner was the No Committee, which brought together individuals from different civil society organizations. On the right, the ADR was first in favor and

736 A full list of the campaigners I interviewed is presented at the end of this chapter. 737 The 2005 referendum was the fourth referendum in Luxembourg’s modern history, yet the last one was held in 1937. 274

then against the TCE in late March, onwards.738 The ADR campaigners mentioned that they did not have a major campaign due to this shift. The parliament held three public hearings on the TCE where the public and MPs discussed various EU policies.739 The common Yes platform organized over a dozen hearings all around the country, to which they invited one speaker from the No Committee as well. In addition, the No campaign organized separate meetings mainly with participation of No campaigners from France.

Agenda-setting The literature shows that timing matters, as an early start gives the campaigners an advantage to set their themes on the agenda and thereby define the terms of the debate. In Luxembourg, the No campaign started earlier. The left-wing No Committee was established as early as October 2004 and they started inviting influential French figures from November onwards. Hence they had a chance to be on the offensive and place their themes on the agenda.740 Abbes Jacoby, Secretary General of The Greens’ Parliamentary Group, mentioned that the Yes camp was always in a defensive position.741 Similarly, Laurent Mosar, CSV MP, explained that their problem was to always be on the defense, trying to argue against arguments used by the No movements.742 The turning point in the Luxembourg campaign was the French and Dutch rejections in late May/early June. The Yes campaign, and particularly the CSV, started an active campaign during this period. It was only then that the four parties applied all of their strength and met with the population on several occasions.743

738 Corinne Deloy, "Referendum on the European Constitution in Luxembourg 10th July 2005," Robert Schuman Foundation Report (2005).; Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés. 739 Piere Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty," European Parties Elections and Referendums Network, Referendum Briefing Paper 14 (2005). 740 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 23. 741 Personal interview, 10 April 2008. 742 Personal interview, 14 November 2008. 743 Corinne Deloy, "Luxembourg Says 'Yes' to the European Constitution," Robert Schuman Foundation Report (2005). 275

This change was also accompanied by a change in the Yes campaign framing. I discuss this shift in detail below.

Priming and Framing As I argued previously, three factors mattered in the priming and framing stage; strength of the campaign frames, credibility of the speakers, and mobilizational structures. While the left-wing No campaigners started earlier and set the terms of the debate, their frames were only partly effective as they were remarkably inspired by the French left-wing campaign and not well-adapted to the Luxembourg reality. Although the Yes campaigners initially had difficulties in countering the No frames, they changed strategy in the aftermath of the French and Dutch rejections and ‘nationalized’ their framing. As I will demonstrate below, on the No side, the left-wing No Committee suffered from a lack of resources and the right-wing ADR from a lack of credibility. In contrast, the Yes campaign had higher credibility and better resources. Accordingly, when the Yes camp used strong frames in the final phase, they were more influential than the No campaign.

Content of the campaign frames I first describe the content of the Yes and No campaign messages and then discuss their strengths. Below I list the Yes and No campaign frames. I identified the campaign frames primarily based on the written campaign materials.744 Next, I categorized them according to their subjects such as peace, economy, democracy, institutions, and other salient issues like welfare state, immigration, accession of Turkey, or national identity. The main Yes campaign frames can be categorized as: 1. Approving the TCE means approving Europe. The European integration has guaranteed peace in Europe in the last 60 years.

744 Where available, one to two page leaflets listing the campaign arguments provided the best source. If not, I used interview data and secondary sources to complement the available materials. 276

2. The TCE would institutionally enhance the integration project, and make it more democratic, transparent, efficient, and powerful in the world. The new EU President and the Minister of Foreign Affairs, the strengthening of the European Parliament, and the clarification of competences of the Union and the member states are important steps forward. 3. The TCE and the Charter of Fundamental Rights would bring a social dimension to Europe with employment, social progress, and better environmental standards. Reforms brought by the TCE would also help in formulating a common environmental policy, immigration policy, and a common fight against organized crime. 4. The TCE would consolidate the power of small countries. The Commission would have the rotation system for both small and large states, regardless of the size of the member states. National parliaments would have more scrutiny over EU decision-making. Luxembourg would have more power in the Council of Ministers. Luxembourg’s six seats in the parliament would be saved. In short, the TCE would not threaten Luxembourg’s sovereignty. 5. Approving the TCE is good for both Luxembourg and Europe. Without Europe and the Euro, the preservation of the steel industry and the success of the banking sector would have been unthinkable. Approving the TCE would allow Luxembourg to continue its success and consolidate its place in Europe. As such, the vote is a matter of self-preservation and national interest for Luxembourg. The CSV, the LSAP, the DP, and The Greens ran both individual and united campaigns. The main Yes campaigner was the CSV, which used all of the Yes frames outlined above. In the final stage after the French and Dutch rejections of the TCE, the CSV emphasized the fifth Yes frame as their main frame. Their campaign slogan was ‘Jo fir Lëtzebuerg, jo fir d’Verfassung’ (Yes for Europe, Yes for the Constitution). In the last phase, Prime Minister Juncker’s key statements were also used on campaign documents: ‘Mir hunn äis ëmmer an der

277

Geschicht vun eisem Land zur Zäit richteg entscheed’ (In our country’s past, we have always made the right decisions). The key message of the CSV slogans therefore was: ‘What is good for Europe is good for Luxembourg’. The LSAP, the DP and the Greens also used all of the listed Yes frames. However the last frame was not as explicit or emphasized in their official campaign documents. The fifth frame was particularly stressed by the CSV, specifically in the aftermath of the double rejection. The LSAP adopted the slogan ‘Jo’ (Yes). The DP campaigned with ‘Blo seet jo, fir Europa!’ (Blue says Yes for Europe). Finally, The Greens used ‘Jo’ (Yes), as well as ‘Wësse stäerkt’ (Knowing makes stronger, or knowledge is the key). On the other side, the No campaign frames can be categorized as: 1. The TCE is not social enough; it includes market-friendly provisions without sufficient social rights and protection. It is a threat to the Luxembourgish social model and to the competition for Luxembourgish jobs because of social/fiscal dumping and liberalization of public services. The services directive, the Bolkestein Directive, seeks to establish free movement of services within the EU. The TCE would thus advance relocation of industry to countries with more favorable taxing conditions and lower working standards. 2. The TCE is not democratic enough; neither the process nor the result is democratic. 3. The TCE subordinates the EU’s security system to NATO’s and militarizes Europe by increasing military expenditures. 4. The TCE is not environmentally-friendly. 5. The TCE means the end of Luxembourgish sovereignty in a European ‘super-state’. 6. The TCE would further the enlargement process and would specifically advance Turkey’s entry into the Union. Turkish accession is problematic due to political and cultural differences. It would further shift the demographic balance in decision-making at the expense of a

278

small country like Luxembourg. Turkey’s accession would also be draining for the EU budget. The main campaigner on the No side was the left-wing No Committee, as the ADR changed its position during the campaign. The No Committee used the first four No frames listed above and campaigned with the slogan, ‘Non à la constitution, votez non pour une autre Europe’ (Vote No to the constitution for another Europe). The ADR on the other hand, used all of these No frames, stressing the first, fifth and sixth the most. Their slogan was ‘Nee zu dësem Vertrag’ (No to this treaty).

Relative strength of the campaign frames According to the findings of the framing literature I presented earlier, frames that invoke available, accessible and applicable concerns are stronger. They are also more effective if they are concrete, vivid, image-provoking, emotionally- compelling, and contain negative information. Just as in the other cases, the first and second Yes frames were not strong as they were both abstract and technical arguments. Since there was no imminent threat to European peace, the first Yes frame emphasizing peace was not an applicable concern. The second Yes frame similarly presented the TCE as an institutional step towards a better Europe, which was neither available nor accessible as the citizens were not very familiar with the technical content of the treaty. Both of these frames failed to evoke existing contentions or emotions in the society. The third Yes frame was stronger as it partly stressed contentious issues: Employment and social concerns. As such, this frame involved available, accessible and applicable concerns. Unemployment was a key concern in the society as I explained above. Social concerns were not central but important when it came to ‘protected’ public sector employment for Luxembourgers. But there were stronger No campaign frames on the subject, as I will discuss below. The fourth and fifth Yes frames sought to show that the TCE was specifically good for Luxembourg’s national interests. Thus their underlying

279

message was one that was accessible and available to the public. The fourth Yes frame was presented as an institutional argument and therefore should be expected to be less effective. The fifth Yes frame, while its variations appeared sporadically throughout the campaign, was forcefully used by the Yes campaign (specifically by the CSV) in the last phase after the French and Dutch rejections. This last Yes frame was particularly strong as it carried emotional content by evoking a fear of losing benefits gained from Europe. The Prime Minister frequently referred to this frame by reminding Luxembourgers that in the past they have always made the right decisions, signaling that a No vote would be a historic mistake. On the other hand, the first, fifth and sixth No frames were the strongest ones. While the welfare state-related concerns were not very contentious in Luxembourg, the first No frame also brought up the central problem of unemployment by stressing possible loss of jobs. The fifth and sixth No frames emphasized the loss of power and benefits, particularly in relation to the enlargement process. All three invoked available, accessible, and applicable concerns by using the recent contentious issues in the society. The Yes campaign sought to counter these issues in their frames but the No campaign frames were more effective because they carried negative information and invoked the fear of losing jobs and benefits. It was only the last Yes frame which was equally emotionally-compelling. The rest of the No campaign frames were not as strong, just as in the other three cases, since they were abstract and did not use contentious issues. The discussion of diffusion below will show in detail why and how the left-wing No Committee borrowed their frames from the French left-wing No campaign. These frames did not build directly on the particular concerns of Luxembourgers regarding unemployment and immigration – the fear of losing jobs and cultural cohesion. They were not tailored to the local context, for instance, by blaming the EU directly for unemployment, weakening of the national language, or emphasizing the possibility of opening up the public sector employment to the EU citizens. Instead, using the French No campaign framing,

280

the left-wing No campaigners indirectly touched upon some of these issues. The domestic contention was partly activated – indirectly – through the relocation of the industry argument and the privatization of public services. Moreover, strikingly, because Luxembourgers were highly exposed to the French campaign, they confused their country’s problems with the French problems. Regarding the campaign documents, in late March each Luxembourg household received a mini brochure on the referendum. Later in mid-May a document explaining all the articles of the treaty was sent out. Both the Yes and No campaigns used posters and leaflets. The campaign documents once again reflected the pattern in France and the Netherlands: The No campaign posters were more specific with actual argumentation on them, while the Yes campaign documents were mostly abstract with the exception of the CSV’s final posters. Below in Figures 6.7-6.11 posters of the Yes side – the CSV, the LSAP, the DP, and The Greens – show variations of the EU flag. But the CSV’s final poster in Figure 6.11 presents the same smiley face inside the EU flag, but this time coupled with a critical quote by Juncker: ‘In our country’s past, we have always made the right decisions’. In contrast, the No campaign posters in Figure 6.12 and 6.13 portray the TCE as poisonous, carrying the key message that it was detrimental to public services or peace.

281

Figure 6.7: The CSV’s poster Figure 6.8: The LSAP’s poster

Figure 6.9: The DP’s poster

Figure 6.10: The Greens’ poster

Figure 6.11: The CSV’s final poster

282

Figure 6.12: Posters of the UNEL-No Committee

Figure 6.13: Posters of the No Committee

283

This frame analysis was confirmed by the Yes and No campaigners. In contrast with the Spanish Yes campaigners, the Luxembourgish Yes campaigners mentioned the difficulty of the campaign and the misperception that they would easily win. This shows that the initial phase of the Luxembourg campaign followed the French and Dutch campaign patterns. The Yes campaigners frequently mentioned their surprise and stated that it was difficult to defend the treaty facing the No campaign arguments. Ben Fayot, MP and the President of the Parliamentary Group of the LSAP, explained that the political class was naively convinced that the campaign would be very easy given the highly pro-EU attitude of the country.745 Similarly, Charles Goerens, DP MP, stated that the Yes campaign was difficult as people were not familiar with the subject and that once they formed an opinion, it was very difficult to convince them.746 The Yes campaigners mentioned that the No campaign frames, which tapped into the fear of losing jobs and social standards, were successful. Without exception, the Yes campaigners acknowledged that Luxembourgish social standards were high, but added that society was concerned with unemployment referring to the particular geographic, demographic and institutional peculiarities of the country. Thus, the Yes campaigners explained that these arguments reached the workers and youth who were already afraid of losing their jobs. Specifically, the social concerns were mentioned as being very important for Luxembourgers who remained in the protected public jobs. Paralleling the French and Dutch Yes campaigners, the Luxembourgish Yes camp stressed the difficulty of explaining the social progress brought by the TCE in institutional terms. Fayot, LSAP MP, suggested that a ‘very heavy and technical argument’ had to be made to show people how every social step taken at the EU level was a compromise among different countries and how setting standards – even when they were not very high – was actually a major achievement.747 Similarly Goerens, DP MP, stated that it was ‘quite impossible’ to explain that there were several positive elements in the field of social policy in relation to the questions raised by the No

745 Personal interview, 14 November 2008. 746 Personal interview, 10 November 2008. 747 Personal interview, 14 November 2008. 284

campaign.748 Laurent Mosar, CSV MP, also stressed the difficulty of explaining the positive parts in the treaty and suggested that the No campaign arguments on enlargement and potential influx of workers matched Luxembourger’s existing fears of losing jobs to foreigners.749 These remarks show that the Luxembourg campaign followed the French and Dutch campaign patterns to a certain extent. Abbes Jacoby from The Greens stated that the No campaign was ‘argued on a very concrete level on social issues, and partly on national problems’.750 Frédéric Krier of the National Union of Luxembourgish Students (UNEL) – part of the No Committee – stated that the campaign for the Yes vote was basically a Yes to Europe, while the campaign for the No vote was a No against the treaty.751 While these campaign patterns did resemble the French and Dutch ones, the key difference was the shift in the Yes campaign after the double rejection. In June not only did all the Yes parties run a more forceful campaign, but the CSV also shifted its core framing. Moreover, Prime Minister Juncker started to campaign. François Biltgen, the Chairman of the CSV and the Minister of Labor and Employment, explained that the CSV changed strategies after the French and Dutch No votes, and started its own campaign:752 If a group of people, a school, an association, a circle of friends wanted to organize a meeting on the TCE and ask a party official’s involvement, this was made possible. There were various meetings where either the Prime Minister himself or another minister, MP or a specialist would attend to answer questions. Within this short period, the Prime Minister attended more than 60 public meetings and debates.753 Laurent Mosar, CSV MP, referred to their new strategy as running a more ‘aggressive’ campaign.754 Luxemburger Wort, the daily newspaper associated with the CSV, also intensified its Yes campaign, asking famous personalities in the country to

748 Personal interview, 10 November 2008. 749 Personal interview, 14 November 2008. 750 Personal interview, 10 April 2008. 751 Personal interview, 17 November 2008. 752 Personal interview, 12 November 2008. 753 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 72. 754 Personal interview, 14 November 2008. 285

indicate why they were in favor of the treaty.755 Readers were thus invited to identify with tennis players, bankers, directors, doctors, politicians, scientists, economists and so on. The party leaders and the Prime Minister strongly emphasized that the treaty was not dead despite the rejections, suggesting that Luxembourg was as important as France and the Netherlands, and that it had the same right to have a say.756 The Prime Minister repeatedly stated the risks reminding Luxembourgers that their country had benefited vastly from European integration:757 ‘Accept what we have contributed to building in the past and what we want to protect for the future’; ‘Nothing European must be strange to us’; ‘Luxembourg interests have always been European’; ‘Without Europe Luxembourg would be like a fish out of water’; ‘I shall invest all of my energy and determination into getting the Yes vote through in Luxembourg. I shall commit myself passionately to the Yes vote and reveal the populists’. Juncker explicitly warned that only a Yes vote could guarantee the defense of the country’s national interests when Europe would renegotiate.758 He also emphasized that the referendum was really about Luxembourg’s general attitude toward the European project as a whole rather than the constitution itself.759 Importantly, he frequently stated that Luxembourg’s national sovereignty on welfare and tax policies would be protected. He even declared – several times during the campaign – that he would resign if the TCE was rejected in the referendum. As such, the referendum campaign was dramatized and personalized.760 Both the Yes and No campaigners brought this shift up as a decisive moment in the campaign. Erna Hennicot-Schoepges, CSV MP, said ‘without the Juncker effect that was able to motivate the electorate in favor of the Yes (vote),

755 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 68. 756 Ibid. 757 Quoted in Deloy, "Luxembourg Says 'Yes' to the European Constitution."; ———, "Referendum on the European Constitution in Luxembourg 10th July 2005." 758 Quoted in Corinne Deloy, "Referendum on the European Constitution in Luxembourg: A Round up a Week Before the Election," Robert Schuman Foundation Report (2005). 759 Quoted in Ibid. 760 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 71. 286

the No (vote) would probably have won the day’, adding that ‘Luxembourg politicians were obliged to discuss certain topics and provide more real explanations about Europe’.761 On the other side, André Kremer, the Coordinator and Leader of the No Committee, explained this shift in June by referring to the Prime Minister’s ‘historic speech in the parliament’ and the new arguments that nationalized the debate, making it all about Luxembourg.762 At the same time, he added, the CSV became very active in mobilizing the society. Adrien Thomas from the No Committee/UNEL referred to Juncker’s significant influence in the following way:763 ‘… But then Juncker came in. The orientation of the campaign changed. … Juncker became very much involved and dramatized the stakes. Yes to the EU Constitution was good for Luxembourg, Luxembourg would be able to preserve its financial place and banking secrecy.’ Figure 6.14 shows a cartoon by the Luxembourg cartoonist Schneider, on the day of the referendum, which depicts the decisive role played by Prime Minister Juncker.764 The caption reads: Polling booth 1: Europe needs you! ‘Convinced Europeans?’

Figure 6.14: Cartoon on PM Juncker`s role in the Luxembourgish 2005 TCE campaign

761 Quoted in Deloy, "Luxembourg Says 'Yes' to the European Constitution." 762 Personal interview, 12 November 2008. 763 Personal interview, 15 April 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 764 Available at: http://www.ena.lu/ 287

Interestingly, similar to the French and Dutch cases, the No campaigners were more concrete in providing examples to demonstrate their arguments. They pointed to rising unemployment, the Bolkestein Directive, as well as various articles from the TCE. The Yes campaigners on the other hand commonly referred to the success of the banking/financial sector as a net benefit from the EU, stressing that it was made possible through the free flow of capital. Surprisingly banking secrecy, an important asset for Luxembourg’s economic development, was used by both campaigns as an example. The ADR argued that the TCE would jeopardize Luxembourg’s veto power on the subject as the EU was pushing for abolishing it. In contrast, the CSV argued that the veto power would be protected. Overall though, the No campaign frames were only partially effective for two reasons. First, as I will discuss in the diffusion section below, the left-wing No campaigners were strongly inspired by the French and they did not entirely fit the French left-wing argumentation to the Luxembourgish reality. The left-wing frames primarily emphasized the social concerns, referring also to the unemployment issue and the threat to public sector, which were the actual concerns in Luxembourg. Second, the right-wing No campaign frames – despite being strong frames – were not forcefully inserted into the debate as the ADR changed its position in late March. Besides, in the aftermath of the French and Dutch double rejection, the Yes campaign changed strategies and built stronger frames. In the next sections I will show that the credibility of the speakers and mobilizational structures further strengthened the Yes campaign.

Credibility of the Speakers Credible sources strengthen campaign framing. As I argued previously, in referendum campaigns, the impact of credibility can be analyzed by looking at three components: Government popularity, scapegoating the EU, and disagreements within the parties. Beginning with government popularity, as I have shown in detail above, Prime Minister Juncker was highly popular. Further, he is the longest ruling

288

European leader still in power.765 This can be interpreted as a factor increasing the strength of the Yes campaign framing as both the Prime Minister and his government were considered to be credible. Moreover, the Prime Minister put his career on the line. First in December 2004, then on several other occasions, he explicitly stated that he would resign if the treaty was rejected: ‘I signed after having negotiated the Constitutional Treaty on behalf of my country. If the citizens inform me that I should not have done this, I believe that I have to respect universal suffrage and the citizens, and suffer the consequences’.766 However, it is not clear whether this move had a positive or a negative effect on the outcome. As will be seen in the focus group data below, opposition to Juncker’s government received mixed responses, also coming up among the reasons given by the No voters.767 Towards the end of the campaign, a TNS-ILRES survey found that 45% of Luxembourgers considered Juncker’s move to be blackmail.768 Second, regarding the scapegoating of the EU in domestic politics, this habit compromises the speakers’ credibility in a Yes campaign as the positive frames conflict with the previous critical ones. In general, Luxembourg politicians did not share the French and Dutch politicians’ tendency to blame the Union for unpopular measures to achieve electoral gains. When interviewed, some mainstream campaigners accepted that there was occasional scapegoating. Nevertheless Fayot, LSAP MP, explained that it had always been clear that the Luxembourgish economy could not exist without a market and that the internal economic and budgetary problems were not systematically used against Europe by any of the mainstream political parties.769 Third, disagreements within the political parties lessen the credibility of the sources by emitting conflicting signals. In Luxembourg, the ADR was clearly split on the issue. Its leader, Gast Gybérien, changed his position on the TCE

765 Deloy, "Referendum on the European Constitution in Luxembourg 10th July 2005." 766 Quoted in ———, "Referendum on the European Constitution in Luxembourg: A Round up a Week Before the Election." 767 ———, "Referendum on the European Constitution in Luxembourg 10th July 2005." 768 Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." 769 Personal interview, 14 November 2008. 289

three times during the campaign.770 Gybérien had been a member of Luxembourg’s representation in the Convention process that drafted the TCE. Jacques-Yves Henckes, ADR MP, explained the multiple shifts during the campaign:771 The ADR initially declared that it would accept the result of the referendum. While it started out with a pro-TCE stance, it joined the No camp in late March. Yet when the referendum approved the TCE, it switched back to the Yes side to keep its promise to respect the result. Due to these shifts not only was the ADR’s credibility harmed but its No campaign frames were also not forcefully advocated. Similarly, Abbes Jacoby of The Greens explained that their party was divided because of the social policy concerns.772 This split parallels the divisions in the French and Dutch center-left political parties; nonetheless, the Luxembourgish Greens campaigned uniformly for a Yes vote.

Mobilizational Structures As the last factor, the literature finds that mobilizational structures also factor into the success of frames. I divide this concept into financial resources and civil society contribution. First, if campaigners have better financial resources, they can better support their campaign and advance their frames. In Luxembourg, just as in the other cases, the Yes campaign had better resources.773 The state did not provide any campaign funding to political parties. Only the meetings of the united parliamentary Yes campaign were financed by the state but there were no other funds made available to individual political parties. In effect, this led to disproportionate resources for the two sides. The left-wing No Committee, formed by civil society organizations, stated that it had very limited resources. Both the Yes and No campaigners confirmed this statement. Importantly, in clear contrast with France and the Netherlands, the Luxembourgish No campaign was not backed by a strong political party. The Left

770 Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." 771 Personal interview, 17 April 2008. 772 Personal interview, 10 November 2008. 773 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 15. 290

and the Communist Party did not have any parliamentary representation and lacked resources. The ADR on the other hand changed its position in late March and the ADR campaigners explicitly mentioned that they did not have a major campaign. As a result the No Committee ran its campaign with marginal resources. Second, the civil society contribution was limited, far less than the level of French mobilization. On the Yes side, although most of civil society organizations were in favor, there was not much of an active Yes campaign. It was only after the French and Dutch rejections, at the beginning of June, that 66 prominent members of civil society signed a joint statement providing a dozen reasons to vote in favor of the TCE.774 All unions and business associations and the two major trade unions the Confederation of Luxembourg Confederation of Christian Trade Unions (LCGB – close to the CSV) and the Confederation of Independent Trade Unions of Luxembourg (OGBL – close to the LSAP) were in favor of the TCE.775 However, Nico Clement of the OGBL stated that despite having a positive attitude they did not campaign for an unconditional Yes, and never asked voters for a Yes vote.776 Similarly, Nico Wennmacher, the President of the Railways Trade Union personally campaigned for the No side but this did not reflect an institutional position.777 On the No side, the left-wing No Committee was formed much earlier, in October 2004. André Kremer, the Coordinator and Leader of the No Committee, explained that it was an individual-based committee formed through Luxembourg Social Forum, by members of left-wing civil society institutions such as women’s organizations, movement of unemployment, the Union of Railways, the Greenpace, the National Union of Luxembourgish Students (UNEL), and also the political party The Left that had no representation in the parliament.778 This group was not as strong as the French No Committee. Adrien Thomas from the UNEL,

774 Personal interview with Pierre Gramegna (Director-General of the Chamber of Commerce), 13 November 2008. 775 Deloy, "Luxembourg Says 'Yes' to the European Constitution." 776 Personal interview, 17 November 2008. 777 Personal interview, 10 November 2008. 778 Personal interview, 12 November 2008. 291

who participated in both the French and Luxembourgish campaigns, stated that the campaign in France was more grassroots.779 The Yes campaigners frequently mentioned their surprise that this committee gained such ground in a short period time given their lack of resources. The No Committee obliged the political parties to undertake a real campaign in the field, something that is not usual in the Grand Duchy.780 Given the No campaign’s significant efforts, the interest in the subject increased over time, just as in France and the Netherlands. In the last month the TNS-ILRES surveys investigated the level of interest in the debate, showing that 82% of Luxembourgers found it interesting.781 Taken as a whole, the No campaign had limited success in building strong frames. The campaign started out very similarly to the French and Dutch ones. The left-wing No campaign came into the debate first and set its themes on the agenda. Indeed, social concerns became the dominant theme around February.782 The most detailed study on the subject also stresses that until the CSV and the Prime Minister nationalized the debate, the No Committee managed to control the agenda.783 ‘Social Europe’ had been the dominant theme of the campaign until mid-June, matching the anxieties of Luxembourgers to a certain extent. This is clearly visible in the problems The Greens and the LSAP had in persuading their followers to vote Yes, which I will show below.784 Yet, the No frames were only partially effective. First, they relied considerably on the French left-wing frames without adapting them strictly to the Luxembourgish society’s concerns. The left-wing No frames mostly stressed the welfare state-related concerns, mentioning unemployment as part of the ‘relocation of industry’ argument. They did not primarily tap into the contentious issues in the Luxembourg society such as fear of losing jobs and

779 Personal interview, 15 April 2008. 780 Deloy, "Luxembourg Says 'Yes' to the European Constitution." 781 The respondents could select more than one answer. Available at: http://www.tns-ilres.com/cms/Home/News/Publications/Archive/Referendum-vum-10-Juli-2005 782 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 43. 783 Ibid., 187. 784 Ibid. 292

cultural/linguistic cohesion due to the specific geographic, demographic, institutional characteristics of the country. Second, because the ADR changed its position during the campaign, it did not advance its strong right-wing frames vigorously. The Yes campaign was initially forced to take the defensive position and had difficulties in countering the No campaign frames. However in the aftermath of the French and Dutch double rejection, the Yes campaigners (primarily the CSV and the Prime Minister) changed strategy and built stronger frames stressing Luxembourgish national interests. The lack of credibility of the right-wing No campaign and the lack of resources of the left-wing No campaign further strengthened this revitalized Yes campaign. Laurent Mosar, CSV MP, explicitly stated that the Yes campaign was lucky because the No movement was not very well organized and lacked financial resources. He added: ‘I am afraid that if there will be another referendum in Luxembourg, this movement would be much better organized with much more money’.785

Analysis of the Vote Below, based on public opinion and focus group data, I show the extent to which the Yes and No campaign messages were picked up by the news media and echoed by the public. The data analysis below confirms that the No campaign framing was only partially effective and that the Yes campaign’s shift in strategy and framing indeed made a difference.

Shifts over Time As Figure 6.1 demonstrated at the outset, the No vote started increasing mid- March onwards, making the most progress between mid-April and mid-June. Figure 6.15 illustrates the progression of vote intentions from April-onwards in light of the important developments of the period:786

785 Personal interview, 14 November 2008. 786 Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." 293

Figure 6.15: Evolution of vote intentions in Luxembourg and campaign events

The shifts in vote intentions show that there is a visible increase in the No vote both before and after the French and Dutch negative votes. Therefore, while the Yes campaigners changed strategies only after the double rejections, the Luxembourgish No vote intention started to gain ground way before these rejections. Actually the French public opinion polls started to show a negative result since mid-March. These French polls could have had an impact on Luxembourg because the media closely followed the French campaign, as I will discuss below. The inconclusive mid-June European summit on financial matters was interpreted as further decreasing intentions to vote Yes.787 However, in the last phase, Figure 6.15 demonstrates containment of negative vote intentions. Prime Minister Juncker made a highly influential speech on 23 June, the National Day, referred to as a ‘historic speech’ by the No campaigners. Also, the parliament voted unanimously in favor of the TCE at the end of June, as the ADR chose not to attend.788 These key events as well as the revitalized Yes campaign

787 Ibid.; Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 72. 788 Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." 294

have indeed made a difference and have been successful in controlling the rise of negative opinion.

Media Content Analyses There is no detailed media content analysis of the Luxembourg referendum campaign. Yet, Luxembourg media is not similar to that of other cases. It has its own newspaper and television media but has only one national television channel – Luxembourg Television and Radio (RTL), which airs everyday but only from 6pm to 8pm. Because it receives the rest of its television channels from France and Germany directly, the French media content analysis accounts for the Luxembourgish case to a certain extent. Furthermore, I interviewed both the director of the national television – RTL, and director of the national radio. Concerning the tone of media coverage, Tom Graas, then director of the RTL, explained that the RTL had organized neutral debates between the Yes and No campaigners, and gave them equal coverage.789 Nonetheless Graas stated that he was under political pressure from the Yes campaign to decrease the No campaign coverage once the No vote started to pick up in the polls. The same pressure was mentioned by Marc Linster, then director of the national RTL Radio.790 Regarding topical coverage, Graas stressed the importance of the media coverage of the French campaign. He emphasized that not just the French rejection but also the French referendum campaign itself was covered largely in the media. Indeed, the number of articles published on the subject increased once French public opinion switched, showing a possible No vote.791 Heavy exposure to the French media means that Luxembourgers were familiar with the French campaign, whose central theme was social concerns – deterioration of the French welfare state and public services.

789 Personal interview, 17 November 2008. 790 Phone interview, 11 November 2008. 791 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 55. 295

Public Opinion Data Table 6.4, based on the Eurobarometer post-referendum survey, shows the breakdown of the Yes/No vote by party proximity:792

Yes No voters voters (%) (%) CSV 79 21 LSAP 49 49 DP 55 42 The Greens 48 49 ADR 23 77 The Left 10 86 Table 6.4: Breakdown of the Luxembourgish Yes/No vote by party proximity

This breakdown demonstrates that the CSV was indeed the most successful in terms of persuading its followers, while the LSAP and The Greens faced significant problems. It was mainly the social concerns that conditioned the voting intention among the mainstream, indicating that the left-wing No campaign had some success in priming its core theme. As expected, among the No campaign political parties, The Left scores better than the ADR. That said, the small size of The Left supporters should be kept in mind. In general Table 6.4 illustrates that the Luxembourg case was similar to the French and Dutch campaigns despite being a positive case, as it was primarily the center-left voters who did not follow the voting advice of their parties. Specifically, the TNS-ILRES post-referendum survey asked the voters about their motivations for voting Yes/No.793 Each voter was allowed to choose more than one answer. Table 6.5 presents the answers of the Yes voters:

792 The remaining percentages reflect ‘Blank’ or ‘Don’t Know’ categories. Eurobarometer, "The European Constitution: Post-referendum Survey in Luxembourg." 793 Available at: http://www.tns-ilres.com/cms/Home/News/Publications/Archive/Referendum-vum-10-Juli-2005 296

Reasons to Vote Yes (%) Place of Luxembourg in Europe/Good for Luxembourg 88 The EU will be stronger vis-à-vis the USA, China and Russia 81 To further the European project 81 Step towards political unity in Europe 80 Enhancing the functioning of the EU institutions 63 My political party supports a Yes vote 32 The Prime Minister’s resignation 22 People’s recommendations 15 Table 6.5: Reasons to vote Yes in Luxembourg

The responses show that the main reasons to vote Yes were not particularly related to the TCE, paralleling the other three referendum cases. Table 6.5 shows that an overwhelming majority of Luxembourgers echoed the fifth Yes frame – the key idea that what was good for Europe was good for Luxembourg. This ranking confirms the framing analysis above. The fifth frame was explicitly and forcefully used by the CSV and the Prime Minister after the French and Dutch rejections, and it was strong as it reminded the voters that Luxembourg owed its progress to Europe and that a No vote would be a grave mistake. The EU’s power in the world, furthering the integration project, and European political unity all reflect broad support for the Union rather than specific references to the content of the treaty. Improvement of the EU institutions is the only response that is directly connected to the TCE. This answer reflects the second Yes frame which highlighted the institutional advances brought on by the treaty, which was a weak frame. It is surprising as 63% of the public selected this statement, yet this answer does not rank high among the given reasons. As for the rest of the Yes campaign frames, following the framing analysis above, the first Yes frame on peace was not strong. The third Yes frame specifically on social progress was indeed not echoed by the public, as there was a stronger No campaign frame on the subject – as well as the heavy exposure to the French campaign which I discuss in detail below. Similarly, the fourth Yes frame on the consolidation of small states’ power was not stated either. But this frame could be partly reflected in the first survey

297

answer as it refers to the beneficial position of Luxembourg. Finally, the Prime Minister’s resignation does not seem to have motivated an important part of the Yes voters. Taken as a whole, the Yes motivations show that it was only the fifth Yes campaign frame which fared extremely well and was overwhelmingly echoed by the public, confirming the campaign analysis above. This particular frame was used by the Yes campaign after the French and Dutch rejections to nationalize the debate and raise the stakes by emphasizing Luxembourg’s national interest in saying Yes. The results clearly demonstrate that this strategy worked and successfully contained the No vote intention in June, as shown in Figures 6.1 and 6.15. Table 6.6 lists the results for the No voters: Reasons to Vote No (%) Enlargement 68 The TCE should be renegotiated 65 No trust in the development of the EU 61 The text is too complicated 59 The TCE is too liberal 39 Turkey 39 The TCE does not push the European integration far enough 33 I had enough of the EU 29 Table 6.6: Reasons to vote No in Luxembourg

The reasons provided by the No voters generally confirm the framing analysis. The second, third and fourth No frames were weak indeed. The public did not refer to the undemocratic, militarizing or anti-environmentalist character of the TCE, just as in the other three referendum cases. The primary reason for rejecting the treaty was enlargement, by 68%. This answer initially seems to reflect the sixth frame (used by the ADR) which opposed the treaty because of the enlargement process and Turkey’s potential accession. However, Turkey’s membership does not rank particularly high among the answers. A possible explanation could be that the answer ‘enlargement’ reflected a broader phenomenon that matched Luxembourg’s fears of losing jobs due to relocation of industry to new members with more favorable taxing conditions. The

298

Eurobarometer post-referendum survey approves this assessment, as in this study, the top two No vote motivations were ‘the negative effects on the employment situation in Luxembourg/ relocation of Luxembourg’s enterprises/ loss of jobs/ I am against the Bolkestein directive’ and ‘the economic situation in Luxembourg is too weak/ there is too much unemployment in Luxembourg’.794 Data from the focus group study below will also confirm this assessment. This means that the Luxembourgish No voters referred to enlargement primarily in relation to the relocation of industry, unemployment, and fear of losing jobs, and not as much in relation to the Turkish accession. As for the other No campaign frames, the first frame emphasizing the loss of social standards was echoed by 39%, which is not a very high percentage. This demonstrates that the borrowed Luxembourg left-wing No campaign’s frames were not greatly effective on their own. Nonetheless, as the top ranking ‘enlargement’ response was linked with the relocation of industry and unemployment, the first No frame highlighting competition for Luxembourgish jobs as a result of social/fiscal dumping and liberalization of public services could also be partly reflected in this answer. Taken together, these answers show the partial activation of latent fears in the society by these borrowed frames. The first and sixth No frames, combined, are echoed by an important portion of the No voters. On the other hand, the fifth frame arguing against the TCE on grounds of losing sovereignty did not come up in the answers, despite being a strong frame. Even so, the Eurobarometer survey found that 9% of the No voters cited loss of national sovereignty as a reason.795 Interestingly, 33% of the No voters rejected the TCE because it was not advancing integration enough. In striking contrast, 61% – a remarkably high percentage –stated lack of trust in the EU, which is highly unexpected given the positive attitude of Luxembourgers in regular Eurobarometer surveys. In general, the No voters seem to be motivated primarily by enlargement and social concerns, connected by fear of losing jobs to new member states due to relocation of

794 Eurobarometer, "The European Constitution: Post-referendum Survey in Luxembourg." 795 Ibid. 299

industry and social dumping. The answers confirm the frame analysis with the exception of the fifth No frame on losing sovereignty. However, the variety of the answers provided and the lack of a clear and plain theme among them shows that the No campaign only had limited success. The match between the No campaign frames and the reasons behind the No vote is not as direct as in the French and Dutch cases. Under the leadership of Patrick Dumont, Fernand Fehlen, Raphaël Kies, and Philippe Poirier, the University of Luxembourg has carried out a detailed analysis of the voting behavior for the Luxembourg parliament.796 This project involved a detailed focus group study, which highlighted four main concerns that were decisive in the final vote: Social cohesion, employment, education, and enlargement. The study finds that these issues were not built around the categories and policies specified in the TCE, which once again shows that the institutional and technical content of the text did not capture the public’s attention. Beginning with social cohesion, the research showed that both culturally and economically, this was a primary concern for Luxembourgers regardless of their education level. Most respondents mentioned fears of decreasing solidarity and cohesion in Luxembourgish society due to increasing immigration. This was coupled with the concern of becoming foreigners in their own country due to the increasing use of French instead of Lëtzebuergesch. Economically, Luxembourgers stressed the importance of maintaining the Luxembourgish social model in the face of increasing foreign labor. Instead of referring to the Charter of Fundamental Rights incorporated into the TCE, they emphasized the necessity of having national public institutions to ensure protection against economic globalization.

796 This study uses detailed focus group data collected by the TNS-ILRES, qualitative data on the campaign strategies and speeches of the political and economic actors, and Eurobarometer survey data on the motivations behind Yes/no votes. Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés. 300

Second, concerning employment, most Luxembourgers brought up the potential loss of national autonomy in the matter. Some expected relocation of industry to new member states resulting in further unemployment. Third, education was linked to the same set of fears. Luxembourgers advocated better school education to control the need for foreigners in the job market, particularly in the service sector. They referred to their worries concerning the future of their children. Finally, enlargement came up as a concern regardless of respondents’ education or profession. What was referred to by enlargement was not only Turkey and the potential new members but also the recent 2004 round which brought ten new Central and Eastern European states in. The main criticism of enlargement was based on fear of competition for jobs, referring to the relocation of industry and outsourcing. Particularly on the question of Turkey, they stated two problems. The first was once again related to its size and economy, going back to the same fear of competition for jobs. The second problem was a cultural one; that it did not fit in. However, importantly, nobody in the focus groups stated Turkey as the explicit reason to vote against the TCE. The specification of these four key concerns, and the way they are elaborated matches the survey data analysis above. Enlargement was indeed primarily referred to as an economic concern, in relation to the fear of losing jobs due to relocation of industry to the recently admitted and future member states. Yet, this study also marks contentious issues unreached by the No campaigners, for instance the need for a better education system to decrease competition with foreigners for jobs, or fear of losing cultural and linguistic cohesion. Lastly, the focus group study also confirmed mixed reactions to Juncker’s resignation statement. Overall, the public opinion data shows that none of the frames were as clearly successful as the fifth Yes campaign frame which stressed the Luxembourgish national interest in the European integration, brought in by the CSV and the Prime Minister in the last phase of the campaign. Nevertheless, the left-wing No campaign dominated the campaign agenda and placed an ‘anti-social

301

TCE’ frame at the core of the campaign until the CSV and Prime Minister joined the campaign and nationalized the debate. The problems faced by the center-left political parties in persuading their followers serve as an indicator of this.797 This social theme matched Luxembourgers’ anxieties to a certain extent, not through the deterioration of the welfare state as in France, but instead through the relocation of industry and unemployment concerns. Compared to France, this frame was not as successful in Luxembourg. An important reason could be that the left-wing No campaign decidedly used the French left-wing No campaign framing, and did not build immediately on the central and peculiarly defined contentious issues of Luxembourg such as unemployment and immigration, based on fear of losing jobs and cultural cohesion. However, they still indirectly touched upon some of these nerves and activated these latent fears. The survey answers show that Luxembourgers essentially picked up the parts of the arguments that matched their concerns the best. Furthermore, as the diffusion section will show below, heavy exposure to the French campaign also enhanced the strength of the No campaign frames. The data also confirms that the right-wing No campaign by the ADR was not as successful. Their particular emphasis on the Turkish accession or the loss of national sovereignty captured the public’s attention only to a small extent. This is not surprising as the ADR had changed its position during the campaign and did not forcefully advance its frames in the debate. It is striking to see how the Luxembourg vote, which is categorized as a positive case by both the academic literature and the policy world, was actually quite similar to the French and Dutch campaigns. Just like them, there was an episodic dip in public opinion. The fall 2005 Eurobarometer surveys found the public support for the EU restored to the usual highly positive level.798 What was unique to the Luxembourg case was the diffusion dynamics, which I discuss next.

797 Ibid., 187. 798 Eurobarometer, "Standard Eurobarometer 64: Public Opinion in the EU, Autumn 2005." 302

Diffusion On 10 July 2005, a little over a month after the French-Dutch rejections, Luxembourg became the last country to hold a referendum. I argue that the later a country held its referendum relative to other states, the more the previous referenda results and campaigns in the other states influenced its campaign dynamics and public. The Luxembourg case is the most important case to analyze in this regard as it came last. Indeed, it was the only case among the four where diffusion was substantial. Yet it was not influenced equally by the three previous referendum cases. The key question is: Why was this case influenced by the French but not by the other two referenda? I argue that cross-case influences are conditional, and depend on the existence of transmission belts between the states such as shared language/culture, common media sources, and collaborative networks/transnational linkages. The more that these channels are open, the more the later cases can be influenced by previous ones. I will show that Luxembourg’s peculiar geographical and institutional set up created several important diffusion channels with only one of them: France. Further, I propose that campaign intensity mattered. As I have discussed in great detail, the French case was the momentum case with the most intense campaign of all and thereby put significant amount of pressure on the existing diffusion channels. How can this diffusion effect be empirically documented? First, the interview data clearly proves that the main No campaigner, the Luxembourgish left-wing No campaign, followed the French left-wing No campaign as a model, and benefited from it both in terms campaign arguments and logistics. On the other hand, the actual influence on the vote is very difficult to trace empirically. Nevertheless, there are some empirical indicators that point to the French campaign’s influence on Luxembourgers, in the public opinion and focus group data. In this section I will show that the explanation provided by the independent variable – relative No campaign strength – is not complete without paying attention to the diffusion process. First, the very existence of the No campaign was dependent on the support of the French campaign. Second, the key shift in the

303

Yes campaign strategy only took place after the French and Dutch rejections increased the stakes. Thus diffusion conditioned the impact of the independent variable and directly amplified the effectiveness of the campaigns. Below I first discuss the existence of the diffusion channels. Based on interview data this analysis will show the considerable support the No campaign received from the French No campaign. Next, based on public opinion data, I discuss the empirical evidence pointing to the limited diffusion effect on the actual vote. The first diffusion channel is the shared language and culture. Luxembourg had a close connection with France linguistically and culturally, which it did not have with Spain or the Netherlands. Surveys on the use of language in Luxembourg show French as the best known language.799 French serves as the oral code linking different communities in Luxembourg, while Lëtzebuergesch is the dominant language used among Luxembourgers. My interviewees mentioned this connection in relation to various campaign decisions such as inviting speakers or selecting campaign materials. To start with the invited speakers, both the Yes and No campaigners stated their preference to invite French speakers given the connection between the two countries. The Yes campaigners invited the former French Prime Minister Michel Rocard and the Franco-German Greens MEP, Daniel Cohn-Bendit. The left-wing No campaigners similarly invited highly influential left-wing French politicians such as Henri Emmanuelli, José Bové, Yves Salesse, Raoul-Marc Jennar, and Francine Bavay. In fact, Luxembourgish social democrats protested the invitation of these French No campaigners. The LSAP’s chairman, Alex Bodry, stated that it was ‘inadmissible for a leader from the French Socialist Party to come to Luxembourg to defend a position that was contrary to that democratically adopted by the Luxembourg Socialists’.800 In contrast only one Dutch No campaigner, Erik Wesselius, came as a representative of the Dutch No Committee.

799 Stell, Luxembourgish Standardization: Context, Ideology and Comparability with the Case of West Frisian, 29. 800 Quoted in Deloy, "Referendum on the European Constitution in Luxembourg: A Round up a Week Before the Election." 304

The Luxembourgish left-wing No campaigners openly acknowledged the support they received from the French left-wing No campaign. This support was both ideational and material. They explicitly mentioned the difficulty of organizing the campaign without such support, as Luxembourg has a very small community. Therefore the French campaign’s support was crucial for the very existence of Luxembourg’s main No campaigner, as this small country did not have the French mobilizing potential in terms of civil society structure or resources. First in terms of ideas, Adrien Thomas from the UNEL explained that the No Committee was heavily influenced by the French debate. He stressed that Luxembourg did not have a tradition of following the Dutch and that traditionally, French culture was influential.801 Similarly, Henri Wehenkel of The Left stated that the Netherlands was ‘far’ and stressed the linguistic and cultural links to France.802 He added: I went to Paris, to bring the propaganda that was unused after the French referendum. I brought 3,000 texts of the Constitution with the explanation and examples, from the French Communist Party. At the same time, the No Committee brought many famous French speakers, as French politicians are well known here.

French campaigners also confirmed that their unused No campaign materials (leaflets and posters in French) were delivered to Luxembourg to be used in their No campaign. Moreover another No campaigner, the President of the Railways Trade Union, Nico Wennmacher, explained that he was following the French trade unionists’ arguments on the subject as they worked together in the Grand Region.803 Second, in terms of material support, the ATTAC France has financially contributed to the ATTAC Luxembourg’s campaign to a small extent.804

801 Personal interview, 15 April 2008. 802 Personal interview, 15 April 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 803 The Grand Region is the area combining Luxembourg with neighboring parts of France, Germany and Belgium. Personal interview, 10 November 2008. 804 Personal interview with Pierre Khalfa from ATTAC France, 22 September 2008. 305

In general, both the Yes and No campaigners commonly referred to the influence of France as being more important than the other cases because of the cultural and linguistic links. Henckes, MP of the ADR, said that their No campaign benefited from the French campaign as Luxembourgers followed the French debate, adding that ‘people did not follow the Dutch debate as there was No language connection’.805 Goerens, DP MP, also explained that although the Netherlands was their first ally after the Second World War, their culture was closer to France and that people watched French television, finding Dutch less understandable.806 Similarly, LSAP MP Fayot highlighted the importance of the French No campaign websites which were followed closely by Luxembourgers.807 The second diffusion channel is common media channels. Luxembourgers receive French television channels and newspapers on a daily basis. They do not have a similar connection to Spain or the Netherlands. Regarding television, the local broadcaster RTL operates six channels, but only one in Lëtzebuergesch.808 As I explained above, this national television channel – RTL airs only from 6pm to 8pm. The rest of the television channels are received directly from France and Germany. Tom Graas, then director of the national RTL TV News, stressed that not only the results and aftermath of the French referendum but also the French referendum campaign itself was covered remarkably in the media: ‘I think that if we had had another referendum before the French, we would have seen a Yes of 80%.’809 Both the Yes and No campaigners frequently mentioned that the voters were exposed to arguments from France due to common television channels and newspapers. They stated that the main background information they received was from other countries because of the limited air time of the Luxembourgish TV channel. Biltgen, CSV MP and the Minister of Labor and Employment, stated that the referendum debate in Luxembourg became largely run by foreign press due to

805 Personal interview, 17 April 2008. 806 Personal interview, 10 November 2008. 807 Personal interview, 14 November 2008. 808 Stell, Luxembourgish Standardization: Context, Ideology and Comparability with the Case of West Frisian, 30. 809 Personal interview, 17 November 2008. 306

the French influence.810 While this might be a common problem in a globalizing world, Luxembourg’s peculiar media set up has magnified the impact of the foreign press. The third and the final channel, collaborative networks and transnational linkages, was crucial yet not confined to France-Luxembourg. Below I assess the impact of institutional networks and personal connections by focusing on the European Parliament groups, the European anti-globalization network, ad hoc European networks, and existence of mobile communities. While all four cases shared the institutional networks, only France and Luxembourg had personal connections. To start with the institutional networks, just as the previous cases, Luxembourg was part of the institutional structure bringing together all European countries: the EP groups and the European anti-globalization network. Political parties in Luxembourg are well-integrated into the EP groups. All campaigners from political parties explained that there were discussions with their EP party groups which included similar Spanish, French and Dutch political parties. Similarly to what was stated in the other cases, the campaigners brought up occasional meetings in which they shared arguments with their counterparts but also emphasized that the campaigns were run separately. Once again, paralleling the other cases, the far left No campaigners mentioned stronger collaboration. Another key institutional network is the European anti-globalization network. Left-wing civil society organizations in all four referendum cases were connected via this European network. Just as in the other countries, the ATTAC and European Social Forum meetings (specifically the 2003 ESF in Paris, and the 2004 ESF in London) were brought up as significant platforms where the TCE was debated and analyzed before the campaigns started. In all cases, the ATTAC France and French left-wing theoretical analysis of the TCE were crucial sources. For Luxembourg, this impact was amplified by the special cultural and linguistic connection between the two countries. As I explained in the previous paragraphs, the ATTAC Luxembourg and left-wing No campaign benefited from the ATTAC

810 Personal interview, 12 November 2008. 307

France in terms of campaign arguments, campaign materials, invited speakers, and to a small extent, financial resources. André Kremer of the No Committee stated that in the formulation of their arguments, they were inspired by and benefited from their French contacts’ intellectual work. Particularly, brochures and leaflets developed by the French Communist Party, the ATTAC France, and Raul Marc Jennar were used in the Luxembourg No campaign meetings.811 Anne-Marie Berny from the ATTAC- LUX explained that most of the ATTAC-LUX members were French living in Luxembourg.812 While she mentioned that the ATTAC France, Germany and Belgium helped them, she stressed that that they have been in regular contact with the ATTAC France throughout the campaign. Adrien Thomas from the UNEL explained that some ATTAC France members even came to Luxembourg to help distribute flyers in mailboxes.813 Therefore the anti-globalization network in Europe has spread the arguments which originated in the Paris ESF and the ATTAC meetings and formed a left-wing master frame, advocating an alternative ‘social’ EU. Among the three cases, this frame was particularly influential on the Luxembourgish left- wing No campaigners due to the special cultural and linguistic connection between the two countries. However, a key issue with this master frame was the adaptation to local reality. While the Dutch far left political party strategically sought to fit this master frame to its national framework, Spanish and Luxembourgish far left campaigners did not do so. Even though the Luxembourgish No campaigners benefited significantly from the French No campaign, borrowing frames worked both positively and negatively. Given the small size of the country, their lack of resources and the absence of a supporting political party, the Luxembourgish left-wing campaigners mentioned the difficulty of building a No campaign from scratch without the support of the French No campaign. Hence the overall impact of adopting the French left-wing No frames should be seen as strengthening the campaign. But the degree to which

811 Personal interview, 12 November 2008. 812 Personal interview, 17 April 2008. 813 Personal interview, 15 April 2008. 308

these borrowed arguments fit the local reality matters. The Luxembourg No campaign chose to keep this master frame intact. In short, they did not shape them to stress the peculiar contentious issues of the Luxembourgish society to make them stronger. This criticism was voiced by several analysts.814 Yet, the fact that the public was highly exposed to the French campaign through diffusion channels such as media channels and personal connections might have worked in favor of the Luxembourg No campaigners as the public was confused on French and Luxembourgish problems, as I will explain below. The second aspect within the third channel is personal connections. As discussed at the outset, a third of Luxembourg’s permanent population is composed of citizens of other countries. 2004 census data shows that among the foreign residents of Luxembourg French forms the second largest group after the Portuguese.815 Moreover, Luxembourg’s peculiar geographic, demographic and institutional set up created a large mobile community between Luxembourg and France. None of the other cases shared a similar personal connection. Luxembourg and France were connected by two particularly mobile communities: Cross-border employees and students. I discuss each in turn. Regarding cross-border employment, the European Employment Services (EURES) data shows that in 2005 around 118,385 cross-border employees came into Luxembourg on a daily basis from France, Germany and Belgium.816 The first line of Table 6.7 shows the number of cross-border employees who leave their country of residence on a daily basis to work in Luxembourg.817

814 Personal interviews with Jürgen Stoldt, political communication expert and Editor of Forum, on 11 November 2008, also see Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 187. 815 Denis Scuto, "EUDO Citizenship Observatory Country Report: Luxembourg," European University Institute Robert Schuman Center for Advanced Studies (2009). 816 European Employment Services (EURES), "Les Cahiers Transfrontaliers d’EURES." 817 The figures reflect the situation in 2004 and 2005. Ibid. 309

Table 6.7: Number of incoming cross-border employees

This data demonstrates that more than 50% of these cross-border employees come from France. As can be seen, Dutch workers do not share a similar position in Luxembourg. The reverse situation can be seen in the first line of Table 6.8 presenting the number of Luxembourgers who leave Luxembourg on a daily basis to work in the neighboring countries.

Table 6.8: Number of outgoing cross-border employees

310

These figures show that, despite in much smaller numbers compared to the incoming employees, there are Luxembourgers who do the opposite and go to France to work on a daily basis. Alternatively, the EURES has also studied the cross-border aspect of temporary jobs in Luxembourg.818 Figure 6.16 below shows the increase in number of temporary employees in Luxembourg over time:819

Figure 6.16: Number of temporary employees in Luxembourg over time

Importantly, this research found that among these employees, the majority was once again from France. Figure 6.17 and Table 6.9 below show the extent to which French workers dominate the temporary workforce in Luxembourg:

818 Franz Clement, "Le Travail Intérimaire au Luxembourg: Un Régulateur du Marché du Travail dans la Grande Région?," European Employment Services (EURES)(2006), http://www.eureslux.org/eures.taf?idnav=136. 819 Ibid. The fluctuation is due to cyclical and seasonal temporary work. 311

Figure 6.17: Breakdown of Luxembourg temporary workforce according to nationality

Table 6.9: Breakdown of Luxembourg temporary workforce according to nationality

Taken as a whole, the data makes it clear that the French dominate the cross- border worker population in Luxembourg. These various groups are in constant interaction with Luxembourgers on a daily basis. The second community responsible for personal connections is the mobile students. Luxembourg’s higher education has been embryonic both in structure and objective; as a result, Luxembourgish students have had to pursue their studies abroad, mainly in France, Germany and Belgium.820 The first university in

820 The statement does not apply to students who train as primary school teachers. For a detailed discussion on student mobility and the reasons behind this particular policy see Germain 312

Luxembourg, the University of Luxembourg, was founded only in 2003. Thus the number of students registered at foreign universities and those registered in France rose over time.821 Krier from the UNEL explained that they had members who studied in the neighboring countries and that specifically, during the campaign, those in France were very important in organizing the Luxembourgish campaign.822 Adrien Thomas from the UNEL who was a doctoral student at Sorbonne took part in both campaigns and explained that most students came back to Luxembourg, before the referendum, after classes were over.823 Consequently, a lot of the Luxembourgish students who were actively involved in the French campaign were also involved in the Luxembourgish campaign. Thomas explained that the No Committee was also inspired by the French through the ATTAC Campus network:824 The UNEL had its main base in Paris in 2005. … We attended the ATTAC Campus meetings, where French members of ATTAC in Paris exposed their point of view on the Constitution. We took notes and building on those notes and on articles we read in the left-leaning newspapers, we wrote articles in Luxembourg. We did not take the Constitution and interpret it for ourselves. We were very much inspired.

Thus, Luxembourg’s peculiar conditions created not only a mobile group of cross- border employees but also a mobile student community. These groups connected France and Luxembourg on a personal level, carrying the influence back and forth in their daily interactions. Tom Graas, then director of RTL TV News, explained how the French commuters had a big influence as the discussions were carried everywhere from offices to restaurants.825 This discussion of diffusion channels establishes that France and Luxembourg had a significant connection. While the interview data I presented clearly shows that the French left-wing No campaign served as a model for the

Pondelinger, "Country Report: ELA – Luxembourg," European Journal for Education Law and Policy 3 (1999). 821 Gerald Newton, ed. Luxembourg and Lëtzebuergesch: Language and Communication at the Crossroads of Europe (Oxford: Oxford University Press,1996), 29. 822 Personal interview, 17 November 2008. 823 Personal interview, 15 April 2008. 824 Personal interview, 15 April 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 825 Personal interview, 17 November 2008. 313

Luxembourgish left-wing No campaign, the question of the diffusion effect on the public opinion remains: How can the impact on the actual voting behavior be empirically shown? There are no direct survey questions which asked Luxembourgers the extent to which they were influenced by the Spanish, French and the Dutch votes. A direct question as such may not even be fully reliable as it would be likely for any country’s nationals to reject the influence of another country on their vote. But there is some indirect empirical evidence signaling the French campaign’s influence on Luxembourgers’ vote choice. First of all, campaigners and researchers agree that diffusion indeed had an impact. Philippe Poirier, one of the most prominent researchers on the referendum, stated early in the campaign that the French and Dutch referenda could have a substantial effect on public attitude in Luxembourg and that a Yes vote below 60% would have to be interpreted as very weak for traditionally Europhile Luxembourg.826 It was common for the Yes campaigners to refer to this influence as contamination, and for the No campaigners as support or inspiration. One way to look into this impact is the shift in vote intentions. As I discussed above, Figures 6.1 and 6.15 demonstrate that there is a visible increase in the No vote before and after the French and Dutch negative votes. French public opinion started to reject the TCE since mid-March. Given the extensive media coverage, the increase in Luxembourg No vote intentions could be seen as related to the French campaign and the subsequent double rejection. Moreover Table 6.10, based on the University of Luxembourg study, shows the cross-tabulation of time of vote decision during the campaign and the Yes/No vote:827

826 Quoted in Deloy, "Referendum on the European Constitution in Luxembourg: A Round up a Week Before the Election." 827 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 127. 314

Time of vote decision during the campaign Yes No voters voters (%) (%) The moment referendum was announced 32.2 25 Early in the campaign 20.9 24.5 After the French and Dutch rejections 3.4 3.9 During the last weeks of the campaign 20 26.5 The last week of the campaign 18.3 14.7 The day of referendum 5.1 5.1 Table 6.10: Timing of vote decision in Luxembourg – the Yes/No breakdown

Based on this data, around half of Luxembourgers took their decision in the last phase of the campaign, after the French and Dutch rejections. In short, most Luxembourgers made their decisions rather late. Another way of analyzing the impact is to look at the geographical distribution of the negative vote. Of the nine communes of Luxembourg that voted against the TCE, seven of them are located right at the border with France, as can be seen in Figure 6.18:828

Figure 6.18: Geographical (communal) distribution of the Luxembourgish no vote

828 Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." Detailed data on the voting pattern of each commune is available at: http://www.verfassung-fir-europa.lu/fr/referendum/resultat/index.html 315

While at first glance this seems to be the perfect evidence of French influence on Luxembourgers, the socioeconomic profile of this region makes the connection tricky. The south of Luxembourg is traditionally working class, what used to be the heart of mining and metallurgical industry, and today remains the bastion of left-wing parties.829 In all four referendum countries, the socioeconomic breakdown of the No vote shows that the manual workers tended to vote negatively. Specifically, the Eurobarometer post-referendum study in Luxembourg shows that 67% of the manual workers voted No.830 Thus, socioeconomic status could be an alternative explanation for the geographical concentration of the No vote along the French border. However, the two factors are mutually reinforcing. The overall tone of the French campaign, which I discussed in great detail in the French case chapter, was primarily left-wing, centering on the deterioration of the welfare system and social concerns in general. This particular framing could be expected to have a greater impact on the working class. Therefore, while it is impossible to separate the two factors causally, the empirical evidence could be interpreted to show some influence of the French campaign. To clarify, the No voters in Luxembourg were not limited to the working class. There were people from every commune that voted negatively. Figure 6.19 marks the areas according to the No vote percentages:831

829 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 83. 830 Eurobarometer, "The European Constitution: Post-referendum Survey in Luxembourg." 831 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés, 83. 316

Figure 6.19: Geographical (percentile) distribution of the Luxembourgish no vote

A highly interesting piece of empirical evidence comes from the focus group study carried out by the TNS-ILRES for the University of Luxembourg and the Luxembourg parliament. During the focus group meetings, where people were asked to elaborate on their reasons to vote positively or negatively, Luxembourgers confused Luxembourgish unemployment figures with French figures.832 This demonstrates that Luxembourgers were heavily subjected to French background information due to the diffusion channels. The following quote by Laurent Mosar, CSV MP, summarizes the same issue in relation to the

832 Personal interviews with Dr. Philippe Poirier on 17 April 2008, and Charles Margue – Director of Studies of the TNS-ILRES on 18 November 2008. 317

No campaign argument stressing the need to prevent the liberalization of public services:833 … Luxembourgers are mostly employed in the public services. That was also an argument here, but perhaps not like in France. It was not a main argument. … The problem was that Luxembourgers watch French and German TV, especially French TV. Those arguments used in France were finally also used in Luxembourg. The situations in France and Luxembourg are different but people watch TV, hear something on liberalization of public services and think that this is happening in Luxembourg too.

This quote also shows how the campaigners had difficulty in limiting the campaign debate to arguments originating from Luxembourg. While the Luxembourgish No campaigners transplanted the French campaign’s arguments directly into the Luxembourg campaign without fitting them fully to the national context, the fact that the public was highly exposed to the French campaign might have lessened this problem. It appears that through media channels and personal connections, the public conflated French concerns with Luxembourgish ones. An interesting spin on the diffusion argument is that Luxembourg’s Yes campaign changed strategies because of diffusion, since they only did so after the French-Dutch double rejection. As I have shown in detail above, the switch in Yes campaign’s framing was a key factor in the sustainment of positive attitude. This influence was not based particularly on diffusion channels; it was rather based on sequencing. The rejections raised the stakes for the Luxembourgish Yes campaign. The French and Dutch Yes campaigners commonly referred to their underestimation of the No campaigns, which was exactly what Luxembourgish Yes campaigners repeated for the initial phase of the campaign. It was the two consecutive rejections, only a few days apart from one another, that made Luxembourgish Yes campaigners realize the potential consequences. While sequencing is very much part of the diffusion logic, this effect was less substantial than the diffusion observed in the left-wing No campaign.

833 Personal interview, 14 November 2008. I made minor changes in the grammatical structure of this quotation for clarity purposes. 318

Therefore, the explanation provided by the independent variable – relative No campaign strength – is incomplete without paying attention to the diffusion process. Diffusion conditioned the impact of the independent variable significantly and amplified campaign strength. The interview data clearly demonstrate that the French and the Luxembourgish left-wing No campaigners were closely connected, to a degree much higher than among the Yes campaigners. As such, the Luxembourgish left-wing No campaign – the main No campaign – benefited from the French No campaign both in terms of framing and logistics. Also, the Yes campaign’s shift in strategy could be interpreted as part of the diffusion logic. There might not be sufficient empirical evidence to show the direct influence of the French referendum campaign on the Luxembourgish voting behavior but the use of a counterfactual makes it clearer. Given the diffusion analysis above, it is hard to disagree that had they gone first or had the French and the Dutch approved, the Luxembourgish campaign would have unfolded much differently. As such, the Luxembourg case was indeed the diffusion case.

The Diffusion Campaign The Grand Duchy’s positive vote was welcomed enthusiastically by European leaders: ‘The people of Luxembourg can be proud of the positive result of the referendum. Yes to the European Constitution is also a declaration of an acceptance of Europe when the Union is experiencing a difficult moment’, declared German Chancellor Gerhard Schröder.834 Similarly, Prime Minister Juncker announced: ‘Luxembourg voted with a clear head. Luxembourg said Yes to the constitutional treaty. You can understand why I am happy. It is the expression of universal suffrage by a small population but a great nation. Consequently all the importance that was granted to the expression of the majority of the French and the Dutch is returned to the Luxembourg election’.835 Actually, the vote was neither strongly positive nor conducted with a ‘clear head’.

834 Quoted in Deloy, "Luxembourg Says 'Yes' to the European Constitution." 835 Quoted in Ibid. 319

Nonetheless, both the policy world and the academic literature simply categorized the Luxembourgish vote as a clear Yes vote. This chapter argued against this misperception: 43% of Luxembourgers, one of the most Europhile nations in the EU, voted negatively. Luxembourg was not able to maintain its early positive attitude as successfully as Spain did. The trajectory taken by the Luxembourgish campaign was more complex than the previous cases, simply because it came last. The key point is that the Luxembourgish campaign followed the same pattern the French and Dutch campaigns did at the beginning: The No campaign came into the debate first and set its themes on the agenda, forcing the Yes campaign to take the defensive position. The No campaign’s borrowed frames had limited success in activating peculiar latent fears in the society. Initially, just as in France and the Netherlands, the Yes campaign had difficulty responding to these No campaign arguments. However, the French and Dutch rejections caused a turning point in the campaign. The Yes campaign switched strategies, built stronger frames and centered the campaign on Luxembourg’s national interests. This new framing was further strengthened by better credibility and resources. The public opinion data confirms these patterns, showing a gradual increase in the No vote intentions until it is contained in the last phase by the revitalized Yes campaign. Because Luxembourg was the last referendum campaign, it was the only case where the conditioning variable – diffusion – had a substantial influence. The assessment of diffusion channels among the four referenda demonstrates that France and Luxembourg share key channels due to Luxembourg’s peculiar geographic, demographic and institutional structure. It was through these channels that France – the momentum case – had a considerable impact on Luxembourg. The Luxembourgish No campaign’s very existence was strengthened by the reliance on the French campaign but the borrowed frames were not tailored to the specific context of Luxembourg. Interestingly, because the public was profoundly subjected to the French campaign through the media and personal connections, the society appears to have conflated the French and Luxembourgish reality to some degree. This confusion could be read as remedying the adaptation problem

320 to some extent. Furthermore, the critical change in the Yes campaign can also be seen as a result of the sequencing logic. Consequently, it is difficult to understand the complex Luxembourgish vote fully without paying attention to diffusion effects. The influence of the independent variable – relative No campaign strength – is conditioned by diffusion. The diffusion process has an analytical input both in explaining the gradual rise of the negative vote and partly in its containment in the final stage. The Luxembourgish case demonstrates that that when present diffusion amplifies the effectiveness of the campaigns. Both Spain and Luxembourg had popular governments and pro-EU sentiment, but the No vote was much higher in Luxembourg because of the No campaign’s relative strength and diffusion effects. In the absence of both, the Luxembourg vote totals would likely have paralleled Spain's quite closely. A related question could be, ‘Is this diffusion effect unique, limited to the Luxembourg-France connection?’ The answer should be negative. Had Germany or Belgium held their referenda before Luxembourg, these would probably have had a similar impact as they also share peculiar diffusion channels with Luxembourg. The TNS-ILRES focus group study showed that Luxembourgers complain about the doubling of prices after the Euro in the exact same way as Germans discuss it; referring to ‘Teuro’, a German term combining ‘teuer’ – expensive – with ‘euro’.836 Perhaps the key finding of studying the Luxembourg case is that no country in the EU, regardless of their high support for domestic or European levels of governance, should automatically be expected to vote positively in EU referenda. Strategic No campaigns, coupled with diffusion effects under certain conditions, can cause an episodic dip in the public opinion towards the Union. Equally important is the impact of the revitalized Yes campaign, which indicates that these dips can be contained with the right use of framing and resources.

836 Dumont et al., Le Référendum sur le Traité Etablissant une Constitution pour l’Europe: Rapport Elaboré pour la Chambre des Députés. 321

PERSONAL INTERVIEWS

Political parties:

Yes campaigners: François Biltgen (CSV, MP, then Chairman of the CSV and Minister of Labor and Employment, Minister for Religious Communities, and Minister for Culture, Higher Education and Research) Laurent Mosar (CSV, MP) Ben Fayot (LSAP, MP, President of the Parliamentary Group) Charles Goerens (DP, MP, Minister of Foreign Affairs in 2004) Abbes Jacoby (The Greens, Secretary General of the Parliamentary Group) Dan Michels (The Greens, Parliamentary Attaché)

No campaigners: Jacques-Yves Henckes (ADR, MP) Henri Wehenkel (The Left)

Civil society:

André Kremer (Coordinator and Leader of the No Committee) Pierre Gramegna (Director-General of the Chamber of Commerce) Nico Clement (Confederation of Independent Trade Unions of Luxembourg OGBL) Nico Wennmacher (President of the Railways Trade Union FNCTTFEL-Landesverband) Tom Graas (then Director of the national RTL TV News) Marc Linster (then Director of the national RTL Radio) Anne-Marie Berny (ATTAC-LUX) Adrien Thomas (National Union of Luxembourgish Students UNEL) Frédéric Krier (National Union of Luxembourgish Students UNEL) Alfred Groff (Luxembourg Social Forum) Jürgen Stoldt (Political communication expert and Editor of Forum) Thomas Rupp (Organizer of the European No Campaign)

322

CHAPTER 7 CONCLUSION

Public sentiment is everything. With public sentiment, nothing can fail; without it, nothing can succeed. Consequently he who moulds public sentiment goes deeper than he who enacts statutes or pronounces decisions. He makes statutes and decisions possible or impossible to be executed.837

Abraham Lincoln, 1858

EU referenda results guide the future of European integration. This research shows that the national elite need to pay particular attention to public sentiment in EU referendum campaigns. The political advantage rests with the No side in a referendum campaign, even when the referendum initially enjoys widespread public support.838 The No campaign only needs to raise doubts in the minds of voters, play upon existing fears or link the proposal to unpopular issues. Public opinion can therefore be molded by strategic No campaigns that tap into the existing contention in society, even when these campaigns bring together ideological opposites from the far left and far right. Yes campaigners facing such strategic campaigns need to fight back with effective arguments. Yet, framing is not persuasion. Persuasion goes deeper as it implies a change in the individual’s evaluation of the subject. Framing does not alter one’s evaluation. By placing the emphasis on particular sub-issues within the broad subject, it instead increases their salience and thereby only shapes the way of thinking on the matter. When subjected to effective framing, individuals think along particular lines but the content of their beliefs may not necessarily change.839 Accordingly, the success of framing does not depend on its accuracy or

837 Excerpt from the First Debate, Ottawa, Illinois, 21 August 1858. Quoted in Roy Basler, Abraham Lincoln: His Speeches and Writings, Second ed. (Cleveland: Da Capo Press, 2001), 458. 838 LeDuc, "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro." 839 Druckman, "On the Limits of Framing Effects: Who Can Frame?," 1044. See also Chong and Druckman, "Framing Theory," 116. In cases where the individual does not hold a clear attitude, frames can introduce new considerations as well as emphasizing existing beliefs. Therefore, some framing messages can also be persuasive. 323 intellectual content but instead on it its appeal to audiences.840 This distinction is essential in understanding the 2005 TCE referenda results. In all four cases, post- referendum surveys found that public opinion towards the Union returned to the previous positive levels. What happened was an episodic dip in public support during the referendum campaigns, the extent of which was shaped by the mobilization capacity of the No campaigners in each country. I argue that the relative strength of the No campaign in relation to the Yes campaign is the main variable explaining the referendum results, but that diffusion processes may condition the magnitude of this strength in second-mover states.

Relative No campaign strength Referendum result (Magnitude of the No vote)

Diffusion effects

Figure 7.1: Analytical Framework

The degree to which the No campaigners successfully linked contentious issues to the referendum proposal affected the magnitude of the No votes. Strong No campaigns are those that engage in strategic agenda-setting and priming/framing. Strong No campaign framing, backed by credible sources and powerful mobilizational structures, countered early positive public opinion. As such, the strength of the campaign frames is the key component. Similarly, better credibility and resources on the part of the campaigners amplify the power of these frames by giving the speakers more standing and opportunities to advance them. Diffusion may further condition the strength of referendum campaigns through networking and learning processes across campaigners in first- and second-mover countries. Such cross-case influences are not automatic and depend on the existence of transmission belts between the states such as shared

840 Chong and Druckman, "Framing Theory," 111. 324

language/culture, common media sources, and collaborative networks/transnational linkages. Where these channels were open, the earlier cases influenced the later ones. Furthermore, campaign intensity matters. The more intense the campaign, the heavier was the influence transmitted by the existing diffusion channels. Campaign intensity instigated the transmission of influence. Consequently, where diffusion effects were significant, this influenced the strength of campaigns and the choice of campaign strategies. Campaign materials from 2005 provide an interesting illustration of these cross-case influences. Primarily via the European anti-globalization network, similar posters and slogans were circulated throughout Europe. Strikingly the three identical posters in Figure 7.2, blaming the TCE for the militarization of Europe, originate from different countries. The slogans read ‘Europe in a bad constitution’. The first photo is taken in Germany. On 12 May 2005 – before the French and Dutch rejections – as German MPs voted on the TCE, demonstrators stood on the steps of the Bundestag holding posters against the treaty.841 The second poster is taken directly from the brochure of the Dutch left-wing civil society No campaigner – ConstitutionNo. It presents the same poster and slogan in Dutch. The third one is a poster used by the Luxembourgish left-wing No campaigner, which is the same one, but in German.

841 Available at: http://www.ena.lu 325

Figure 7.2: Identical posters – against the TCE – in different European countries

The combination of campaign strength and diffusion effects leads to the following distribution of the four referendum cases in 2005:

High Diffusion Low Diffusion

Strong No Campaign France The Netherlands Weak No Campaign Luxembourg Spain

Table 7.1: The interaction between campaign strength and diffusion

Among the four cases, only Luxembourg shared significant diffusion channels with a previous case: France. Therefore while the independent variable alone explains the results in the first three cases, it only partially accounts for the referendum result in Luxembourg. Put differently, diffusion effects conditioned campaign strength in Luxembourg. Tables 7.2 and 7.3 present the findings from the four case studies. Table 7.2 summarizes the relative Yes and No campaign strength in each country. The check marks indicate whether the Yes or the No campaigners held a relative advantage in that particular field. Where both have the mark, it means that they had relative advantage at different phases of the campaign. If neither does, it means that they had similar standing and that this component was insignificant in the analysis. Table 7.3 summarizes the impact of diffusion effects. Here, the

326

check marks denote which countries share the specified diffusion channel and the final row indicates the existence of an intense campaign. Spain France The Luxembourg Netherlands Yes No Yes No Yes No Yes No Agenda- Timing Setting Priming Strength of the Framing campaign frames Credibility of the speakers Mobilizational structures Table 7.2: The relative campaign strength in 2005 EU Constitutional Referenda

Spain France The Luxembourg Netherlands Diffusion Shared language/culture channels Common media channels

Collaborative European networks/ Parliament groups transnational European linkages anti-globalization network (Institutional Ad hoc European networks and networks personal Mobile connections) communities Campaign Intense campaign intensity Table 7.3: Diffusion effects in 2005 EU Constitutional Referenda

To summarize, in Spain, the No campaign did not hold a relative advantage in any of the categories. The Yes and No campaigns both started around the same time and neither had credibility problems. Yet the No side suffered from weak frames and weak mobilizational structures. The only exception were the two regional No campaigns, which used stronger frames building on the contentious language recognition issue. As a result, the No vote

327

percentage was significantly higher in the regions (17% national vs. 33% in the Basque Country and 28% in Catalonia). The diffusion channels were limited to the institutional networks common to all four cases. While the left-wing No campaigners adopted the master frame of the European anti-globalization network – blaming the TCE for being market-friendly – they failed to adapt this frame to their national framework. The result was a quiet and non-controversial case, in which the Spanish Yes campaigners did not have a challenging task to sustain initial positive public opinion. With such a low-profile campaign, the Spanish case did not put pressure on existing diffusion channels and thereby did not influence the subsequent cases. In clear contrast, in France, the No campaign was strong across the board. The No campaigners started earlier and successfully put their themes on the national agenda, forcing the Yes campaigners to take the defensive position. The No campaign frames were strong as they centered on existing contentious issues in society, linking the TCE to declining social protection, increasing immigration and potential Turkish accession. Their excellent civil society mobilization drive further strengthened these frames by advancing them forcefully at the local level. The Yes side, on the other hand, suffered from weak frames and weak credibility as the center-left political parties were de facto divided between the two camps despite their official pro-TCE standing. Consequently, the No campaigners succeeded in linking the existing contention to the referendum proposal, and the Yes campaign could not sustain the early positive public opinion. Regarding diffusion, the French campaign was crucial. Being the most intense of all, it became the momentum case. The anti-globalization network, common to all four cases, primarily relied on the French left-wing criticism of the TCE and spread it to the other cases. The early and intense discussion in France influenced even Spain, whose referendum preceded France's. Most striking, however, was France’s impact on Luxembourg, with which it shared exceptional diffusion channels. The Dutch picture was similar to the French one. The Dutch No campaign was stronger than the Yes campaign in every realm. The No campaigners

328 benefited from an early start and determined the terms of the debate. Their frames effectively linked existing contentious issues to the European Constitution, highlighting primarily the danger of losing sovereignty in a European ‘super- state’. Running a vigorous campaign with the Socialist Party’s financial and institutional resources, the No campaign successfully struck a sensitive nerve given the recent political assassinations and the unpopular welfare state reforms. The No campaign’s success coupled with a hesitant Yes campaign that had weak framing and questionable credibility resulted in the turmoil case. Despite being the third country to hold a referendum, diffusion effects were minimal in the Dutch campaign because available diffusion channels were limited to the institutional networks common to all four cases. The one important aspect was the Dutch left-wing No campaigners' success in strategically tailoring the anti- globalization master frame to the Dutch local context. By doing so, they strengthened this borrowed frame. Among the four cases, the Dutch left-wing No campaigners were the only ones to pay attention to this task. France was essentially the origin of this master frame and thereby local adaptation was not an issue for the French. But the Spanish and Luxembourgish left-wing No campaigners overlooked this aspect. Luxembourg came last. Given the substantial diffusion channels it shared with France, it became the diffusion case. This case initially resembled the French and Dutch cases as the No side started early, set the campaign themes and had limited success in linking certain contentious issues – unemployment and decreasing protection of public services – to the European Constitution. But in addition, because of close links between Luxembourg and France, the main Luxembourgish No campaigner – the left-wing committee – took its lead from the French left-wing No campaign. This committee received campaign support from France both in terms of borrowed campaign frames and logistics. As such, the Luxembourgish No campaign’s existence depended on the French campaign, at the same time it neglected to adjust these borrowed frames to fit local conditions,

329

compromising the strength of the No campaign.842 Moreover, while diffusion had mixed effects on the No campaign, it arguably only strengthened the Yes campaign. The Yes campaigners consciously changed their campaign strategy and used stronger frames after the French and Dutch rejections, marking a key turning point in the campaign. Overall, in Luxembourg diffusion led both to an increase in the No vote and its ultimate ‘containment’ in the final stage of the campaign. Only in this last case did diffusion significantly condition the outcome.

Contribution to the EU Literature This analysis reveals the misleading nature of the dichotomy between ‘second- order’ and ‘attitude’ models in the literature, models that attribute EU referenda results nearly exclusively to either domestic or European factors. The second- order model points to national factors, such as the level of satisfaction with the government and voter identification with parties holding office, as determinants of the voting behavior.843 As such, issues relating to the EU are relegated to second place. The attitude model, on the other hand, proposes that the vote choice reflects voters’ underlying attitudes towards European integration.844 The fewer areas voters want included in the European integration, the more they vote No. Recent work in the field has started to look into campaign intensity and effectiveness but still seeks to determine the conditions under which one of these models can explain the results, mostly finding proof for the attitude model.845 By examining the four cases comparatively for the first time in a systematic analysis, my research uncovers key factors neglected by this literature. Existing studies overemphasized domestic dissatisfaction with the government as a causal factor in the French and Dutch cases. A closer look demonstrates that fall

842 Strikingly though, some Luxembourgers confused French problems with their own problems as a result of the heavy exposure to the French campaign. This confusion could be interpreted as lessening the adaptation problem to some extent by strengthening the borrowed frames. 843 Franklin, van der Eijk, and Marsh, "Referendum Outcomes and Trust in Government: Public Support for Europe in the Wake of Maastricht." 844 Siune, Svensson, and Tonsgaard, "The EU: The Danes Said “No” in 1992, But “Yes” in 1993: How and Why?." 845 For a comprehensive comparative study see Hobolt, Europe in Question: Referendums on European Integration. 330

2004 public opinion polls reflected positive attitudes towards the TCE in all four cases. The French and Dutch governments may have been relatively more unpopular than the Spanish and Luxembourgish governments, but this unpopularity pre-dated the early, uniformly positive public opinion polls on the referenda. Similarly, although the most comprehensive single-case studies of the French and the Dutch referenda also emphasized domestic dissatisfaction with certain EU policies such as market-friendly economic policies, EU enlargement, or the common currency, Euro, as causal factors in the rejections, my case studies reveal that these EU policies were equally contentious domestically at the time of the initial, pro-referendum public opinion polls.846 Therefore, the distinction between the two models is a false dichotomy. Successful No campaigns built a connection between existing contentious issues (at both the domestic and the EU levels) and the referendum proposal, claiming that these problems were caused or would be exacerbated by the proposed European treaty. In the absence of active No campaign framing, such contentious issues remained dormant and unlinked to the referendum in the public’s mind. The work based on second-order and attitude models, specifically those that do not study campaigns, are also too static and simplistic to capture the lengthy opinion formation process. While the level of government unpopularity and the level of support for the European integration are indeed critical to understanding referendum outcomes, these variables should instead be analyzed in relation to the campaigns. This is why my research focuses on the ‘long campaign’ period, starting from the initial positive polls until the day of the referendum. Dissatisfaction with the government may affect the campaign in two ways; it provides potential campaign material to the No side, and decreases the credibility of the government's Yes campaign. Similarly, public dissatisfaction with certain EU policies becomes activated and linked to the referendum proposal only when the No campaigns tap into them.

846 See Brouard and Tiberj, "The French Referendum: The Not So Simple Act of Saying Nay."; Brouard and Sauger, "Comprendre la Victoire du "Non" : Proximité Partisane, Conjoncture et Attitude à l’égard de l’Europe."; Aarts and van der Kolk, "Understanding the Dutch 'No': The Euro, the East, and the Elite."; Schuck and de Vreese, "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign." 331

Another important flaw in the literature is to simply categorize the Yes victories in Spain and Luxembourg as positive cases. Such categorization conceals the similar dynamics in the successful and unsuccessful referendum campaigns. The Spanish case demonstrated that in regions where the No campaigns were stronger, the No vote was indeed higher. Moreover, in Luxembourg the No vote percentage was as high as 43%. Earlier in the campaign, the No side managed to set its campaign themes on the agenda and forced the Yes campaign to defend the treaty, in a very similar fashion to the French and Dutch cases. Treating these referenda dichotomously as positive and negative cases hides these important similarities. Finally, methodologically, this study is the first of its kind to analyze campaign strategies in EU referenda by interviewing the campaigners and exploring campaign materials extensively and systematically in four cases. The extant literature either focuses on campaign strategies on an ad hoc basis (mostly in single case studies) or neglects them completely. This research also uses public opinion data differently, to spot variations in opinions on the same topic before and after the campaign. The literature suggests analyzing opinions expressed in a neutral control condition, where participants express their preferences without having been exposed to framing, to measure the framing effect.847 By comparing the early polls on the TCE with the referendum results, this study traces the evolution of public opinion in relation to campaign framing.

The Double EU Referenda The ‘double EU referenda’ present a valuable opportunity for further research and testing of my approach. What explains the success of a repeat referendum that approves the very same referendum question rejected by voters less than a year earlier? This has happened three times. In Denmark (on the Maastricht Treaty in 1992-1993) and in Ireland (on the Nice Treaty in 2001-2002 and on the Lisbon

847 Chong and Druckman, "A Theory of Framing and Opinion Formation in Competitive Elite Environments," 106. However they discuss this issue for experimental designs, and suggest using a control group in the experiments. 332

Treaty in 2008-2009), the public rejected the referendum proposal in an initial vote but approved it in a subsequent vote. Given my research findings, it seems likely that the Yes campaigners learned from their previous mistakes and therefore engaged in more effective framing in the second campaigns. I propose that when both Yes and No campaigners use strong frames, resources matter.848 As such, resources matter only conditionally. The relative level of resources available to each side is not the deciding factor as long as one side uses stronger framing. The government and all the mainstream political parties almost always campaign on the Yes side as Euroscepticism is confined to the extremes of the political spectrum.849 Consequently, the Yes campaigners in EU referenda have greater resources than the No campaigners. This research showed that despite having fewer resources, the No campaigns win if they employ stronger frames.850 Thus I posit that if – and only if – both Yes and No campaigners are effective framers, resource levels become crucial in determining who wins. A systematic study of all six referenda campaigns could tackle the question of ‘double campaigns’ by breaking the concept into two dimensions: campaign framing and campaign resources. While no work has compared all six campaigns, existing studies point to the importance of increasing campaign intensity in the second round of referenda.851 The most complete research brought together four of the cases (Denmark 1992-1993, Ireland 2001-2002) and showed that in the second referenda campaigns both the quantity and the type of information available to the public were different.852

848 I define resources broadly as mobilizational structures including financial and institutional resources. 849 Taggart, "A Touchstone of Dissent: Euroscepticism in Contemporary Western European Party Systems ". 850 de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue, 58. De Vreese and Semetko reach the same conclusion in their comprehensive study of the 2000 Danish referendum on the Euro. 851 Hobolt, "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums.", ———, "How Parties Affect Vote Choice in European Integration Referendums.", de Vreese, "Political Parties in Dire Straits? Consequences of National Referendums for Political Parties.", Garry, Marsh, and Sinnott, "'Second-order' versus 'Issue- voting' Effects in EU Referendums: Evidence from the Irish Nice Treaty Referendums." 852 Hobolt, Europe in Question: Referendums on European Integration. 333

The European Parties Elections and Referendums Network (EPERN), a network of scholars researching the impact of European integration on parties, elections and public opinion, has published briefing papers on the two Irish referenda.853 These papers confirm that the first referendum on the Lisbon Treaty followed the pattern of the French and Dutch cases. Since the Irish held favorable attitudes towards the treaty six months before the referendum date, the negative result was unexpected.854 Similarly, these briefing papers find that the Yes campaign’s relative weakness compared to the No campaign (specifically in framing) contributed to the result. As Holmes argues: 855 But perhaps the biggest problem throughout for the Yes campaign was finding a positive argument for Lisbon. The various groups emphasized the general advantage of being a part of the European Union, with little reference to any specifics of the Treaty. … One feature that showed up clearly was that voters felt that the No side had won the arguments - 68% of all voters, including as many as 57% of Yes voters, thought the No campaign had been more effective. … All of these arguments indicate the success of the No side in setting the agenda of the campaign, and in winning the subsequent debates.

In another similarity with the French and Dutch cases, the center-left political parties had significant difficulty in aligning their voters with their party choice. The majority of their followers voted against the treaty. In contrast, the voters of the No campaign parties remained in line with their party’s position. Regarding resources, the Yes side had strong access to financial and institutional resources as it was formed by the government, the largest political parties as well as a broad range of social forces. However, these resources were not mobilized forcefully. For instance, most of the civil society organizations ran a ‘half-hearted’

853 The EPERN is based at the University of Sussex. 854 Michael Holmes, "The Referendum on the Treaty of Lisbon in the Republic of Ireland, 12 June 2008," European Parties Elections and Referendums Network, Referendum Briefing Paper 16 (2008): 5. 855 Ibid. Holmes argues that the Yes campaigners issued warnings about the loss of influence Ireland would suffer if it rejected the treaty, which was somewhat self-defeating because their opponents were arguing that the EU was marginalizing small countries. Moreover, he suggests that there was a tendency to denigrate the opposition, with, for example, Bertie Ahern dismissing No campaigners as loolah’s (meaning lunatics). According to Holmes, both of these strategies were counter-productive. 334

campaign.856 The No campaign had fewer resources but was backed by a few political parties and the Libertas lobby group founded by Irish businessmen. These resources were used effectively to promote their strong framing. The briefing paper on the second Irish referendum, approving the Lisbon Treaty, is equally striking. In the period leading up to the second referendum, the Yes campaign worked hard and negotiated with the EU in order to secure some guarantees on the rejected treaty. They reassured the public concerning the No campaign’s arguments in the first round by guaranteeing a commissioner, Irish competency over tax rates, abortion and neutrality, and finally by reaching an agreement on workers rights. These guarantees resulted in stronger Yes campaign framing.857 Furthermore, the paper points to a revitalized Yes campaign with better organization and resources, enhanced use of new media sources and high profile figures, and particularly stronger framing:858 Many commentators … were openly and strongly critical of the Yes campaign for its failure to develop any momentum and to counter the perceived false claims of the No side (in the first round). … In the immediate aftermath of the rejection of Lisbon, pro-European activists came together to plan out how the succeeding Lisbon referendum would be won. For them the period between the two referendums was spent fundraising and supervising the formation of various groups to contest the second referendum. … The strategic goal of the Yes side was to make the referendum not on Lisbon but on Irish membership of the EU. To move away from trying to sell a largely bureaucratic Treaty and shift the emphasis to what the electorate could plainly understand: Irish membership of the EU, which the Irish public is enthusiastic about and fully aware of the benefits.

At the same time, the No campaign experienced a decrease in size and capacity. The No side lacked the targeted campaign strategy of the Yes side. In the second

856 Ibid.: 4. 857 These guarantees were not voted on in the referendum, but they were used by the government as justification for running a second round referendum on the same treaty. Essentially the government demonstrated that it had listened to the Irish people and had convinced the EU member states to address their concerns over Lisbon with these legally binding guarantees. See John FitzGibbon, "The Second Referendum on the Treaty of Lisbon in the Republic of Ireland, 2nd October 2009," European Parties Elections and Referendums Network, Referendum Briefing Paper 17 (2009): 2. 858 Ibid.: 2-6. 335

round, strikingly, 67% of voters found the Yes side to be more convincing.859 FitzGibbon explicitly argues that during the second campaign, the Yes side succeeded where it had failed in the first campaign, by setting the agenda and countering the No campaign framing. Furthermore, in the second campaign, not only did the Yes side use its financial and organizational resources more extensively and vigorously but the No side had fewer resources due to their lessened capacity.860 FitzGibbon once again openly presents the substantial resources of the well-organized Yes campaign as an important difficulty for the No side. Therefore I argue that where both campaigns had strong frames, the Yes side's better resources made the difference. The close resemblance of the first Irish referendum campaign to the French and Dutch TCE referenda cases demonstrates the generalizability of my findings.861 Public opinion is volatile and shifts considerably during campaigns, culminating in outcomes not predicted by pre-campaign polls. The 2000 Danish referendum that rejected the Euro, despite being a single issue referendum, also had key similarities with the TCE referenda regarding political party preferences and the issues used by the Yes and No campaigners.862 De Vreese and Semetko argue that in this case, the No campaign similarly and successfully broadened the debate to include the welfare system, sovereignty, immigration, whereas the Yes campaign centered on the economy and Europe’s importance in peace and prosperity. This study's lessons on diffusion are generalizable as well. Europe is full of countries that share similar cultures, media sources, institutional and personal

859 Ibid.: 11. 860 The Socialist Party, Sinn Féin and Cóir supported the No campaign with their resources but Ganley’s Libertas initially refused to join the campaign. Ganley entered the campaign only in the last phase. 861 This must be a cautious generalization as the type and question of the referendum might matter. The Lisbon Treaty is only an updated version of the TCE, meaning that the two documents are almost the same. The TCE and Lisbon Treaty referenda can be considered similar to referenda on other treaty amendments such as the Maastricht Treaty or the Nice Treaty, but may be somewhat different from single issue referenda such as those on adoption of the Euro. The literature finds that particularly in referenda on complex international treaties or large packages of constitutional provisions voters do not have well-formed opinions. LeDuc, "Opinion Change and Voting Behaviour in Referendums." 862 de Vreese and Semetko, Political Campaigning in Referendums: Framing the Referendum Issue. 336

connections. For instance, the briefing papers on the two Irish referenda highlight the UK’s influence on Ireland via common media channels.863 In another example, my interview data from the Netherlands confirms collaboration among the Dutch and Irish No campaigners via the EP groups during the first Irish referendum campaign on the Lisbon Treaty. The literature similarly points to the importance of sequencing at the 1994 EU membership referenda in Austria, Finland, Sweden, and Norway.864 Therefore diffusion effects can be observed across Europe, during various periods of time.

Broader Theoretical Implications This research has implications beyond the EU literature as well. It makes contributions to four different areas of study; political campaigns, contentious politics, diffusion, and international organizations. Regarding political campaigns, the earlier literature found that campaigns have minimal influence mainly based on the study of United States elections. Given the limitations of these early explanations in accounting for election outcomes, recently scholars began looking into the role of strategic communications. However, this literature is still emerging and relatively few studies have been carried out on referendum campaigns. This study confirms that cognitive processes affect the election outcome through strategic agenda-setting, priming, and framing. Referendum campaigns provide an excellent opportunity to study these effects as in such campaigns, voting behavior exhibits greater volatility than regular elections.865 Especially where the referendum proposal is unfamiliar, complex or technical, voter predispositions such as partisanship or ideology remain weak making campaigns all the more influential. The 2005 EU constitutional referenda campaigns, given the public’s high level of unfamiliarity with the treaty, significantly shaped public opinion on the subject. Importantly, studies exploring campaign effects only in the short, intense, final phase fail to

863 See Holmes, "The Referendum on the Treaty of Lisbon in the Republic of Ireland, 12 June 2008." 864 Jahn and Storsved, "Legitimacy through Referendum? The Nearly Successful Domino-Strategy of the EU-Referendums in Austria, Finland, Sweden and Norway," 21. 865 LeDuc, "Opinion Change and Voting Behaviour in Referendums." 337

capture the full impact of a lengthy campaign. Instead, this study analyzes the long campaign starting from the early polls on the subject to fully understand the evolution of the public’s opinion. Moreover, by studying campaign strategies systematically in four cases based on both campaign materials and in-depth interviews, this research contributes to our understanding of strategic framing, and the factors that contribute to frame strength. This study also has important implications for the contentious politics literature. Bringing together the findings of the public opinion and social movements literatures in the analysis of campaign framing, it shows that bridging these literatures is fruitful for building a comprehensive theoretical framework. While election campaigns are outside of the traditional scope of social movements studies, this study investigates how campaign actors framed the political opportunity they were given by the announcement of referenda, and what financial and institutional resources they used in this process. In turn, the detailed findings of the public opinion literature on frame strength enhance our understanding of the broader concept of frame resonance in the social movements literature. Third, this study makes a significant contribution to the diffusion literature by analyzing cross-case influences, also known as ‘Galton’s problem’, in political campaigns. So far, no study systematically analyzes the diffusion process in referendum campaigns. Drawing a parallel to the modular phenomena literature, this research shows that diffusion in such campaigns follow a similar path as the campaigners’ actions are based in significant part on the prior successful example of others. Just as in modular collective action, I find that certain factors facilitate the transmission of diffusion and serve as channels. As such, diffusion is not automatic. In referendum campaigns, the existence of shared language/culture, common media channels, collaborative networks/transnational linkages is necessary to carry the influence across cases. Finally concerning the international relations discipline, this research shows that the EU cannot be studied simply as international cooperation among the member states. As Taggart pointed out, the literature needs to integrate the

338

study of domestic politics with the study of the European integration process more systematically, while being sensitive to the growing interdependence between the two domains.866 More attention should be paid to the domestic level as national politicians are the key actors in explaining the benefits and costs of the integration process to EU citizens. If the EU is aspiring to become a state-like entity, moving beyond an international organization, a comparative politics approach will be necessary to understand the core dilemmas it is facing.

Policy Implications for Europe The false dichotomy between the second-order and attitude models in the EU referenda literature has important policy implications. If the referendum vote represents national concerns instead of underlying attitudes towards Europe, then the use of direct democracy on EU questions is problematic. Despite being divided, the literature has found more evidence for the attitude model. This implies that citizens are capable of voting in line with their interests concerning Europe under certain conditions.867 Sharing the view that the use of direct democracy is justified, I take a step further and propose that the campaign argumentation is what matters the most as public opinion is highly susceptible to framing effects. The attitudes on the referendum proposal can be molded by strategic No campaigns which link this proposal to the existing contention relating to both domestic and EU levels of governance. The regular success of No campaigns across Europe may at first glance appear as a major barrier to increasing European integration. Nevertheless, the EU can arguably benefit from increasing contestation, even when that contestation challenges the EU itself. As Tilly shows, the history of democracy and contention are deeply intertwined in Europe.868 The EU has long been a non-issue in European democracies, based on a ‘permissive consensus’ among European

866 Taggart, "Keynote Article: Questions of Europe - The Domestic Politics of the 2005 French and Dutch Referendums and their Challenge for the Study of European Integration," 8. 867 See Hobolt, Europe in Question: Referendums on European Integration. 868 Charles Tilly, Contention and Democracy in Europe, 1650-2000 (Cambridge: Cambridge University Press, 2004). 339

citizens.869 The member-state governments negotiated complex, technical EU policies on behalf of their inattentive citizens, while the public tacitly approved these developments. As a result, European integration was not initially politicized and national politicians were not accustomed to discussing the Union as part of their electoral campaigns.870 Rising Euroscepticism since the 1990s showed that this permissive consensus has collapsed. Yet, this does not by any means indicate the end of European integration. It has only entered a new phase. Given the significant increase in the Union’s competences, the member states now need to explain EU policies to their publics and defend them when necessary, with real arguments. Indeed, the previous lack of politicization on the EU, which is becoming a state-like entity, was not only undemocratic but also dangerous for the integration project itself. As this research has shown, negative and more immediate linkages to existing and thereby available frames in EU referenda campaigns can easily win over traditional pro-EU arguments resting on positive abstract, non- immediate benefits.871 Indeed, the leader of the Dutch Labor Party Wouter Bos stated: ‘I have never seen as much debate in the Netherlands about Europe and about this country’s place in Europe. I prefer it when there is open discussion rather than the continuation of covert discontent’.872 Even the formation of pan- European social movements, although they strongly oppose the EU’s economic policies, could be welcomed as contributing to increasing public consciousness of the EU and its evolving role. The EU does not need priests but national politicians who can explain the Union to their constituencies with substantive arguments instead of simply relying on its role in creating peace after the Second World

869 Pippa Norris, "Representation and the Democratic Deficit," European Journal of Political Research 32, no. 2 (1997). 870 Claes de Vreese, "No News Is Bad News! The Role of the Media and News Framing in Embedding Europe " WRR Scientific Council for Government Policy Web Publications 20 (2007).; Catherine de Vries, "Sleeping Giant: Fact or Fairytale? How European Integration Affects National Elections," European Union Politics 8, no. 3 (2007).; Christoph Meyer, "Political Legitimacy and the Invisibility of Politics: Exploring the European Union's Communication Deficit," Journal of Common Market Studies 37, no. 4 (1999). 871 Dekker and Wennekers, "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen.", cited in Dekker and van der Horst, Europe's Neighbours: European Neighbourhood Policy and Public Opinion on the European Union. 872 Quoted in Deloy, "The Dutch Reject the European Constitution." 340

War. As Luxembourg’s Prime Minister Juncker put it: ‘Europe does not make people dream anymore’.873 What the Union needs is effective political communication to its citizens, primarily by national politicians. The future legitimacy and path of European integration depend on it.

873 Quoted in Hausemer, "Luxembourg's Referendum on the European Constitutional Treaty." 341

BIBLIOGRAPHY

Aarts, Kees, and Henk van der Kolk, eds. Nederlanders en Europa: Het Referendum over de Europese Grondwet. Amsterdam: Utigeverij Bert Bakker, 2005. ———. "Op Weg Naar 1 Juni." In Nederlanders en Europa: Het Referendum over de Europese Grondwet, edited by Kees Aarts and Henk van der Kolk. Amsterdam: Utigeverij Bert Bakker, 2005. ———. "Understanding the Dutch 'No': The Euro, the East, and the Elite." Political Science and Politics 39, no. 2 (2006). Anker, Hans. "The Netherlands, Referendum on the European Constitution, June 1: A referendum on the Gap between the Citizens and the Political Establishment " In Election Time 2005: The European Yearbook of Political Campaigning 2005, edited by Sepp Hartinger: European Association of Political Consultants and Hartinger Consulting Communications, 2006. Ansolabehere, Stephen, and Shanto Iyengar. Going Negative: How Attack Ads Shrink and Polarize the Electorate. New York: The Free Press, 1995. ———. "Riding the Wave Claiming Ownership over Issues: The Joint Effect of Advertising and News Coverage in Campaigns." Public Opinion Quarterly 58, (1994): 335-57. Arriba, Ana, and Luis Moreno. "Spain - Poverty, Social Exclusion and 'Safety Nets'." In Welfare State Reform in Southern Europe: Fighting Poverty and Social Exclusion in Italy, Spain, Portugal and Greece, edited by Maurizio Ferrera. Abingdon: Routledge, 2005. Baden, Christian, and Claes de Vreese. "Making Sense: A Reconstruction of People's Understandings of the European Constitutional Referendum in the Netherlands." Communications 33, no. 2 (2008): 117-45. Basler, Roy. Abraham Lincoln: His Speeches and Writings. Second ed. Cleveland: Da Capo Press, 2001. Becker, Frans, and Rene Cuperus. "Welfare State Blues: Thirty Years of Reform Policy in the Netherlands." In Public Discourse and Welfare State Reform: The Social Democratic Experience, edited by Vivien Schmidt. Amsterdam: Mets & Schilt, 2005. Beissinger, Mark. "Structure and Example in Modular Political Phenomena: The Diffusion of Bulldozer/Rose/Orange/Tulip Revolutions." Perspectives on Politics 5, no. 2 (2007). Belanger, Eric, and Kees Aarts. "Explaining the Rise of the LPF: Issues, Discontent, and the 2002 Dutch Election." Acta Politica 41, (2006): 4-20. Benford, Robert, and David Snow. "Framing Processes and Social Movements: An Overview and Assessment." Annual Review of Sociology 26, no. 1 (2000). Berezin, Mabel. "Appropriating the 'No': The French National Front, the Vote on the Constitution, and the 'New' April 21." Political Science and Politics 39, no. 2 (2006).

342

Berganza, Rosa. "Media Framing of the Spanish Referendum on EU Constitution." In International Communication Association. Dresden, Germany, 2006. Besselink, Leonard. "Double Dutch: The Referendum on the European Constitution." European Public Law 12, no. 3 (2006). Binnema, Harmen, and Ben Crum. "Resistance to Europe as a Carrier of Mass- Elite Incongruence: The Case of The Netherlands." In Les Résistances à l'Europe: Cultures Nationals, Ideologies et Strategies d'Acteurs, edited by Justine Lacroix and Ramona Coman. Bruxelles: Editions de l'Université de Bruxelles, 2007. Blanco, Víctor Sampedro, Antonia María Ruiz Jiménez, and Bruno Carriço Reis. "El Referendo del Tratado de la UE en la Prensa Española de Referencia: A Favor de la Constitución o en Contra de la Unión Europea." Elcano Royal Institute Working Paper 20, (2005). Brady, Henry, Richard Johnston, and John Sides. "The Study of Political Campaigns." In Capturing Campaign Effects, edited by Henry Brady and Richard Johnston. Ann Arbor: University of Michigan Press, 2006. Braun, Dietmar, and Fabrizio Gilardi. "Taking 'Galton's Problem' Seriously: Towards a Theory of Policy Diffusion." Journal of Theoretical Politics 18, no. 3 (2006). Brewer, Paul. "Value Words and Lizard Brains: Do Citizens Deliberate about Appeals to their Core Values?" Political Psychology 22, no. 1 (2001): 45- 64. Brinks, Daniel, and Michael Coppedge. "Diffusion Is No Illusion: Neighbor Emulation in the Third Wave of Democracy." Comparative Political Studies 39, no. 4 (2006): 463-89. Brouard, Sylvain, and Nicolas Sauger. "Comprendre la Victoire du "Non" : Proximité Partisane, Conjoncture et Attitude à l’égard de l’Europe." In Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, edited by Annie Laurent and Nicolas Sauger. Paris: CEVIPOF, 2005. Brouard, Sylvain, and Vincent Tiberj. "The French Referendum: The Not So Simple Act of Saying Nay." Political Science and Politics 39, no. 2 (2006). Bunce, Valerie, and Sharon Wolchik. "International Diffusion and Postcommunist Electoral Revolutions." Communist and Post-Communist Studies 39, no. 3 (2006). Buruma, Ian. Murder in Amsterdam: The Death of Theo van Gogh and the Limits of Tolerance. New York: The Penguin Press, 2006. Campbell, Angus, Philip Converse, Warren Miller, and Donald Stokes. The American Voter. New York: Wiley, 1960. Capdevila, Arantxa, Lorena Gómez, and Laia Aubia. "Partisan Strategies in Light of the European Constitution in Spain." EpsNet Electronic Journal 5, (2006). Castanares, Juan Carlos, and Antonio Juste. "Spain: In the Center or on the Periphery of Europe?" In Southern Europe and the Making of the

343

European Union, edited by Antonio Pinto and Nuno Teixeira. New York: Columbia University Press, 2002. Cautrès, Bruno. "Les Clivages Socio-politiques sur l’intégration Européenne et le Vote du 29 Mai 2005." In Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, edited by Annie Laurent and Nicolas Sauger. Paris: CEVIPOF, 2005. Chari, Raj. "The 2008 Spanish Election: A Balancing Game." West European Politics 31, no. 5 (2008): 1069-77. Chong, Dennis. Rational Lives: Norms and Values in Politics and Society. Chicago: University of Chicago Press, 2000. Chong, Dennis, and James Druckman. "Framing Public Opinion in Competitive Democracies." American Political Science Review 101, no. 4 (2007): 637- 55. ———. "Framing Theory." Annual Review of Political Science 10, (2007): 103- 26. ———. "A Theory of Framing and Opinion Formation in Competitive Elite Environments." Journal of Communication 57, no. 1 (2007): 99-118. Christin, Thomas, and Alexander Trechsel. "Joining the EU?: Explaining Public Opinion in Switzerland." European Union Politics 3, no. 4 (2002). CIA. "The World Factbook." (2010), https://www.cia.gov/library/publications/the-world-factbook/geos/lu.html. Clement, Franz "Le Travail Intérimaire au Luxembourg: Un Régulateur du Marché du Travail dans la Grande Région?" European Employment Services (EURES) (2006), http://www.eureslux.org/eures.taf?idnav=136. Closa, Carlos. "Business as Usual: EU Policy Under Zapatero -The EU Policy of the Socialist Government 2004-2008." In American Political Science Association. Toronto, Canada 2009. Closa, Carlos, and Paul Heywood. Spain and the European Union. New York: Palgrave Macmillan, 2004. Cobb, Michael, and James Kuklinski. "Changing Minds: Political Arguments and Political Persuasion." American Journal of Political Science 41, no. 1 (1997): 88-121. Collier, David, and Richard Messick. "Prerequisites versus Diffusion: Testing Alternative Explanations of Social Security Adoption." American Political Science Review 69, (1975): 1299-315. Converse, Philip. "The Nature of Belief Systems in Mass Publics." In Ideology and Discontent, edited by David Apter. New York: The Free Press, 1964. Cornelius, Wayne. "Spain: The Uneasy Transition." In Controlling Immigration: A Global Perspective, edited by Wayne Cornelius, Takeyuki Tsuda, Philip Martin and James Hollifield. Stanford: Stanford University Press, 2004. Cornelius, Wayne, Takeyuki Tsuda, Philip Martin, and James Hollifield, eds. Controlling Immigration: A Global Perspective. Stanford: Stanford University Press, 2004. Criddle, Byron. "The French Referendum on the Maastricht Treaty September 1992." Parliamentary Affairs 46, no. 2 (1993): 228-38.

344

Crum, Ben. "The EU Constitutional Treaty in the Netherlands: Could a Better Embedding Have Made a Difference?" WRR Scientific Council for Government Policy Web Publications no. 25 (2007). ———. "Party Stances in the Referendums on the EU Constitution: Causes and Consequences of Competition and Collusion." European Union Politics 8, no. 1 (2007). Darcy, R, and Michael Laver. "Referendum Dynamics and the Irish Divorce Amendment." Public Opinion Quarterly 54, no. 1 (1990): 1-20. Davies, Peter. The Extreme Right in France, 1789 to the Present: From de Maistre to Le Pen. New York: Routledge, 2002. Davis, Kathryn Anne. Language Planning in Multilingual Contexts: Political Communities and Schools in Luxembourg. Amsterdam and Philadelphia: John Bejamins Publishing Company, 1994. de Vreese, Claes. The Dynamics of Referendum Campaigns: An International Perspective. Basingstoke: Palgrave MacMillan, 2007. ———. "No News Is Bad News! The Role of the Media and News Framing in Embedding Europe " WRR Scientific Council for Government Policy Web Publications 20, (2007). ———. "Political Parties in Dire Straits? Consequences of National Referendums for Political Parties." Party Politics 12, no. 5 (2006): 581-98. ———. "Why European Citizens Will Reject the European Constitution." Harvard University Minda de Gunzburg Centre for European Studies Working Paper 116, (2004). de Vreese, Claes, and Hajo Boomgaarden. "Immigration, Identity, Economy and the Government: Understanding Variation in Explanations for Outcomes of EU-related Referendums." In The Dynamics of Referendum Campaigns: An International Perspective, edited by Claes De Vreese. Basingstoke: Palgrave Macmillan, 2007. ———. "Projecting EU Referendums: Fear of Immigration and Support for European Integration." European Union Politics 6, no. 1 (2005). de Vreese, Claes, and Holli Semetko. Political Campaigning in Referendums: Framing the Referendum Issue. Abingdon: Routledge, 2004. de Vries, Catherine. "Sleeping Giant: Fact or Fairytale? How European Integration Affects National Elections." European Union Politics 8, no. 3 (2007): 363-85. de Wilde, Pieter. "Reasserting the Nation State: The Trajectory of Euroscepticism in the Netherlands 1992-2005." RECON Online Working Paper 1, (2009). Dehousse, Renaud. "The Unmaking of a Constitution: Lessons from the European Referenda." Constellations 13, no. 2 (2006). Dekker, Paul, and Sjef Ederveen, eds. European Times: Public Opinion on Europe & Working Hours, Compared and Explained. The Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis 2005. Dekker, Paul, and Albert van der Horst. Europe's Neighbours: European Neighbourhood Policy and Public Opinion on the European Union. The

345

Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis, 2008. Dekker, Paul, and Charlotte Wennekers. "Publieke Opinies over Europa: Tussen Abstracte Steun en Concrete Ergernissen." Internationale Spectator 59, no. 12 (2005): 627-30. della Porta, Donatella. Globalization from Below: Transnational Activists and Protest Networks. Minneapolis: University of Minnesota Press, 2006. della Porta, Donatella, and Mario Diani. Social Movements: An Introduction. Oxford: Blackwell, 1999. della Porta, Donatella, and Hanspeter Kriesi. "Social Movements in a Globalizing World: An Introduction." In Social Movements in a Globalizing World, edited by Donatella Della Porta, Hanspeter Kriesi and Dieter Rucht. London: Macmillan, 1999. Deloy, Corinne. "29th May 2005: A Devastated Landscape." Robert Schuman Foundation Report (2005). ———. "The Dutch Reject the European Constitution." Robert Schuman Foundation Report (2005). ———. "Luxembourg Says 'Yes' to the European Constitution." Robert Schuman Foundation Report (2005). ———. "Referendum on the European Constitution in Luxembourg 10th July 2005." Robert Schuman Foundation Report (2005). ———. "Referendum on the European Constitution in Luxembourg: A Round up a Week Before the Election." Robert Schuman Foundation Report (2005). ———. "Referendum on the European Constitution in Spain on 20th February: An Assessment Just a Few Days Before the Election." Robert Schuman Foundation Report (2005). ———. "Referendum on the Ratification of the European Constitution in Spain." Robert Schuman Foundation Report (2005). ———. "The Spanish Approve the European Constitution by a Wide Majority." Robert Schuman Foundation Report (2005). Druckman, James. "On the Limits of Framing Effects: Who Can Frame?" Journal of Politics 63, no. 4 (2001): 1041-66. ———. "Political Preference Formation: Competition, Deliberation, and the (Ir) Relevance of Framing Effects." American Political Science Review 98, no. 4 (2004): 671-86. Druckman, James, Lawrence Jacobs, and Eric Ostermeier. "Candidate Strategies to Prime Issues and Image." Journal of Politics 66, no. 4 (2004): 1180- 202. Druckman, James, and Rose McDermott. "Emotion and the Framing of Risky Choice." Political Behavior 30, no. 3 (2008): 297-321. Druckman, James, and Kjersten Nelson. "Framing and Deliberation: How Citizens' Conversations Limit Elite Influence." American Journal of Political Science 47, no. 4 (2003): 729-45. Dumont, Patrick, Fernand Fehlen, Raphaël Kies, and Philippe Poirier. Le Référendum sur le Traité Etablissant une Constitution pour l’Europe:

346

Rapport Elaboré pour la Chambre des Députés. Luxembourg: University of Luxembourg, 2007. Duseigneur, Guillaume. "Lorsque l’Union Européenne Questionne l’Identité Partisane: Le Référendum Interne du PS sur le TCE." In Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, edited by Annie Laurent and Nicolas Sauger. Paris: CEVIPOF, 2005. Ederveen, Sjef, and Paul Dekker. Destination Europe: Immigration and Integration in the European Union. The Hague: SCP The Netherlands Institute for Social Research, CPB Netherlands Bureau for Economic Policy Analysis, 2004. Eichenberg, Richard, and Russell Dalton. "Europeans and the European Community: The Dynamics of Public Support for European Integration." International Organization 47, no. 4 (1993). Elkins, Zachary, and Beth Simmons. "On Waves, Clusters, and Diffusion: A Conceptual Framework." The Annals of the American Academy of Political and Social Science 598, no. 1 (2005). Engelen, E. "How to Solve the Riddle of Belated Euro Contestation in the Netherlands?" WRR Scientific Council for Government Policy Web Publications no. 26 (2007). Escribano, Antonio. "The EC and Spanish Immigration Policy." In Spain and EC Membership Evaluated, edited by Amparo Barbado. London: Pinter Publishers, 1993. EurActiv. "Austria, Luxembourg and Switzerland to Change Bank Secrecy Rules " (2009), http://www.euractiv.com/en/financial-services/austria- luxembourg-switzerland-change-bank-secrecy-rules/article-180329. Eurobarometer. "The European Constitution: Post-referendum Survey in France." European Commission: Brussels (2005). ———. "The European Constitution: Post-referendum Survey in Luxembourg." European Commission: Brussels (2005). ———. "The European Constitution: Post-referendum Survey in Spain." European Commission: Brussels (2005). ———. "The European Constitution: Post-referendum Survey in the Netherlands." European Commission: Brussels (2005). ———. "Qualitative Study on EU Citizens and the Euro in the Months Following its Introduction." European Commission: Brussels (2002). ———. "Special Eurobarometer: Face-to-face Interviews with EU 25 on the Draft European Constitution." European Commission: Brussels, November (2004). ———. "Standard Eurobarometer 61: Public Opinion in the EU, Spring 2004." European Commission: Brussels (2004). ———. "Standard Eurobarometer 62: Public Opinion in the EU, Autumn 2004." European Commission: Brussels (2004). ———. "Standard Eurobarometer 64: Public Opinion in the EU, Autumn 2005." European Commission: Brussels (2005).

347

European Employment Services (EURES). "Les Cahiers Transfrontaliers d’EURES." (2005), http://www.eureslux.org/eures.taf?Idnav=70&Idnav=fr. Féron, Élise. "Anti-globalization Movements and the European Agenda." Innovation: The European Journal of Social Sciences 17, no. 2 (2004). Fiske, Susan, and Shelley Taylor. Social Cognition. New York: McGraw Hill, 1984. FitzGibbon, John. "The Second Referendum on the Treaty of Lisbon in the Republic of Ireland, 2nd October 2009." European Parties Elections and Referendums Network, Referendum Briefing Paper 17, (2009). Flood, Chris. "French Euroscepticism and the Politics of Indifference." In French Relations with the European Union, edited by Helen Drake. Abingdon: Routledge, 2005. Follesdal, Andreas, and Simon Hix. "Why There is a Democratic Deficit in the EU: A Response to Majone and Moravcsik." Journal of Common Market Studies 44, no. 3 (2006). Font, Joan. "El Voto de los Españoles en el Referendum sobre el Tratado Constitucional." Elcano Royal Institute Working Paper 22, (2005). Font, Joan, and Elisa Rodriguez. "The Spanish Referendum on the EU Constitution: Issues, Party Cues and Second-Order Effects." In The Midwest Political Science Association Annual National Conference. Chicago, U.S.A. , 2006. Fortuyn, Pim. De Puinhopen van Acht Jaar Paars: Een Genadeloze Analyse van de Collectieve Sector en Aanbevelingen voor een Krachtig Herstelprogramma. Rotterdam: Karakter, 2002. ———. Zielloos Europa: Tegen een Europa van Technocraten, Bureaucratie, Subsidies en Overmijdelijke Fraude. Utrecht: A.W.Bruna, 1997. Franck, Raphael. "Why Did a Majority of French Voters Reject the European Constitution?" European Journal of Political Economy 21, no. 4 (2005). Franklin, Mark. "Learning from the Danish Case: A Comment on Palle Svensson's Critique of the Franklin Thesis." European Journal of Political Research 41, no. 6 (2002). Franklin, Mark, Cees van der Eijk, and Michael Marsh. "Referendum Outcomes and Trust in Government: Public Support for Europe in the Wake of Maastricht." West European Politics 18, (1995). Gabel, Matthew. "Economic Integration and Mass Politics: Market Liberalization and Public Attitudes in the European Union." American Journal of Political Science 42, no. 3 (1998). Gabel, Matthew, and Harvey Palmer. "Understanding Variation in Public Support for European Integration." European Journal of Political Research 27, no. 1 (1995). Gabel, Matthew, and Kenneth Scheve. "Mixed Messages: Party Dissent and Mass Opinion on European Integration." European Union Politics 8, no. 1 (2007): 37-59. Gallouj, Camal, and Karim Gallouj. "The French Social Protection System: Current State and Future Prospects." In The Handbook of European

348

Welfare Systems, edited by Klaus Schubert, Simon Hegelich and Ursula Bazant. Abingdon: Routledge, 2009. Gamson, William. "Political Discourse and Collective Action." In International Social Movement Research, edited by Bert Klandermans, Hanspeter Kriesi and Sidney Tarrow. Greenwich: JAI Press, 1988. Garry, John, Michael Marsh, and Richard Sinnott. "'Second-order' versus 'Issue- voting' Effects in EU Referendums: Evidence from the Irish Nice Treaty Referendums." European Union Politics 6, no. 2 (2005). George, Alexander. "Case Studies and Theory Development: The Method of Structured, Focused Comparison." In Diplomacy: New Approaches in History, Theory, and Policy, edited by Paul Gordon Lauren. New York: Free Press, 1979. Gerstlé, Jacques. "The Impact of Television on French Referendum Campaign in 2005." Notre Europe (2006). Gil-Escoin, Paloma Villota, and Susana Vazquez. "The Welfare State in Spain: Unfinished Business." In The Handbook of European Welfare Systems, edited by Klaus Schubert, Simon Hegelich and Ursula Bazant. Abingdon: Routledge, 2009. Glencross, Andrew, and Alexander Trechsel. "First or Second Order Referendums? Understanding the Votes on the Constitutional Treaty in Four EU Member States." Manuscript, European University Institute (2007). Goffman, Erving. Frame Analysis: An Essay on the Organization of Experience. New York: Harper and Row, 1974. Gomez-Reino, Marga, Ivan Llamazares, and Luis Ramiro. "Euroscepticism and Political Parties in Spain." In Opposing Europe? The Comparative Party Politics of Euroscepticism: Volume I, edited by Aleks Szczerbiak and Paul Taggart. Oxford: Oxford University Press, 2008. Graber, Doris. Processing Politics: Learning from Television in the Internet Age. Chicago: University of Chicago Press, 2001. Grunberg, Gerard. "Euroscepticism in France, 1992-2002." In Opposing Europe? The Comparative Party Politics of Euroscepticism: Volume I, edited by Aleks Szczerbiak and Paul Taggart. Oxford: Oxford University Press, 2008. Guibernau, Montserrat. "Nations without States in the EU: The Catalan Case." In European Integration and the Nationalities Question, edited by John McGarry and Michael Keating. Abingdon: Routledge, 2006. Guillen, Ana, Santiago Alvarez, and Pedro Adao e Silva. "Redesigning the Spanish and Portuguese Welfare States: The Impact of Accesion into the European Union." In In Spain and Portugal in the European Union: The First Fifteen Years, edited by Sebastian Royo and Paul Manuel. London: Frank Cass, 2003. Guyomarch, Alain, Howard Machin, and Ella Ritchie. France in the European Union. New York: St,Martin's Press, 1998. Hainsworth, Paul. "France Says No: The 29 May 2005 Referendum on the European Constitution." Parliamentary Affairs 59, no. 1 (2006).

349

Hamann, Kerstin. "European Integration and Civil Society in Spain." In Spain and Portugal in the European Union: The First Fifteen Years, edited by Sebastian Royo and Paul Manuel. London: Frank Cass, 2003. Harmsen, Robert. "The Dutch Referendum on the Ratification of the European Constitutional Treaty." European Parties Elections and Referendums Network, Referendum Briefing Paper 13, (2005). ———. "Euroscepticism in the Netherlands: Stirrings of Dissent." In Euroscepticism: Party Politics, National Identity and European Integration, edited by Robert Harmsen and Menno Spiering. Amsterdam: Rodopi, 2004. Harrison, Joseph, and David Corkill. Spain: A Modern European Economy. Aldershot: Ashgate, 2004. Hausemer, Piere. "Luxembourg's Referendum on the European Constitutional Treaty." European Parties Elections and Referendums Network, Referendum Briefing Paper 14, (2005). Hedstrom, Peter, Rickard Sandell, and Charlotta Stern. "Mesolevel Networks and the Diffusion of Social Movements: The Case of the Swedish Social Democratic Party." American Journal of Sociology 106, no. 1 (2000): 145- 72. Hix, Simon. "Dimensions and Alignments in European Union Politics: Cognitive Constraints and Partisan Responses." European Journal of Political Research 35, no. 1 (1999). Hobolt, Sara. "Campaign Information and Voting Behavior in EU Referendums." In The Dynamics of Referendum Campaigns: An International Perspective, edited by Claes De Vreese. Basingstoke: Palgrave MacMillan, 2007. ———. "Direct Democracy and European Integration." Journal of European Public Policy 13, no. 1 (2006). ———. Europe in Question: Referendums on European Integration. New York: Oxford University Press, 2009. ———. "How Parties Affect Vote Choice in European Integration Referendums." Party Politics 12, no. 5 (2006): 623–47. ———. "Taking Cues on Europe? Voter Competence and Party Endorsements in Referendums on European Integration." European Journal of Political Research 46, no. 2 (2007): 151-82. ———. "When Europe Matters: The Impact of Political Information on Voting Behaviour in EU Referendums." Journal of Elections, Public Opinion and Parties 15, no. 1 (2005): 85-109. Hoffmann, Jean-Paul. "Lëtzebuergesch in Luxembourg Today." In Luxembourg and Lëtzebuergesch: Language and Communication at the Crossroads of Europe, edited by Gerald Newton. Oxford: Oxford University Press, 1996. Hollifield, James. "France: Republicanism and the Limits of Immigration Control." In Controlling Immigration: A Global Perspective, edited by Wayne Cornelius, Takeyuki Tsuda, Philip Martin and James Hollifield. Stanford: Stanford University Press, 2004.

350

Holman, Otto. Integrating Southern Europe: EC Expansion and the Transnationalization of Spain. London: Routledge, 1996. Holmes, Michael. "The Referendum on the Treaty of Lisbon in the Republic of Ireland, 12 June 2008." European Parties Elections and Referendums Network, Referendum Briefing Paper 16, (2008). Hooghe, Liesbet, Gary Marks, and Carole Wilson. "Does Left/Right Structure Party Positions on European Integration?" Comparative Political Studies 35, no. 8 (2002): 965-89. Howarth, David. The French Road to European Monetary Union. Basingstoke: Palgrave, 2001. Huber, Evelyne, and John Stephens. Development and Crisis of the Welfare State: Parties and Policies in Global Markets. Chicago: University of Chicago Press, 2001. Hug, Simon. Voices of Europe: Citizens, Referendums and European Integration. Lanham: Rowman & Littlefield Publishers, 2002. Hug, Simon, and Pascal Sciarini. "Referendums on European Integration: Do Institutions Matter in the Voter’s Decision?" Comparative Political Studies 33, no. 1 (2000): 3-36. International Monetary Fund. "IMF Country Report: Luxembourg." 2006. Ivaldi, Gilles. "Beyond France's 2005 Referendum on the European Constitutional Treaty: Second-order Model, Anti-establishment Attitudes and the End of the Alternative European Utopia." West European Politics 29, no. 1 (2006). Iyengar, Shanto. "Agenda Setting and Beyond: Television News and the Strength of Political Issues." In Agenda Formation, edited by Wiliam Riker. Ann Arbor: University of Michigan Press, 1993. Iyengar, Shanto, and Donald Kinder. News That Matters: Television and American Opinion. Chicago: University of Chicago Press, 1987. Jacobs, Lawrence, and Robert Shapiro. "Issues, Candidate Image, and Priming: The Use of Private Polls in Kennedy's 1960 Presidential Campaign." American Political Science Review 88, no. 3 (1994): 527-40. Jahn, Detlef, and Ann-Sofie Storsved. "Legitimacy through Referendum? The Nearly Successful Domino-Strategy of the EU-Referendums in Austria, Finland, Sweden and Norway." West European Politics 18, no. 4 (1995): 18-37. Jauregui, Gurutz. "Basque Nationalism: Sovereignty, Independence and European Integration." In European Integration and the Nationalities Question, edited by John McGarry and Michael Keating. Abingdon: Routledge, 2006. Jerit, Jennifer. "Survival of the Fittest: Rhetoric during the Course of an Election Campaign." Political Psychology 25, no. 4 (2004): 563-75. Jerit, Jennifer, and Jason Barabas. "Bankrupt Rhetoric: How Misleading Information Affects Knowledge about Social Security." Public Opinion Quarterly 70, no. 3 (2006): 278-303.

351

Johnston, Richard, André Blais, Henry Brady, and Jean Crete. Letting the People Decide: Dynamics of a Canadian Election. Montreal: McGill-Queens University Press, 1992. Johnston, Richard, André Blais, Elisabeth Gidengil, and Neil Nevitte. The Challenge of Direct Democracy: The 1992 Canadian Referendum. Montreal: McGill - Queen's University Press 1996. Johnston, Richard, Michael Hagen, and Kathleen Hall Jamieson. The 2000 Presidential Election and the Foundations of Party Politics. Cambridge: Cambridge University Press, 2004. Katz, Richard. "Euroscepticism in Parliament: A Comparative Analysis of the European and National Parliaments." In Opposing Europe? The Comparative Party Politics of Euroscepticism Volume 2, edited by Aleks Szczerbiak and Paul Taggart. New York: Oxford University Press, 2008. Kerschen, Nicole. "The Welfare System of Luxembourg: From Past Dependency to the European Approach." In The Handbook of European Welfare Systems, edited by Klaus Schubert, Simon Hegelich and Ursula Bazant. Abingdon: Routledge, 2009. King, Gary, Robert Keohane, and Sidney Verba. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton: Princeton University Press, 1994. Kleinnijenhuis, Jan, Janet Takens, and Wouter van Atteveldt. "Toen Europa de Dagbladen Ging Vullen." In Nederlanders en Europa: Het Referendum over de Europese Grondwet, edited by Kees Aarts and Henk van der Kolk. Amsterdam: Utigeverij Bert Bakker, 2005. Kopstein, Jeffrey, and David Reilly. "Geographic Diffusion and the Transformation of the Postcommunist World." World Politics 53, no. 1 (2000). Lasserre Capdeville, Jérôme. Le Secret Bancaire: Étude de Droit Comparé: France, Suisse, Luxembourg. Aix-en-Provence: Presses Universitaires d’Aix-Marseille, 2006. Lazarsfeld, Paul, Bernard Berelson, and Hazel Gaudet. The People's Choice: How the Voter Makes Up His Mind in a Presidential Campaign. New York: Columbia University Press, 1944. Lechner, Frank. The Netherlands: Globalization and National Identity. New York: Routledge, 2008. LeDuc, Lawrence. "Opinion Change and Voting Behaviour in Referendums." European Journal of Political Research 41, no. 6 (2002): 711–32. ———. "Opinion Formation and Change in Referendum Campaigns." In The Dynamics of Referendum Campaigns: An International Perspective, edited by Claes De Vreese. Basingstoke: Palgrave MacMillan, 2007. ———. The Politics of Direct Democracy: Referendums in Global Perspective. New York: Broadview Press, 2003. ———. "Saving the Pound or Voting for Europe? Expectations for Referendums on the Constitution and the Euro." Journal of Elections, Public Opinion & Parties 15, no. 2 (2005): 169-96.

352

Lee, Chang Kil, and David Strang. "The International Diffusion of Public-Sector Downsizing: Network Emulation and Theory-Driven Learning." International Organization 60, (2006): 883-909. Lemennicier, Bertrand. "Political Polarization and the French Rejection of the European Constitution." European Journal of Political Economy 21, no. 4 (2005). Lewis-Beck, Michael. Economics and Elections: The Major Western Democracies. Ann Arbor: University of Michigan Press, 1988. Lijphart, Arend. "The Comparable-Cases Strategy in Comparative Research." Comparative Political Studies 8, no. 2 (1975). Lippmann, Walter. Public Opinion. New York: The Free Press, 1997 (1922). Loriaux, Michael. France after Hegemony: International Change and Financial Reform. Ithaca: Cornell University Press, 1991. Lubbers, Marcel. "Regarding the Dutch 'Nee' to the European Constitution: A Test of the Identity, Utilitarian and Political Approaches to Voting 'No'." European Union Politics 9, no. 1 (2008): 59-86. Lupia, Arthur. "Busy Voters, Agenda Control, and the Power of Information." The American Political Science Review 86, no. 2 (1992): 390-403. Madroñal, Juan Carlos. "Spanish Referendum on the EU Constitution: Monitoring Report." Mas Democracia and Democracy International Monitoring Reports (2005). Magleby, David. "Opinion Formation and Opinion Change in Ballot Proposition Campaigns." In Manipulating Public Opinion: Essays on Public Opinion as a Dependent Variable, edited by Michael Margolis and Gary Mauser. Belmont: Brooks/Cole Publishing Company, 1989. Marcus, George, John Sullivan, Elizabeth Theiss-Morse, and Daniel Stevens. "The Emotional Foundation of Political Cognition: The Impact of Extrinsic Anxiety on the Formation of Political Tolerance Judgments." Political Psychology 26, no. 6 (2005): 949-63. Marks, Gary, Liesbet Hooghe, Moira Nelson, and Erica Edwards. "Party Competition and European Integration in the East and West: Different Structure, Same Causality." Comparative Political Studies 39, no. 2 (2006). Marthaler, Sally. "The French Referendum on Ratification of the Constitutional Treaty " European Parties Elections and Referendums Network, Referendum Briefing Paper 12, (2005). Mathers, Andy. Struggling for a Social Europe: Neoliberal Globalization and the Birth of a European Social Movement. Aldershot: Ashgate, 2007. McAdam, Doug. "“Initiator” and “Spin-off” Movements: Diffusion Processes In Protest Cycles." In Repertoires and Cycles of Collective Action, edited by Mark Traugott. Durham: Duke University Press, 1995. McAdam, Doug, John McCarthy, and Mayer Zald. Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structures and Cultural Framings. Cambridge: Cambridge University Press, 1996.

353

McAdam, Doug, and Dieter Rucht. "The Cross National Diffusion of Movement Ideas." Annals of the American Academy of Political and Social Science 528, no. 36-59 (1993). McAdam, Doug, Sidney Tarrow, and Charles Tilly. Dynamics of Contention. Cambridge: Cambridge University Press, 2001. McAllister, Ian. "Elections without Cues: The 1999 Australian Republic Referendum." Australian Journal of Political Science 36, no. 2 (2001): 247-69. McCombs, M, and D Shaw. "The Agenda-Setting Function of Mass Media." Public Opinion Quarterly 36, no. 2 (1972): 176. McLaren, Lauren. "Public Support for the European Union: Cost/Benefit Analysis or Perceived Cultural Threat?" The Journal of Politics 64, no. 2 (2002). Medrano, Juan Diez. Framing Europe: Attitudes to European Integration in Germany, Spain, and the United Kingdom. Princeton: Princeton University Press, 2003. Menendez-Alarcon, Antonio. The Cultural Realm of European Integration: Social Representations in France, Spain, and the United Kingdom. Westport: Greenwood Publishing Group, 2004. Meyer, Christoph. "Political Legitimacy and the Invisibility of Politics: Exploring the European Union's Communication Deficit." Journal of Common Market Studies 37, no. 4 (1999): 617-39. Milner, Henry. "YES to the Europe I Want; NO to This One? Some Reflections on France's Rejection of the EU Constitution." Political Science and Politics 39, no. 2 (2006). Milner, Susan. "Protection, Reform, and Political Will: France and the European Social Model." In French Relations with the European Union, edited by Helen Drake. Abingdon: Routledge, 2005. Muus, Philip. "The Netherlands: A Pragmatic Approach to Economic Needs and Humanitarian Considerations." In Controlling Immigration: A Global Perspective, edited by Wayne Cornelius, Takeyuki Tsuda, Philip Martin and James Hollifield. Stanford: Stanford University Press, 2004. Myers, Daniel "The Diffusion of Collective Violence: Infectiousness, Susceptibility, and Mass Media Networks." American Journal of Sociology 106, no. 1 (2000): 173-208. Nadeau, Richard, and Richard Niemi. "Educated Guesses: The Process of Answering Factual Knowledge Questions in Surveys." Public Opinion Quarterly 59, no. 3 (1995): 323-46. Nadeau, Richard, Francois Petry, and Eric Belanger. "Issue-Based Strategies in Election Campaigns: The Case of Healthcare in the 2000 Canadian Federal Election." Political Communication (2010). Nelson, Thomas, Rosalee Clawson, and Zoe Oxley. "Media Framing of a Civil Liberties Conflict and its Effect on Tolerance." American Political Science Review 91, no. 3 (1997): 567-83. Newton, Gerald, ed. Luxembourg and Lëtzebuergesch: Language and Communication at the Crossroads of Europe. Oxford: Oxford University Press, 1996.

354

Nijeboer, Arjen. "The First Dutch Referendum: A Pre-Ballot Assessment." Notre Europe no. 18 (2005). ———. "Peoples’ Vengeances: The Dutch Referendum." European Constitutional Law Review 1, (2005): 393-405. Norris, Pippa. "Representation and the Democratic Deficit." European Journal of Political Research 32, no. 2 (1997): 273-82. Norris, Pippa, John Curtice, David Sanders, Margaret Scammell, and Holli Semetko. On Message: Communicating the Campaign. London: Sage, 1999. Pelkmans, J, and S van Kessel. "Encapsulating Services in the 'Polder': The Bolkestein Directive in Dutch Politics." WRR Scientific Council for Government Policy Web Publications no. 31 (2007). Perea, Eva Anduiza. "Who Abstained and Why? Voter Turnout for the Referendum on the Treaty to Establish a European Constitution." Elcano Royal Institute Working Paper 34, (2005). Perni, Orietta. "Migration in Spain from Aznar to Zapatero: A Real Change of Direction?" In International Studies Association 47th Annual Convention. San Diego, California, USA. Perrineau, Pascal. "The Conditions for the Re-emergence of an Extreme Right Wing in France: The National Front, 1984-1998." In The Development of the Radical Right in France: From Boulanger to Le Pen, edited by Edward Arnold. Basingstoke: Macmillan Press, 2000. Pfau, Michael, and Henry Kenski. Attack Politics: Strategy and Defense. New York: Praeger, 1990. Piar, Christophe, and Jacques Gerstlé. "Le Cadrage du Référendum sur la Constitution Européenne: La Dynamique d’une Campagne à Rebondissements." In Le Référendum de Ratification du Traité Constitutionnel Européen du 29 Mai 2005: Comprendre le "Non" Français, edited by Annie Laurent and Nicolas Sauger. Paris: CEVIPOF, 2005. Polletta, Francesca, and M. Kai Ho. "Frames and their Consequences." In The Oxford Handbook of Contextual Political Analysis, edited by Robert Goodin and Charles Tilly. New York: Oxford University Press, 2006. Pondelinger, Germain. "Country Report: ELA – Luxembourg." European Journal for Education Law and Policy 3, (1999): 151–53. Pratto, Felicia, and Oliver John. "Automatic Vigilance: The Attention-Grabbing Power of Negative Social Information." Journal of Personality and Social Psychology 61, no. 3 (1991): 380–91. Qvortrup, Matt. "The Three Referendums on the European Constitution Treaty in 2005." The Political Quarterly 77, no. 1 (2006). Rasinski, Kenneth. "The Effect of Question Wording on Public Support for Government Spending." Public Opinion Quarterly 53, no. 3 (1989): 388- 94. Ray, Leonard. "Reconsidering the Link between Incumbent Support and Pro-EU Opinion." European Union Politics 4, no. 3 (2003): 259-79.

355

Reif, Karlheinz, and Hermann Schmitt. "Nine Second-Order National Elections: A Conceptual Framework for the Analysis of European Election Results." European Journal of Political Research 8, no. 1 (1980). Ricard-Nihoul, Gaëtane. "The French 'No' Vote on 29 May 2005: Understanding and Action." Notre Europe 44, (2005). Ricard-Nihoul, Gaëtane, and Morgan Larhant. "How to Explain the Unexpected: An Assesment of the French Constitutional Referendum." Elcano Royal Institute Working Paper 57, (2005). Richard, Arsène, and Ronald Pabst. "Evaluation of the French Referendum on the EU Constitution, May 2005." Democracy International Monitoring Reports (2005). Riker, Wiliam, ed. Agenda Formation. Ann Arbor: University of Michigan Press, 1993. ———. The Strategy of Rhetoric: Campaigning for the American Constitution. New Haven: Yale University Press, 1996. Roes, Theo, ed. Facts and Figures of the Netherlands: Social and Cultural Trends 1995-2006. The Hague: SCP -The Netherlands Institute for Social Research 2008. Rogers, Everett. Diffusion of Innovations. Fourth ed. New York: New York: Free Press, 1995. Royo, Sebastian. Varieties of Capitalism in Spain: Remaking the Spanish Economy for the New Century. New York: Palgrave Macmillan, 2008. Rude, George The Crowd in History, 1730-1848. New York: John Wiley, 1964. Schmidt, Vivien. "Does Discourse Matter in the Politics of Welfare State Adjustment?" Comparative Political Studies 35, no. 2 (2002). ———. The Futures of European Capitalism. Oxford: Oxford University Press, 2002. ———. "The Politics of Economic Adjustment in France and Britain: When Does Discourse Matter?" Journal of European Public Policy 8, no. 2 (2001): 247-64. Schuck, Andreas. "Referendum Campaign Dynamics: News Media, Campaign Effects and Direct Democracy." University of Amsterdam, 2005. Schuck, Andreas, and Claes de Vreese. "The Dutch No to the EU Constitution: Assessing the Role of EU Skepticism and the Campaign." Journal of Elections, Public Opinion, and Parties 18, no. 1 (2008): 101-28. Scuto, Denis. "EUDO Citizenship Observatory Country Report: Luxembourg." European University Institute Robert Schuman Center for Advanced Studies (2009). Shields, J.G. The Extreme Right in France: From Pétain to Le Pen. Abingdon: Routledge, 2007. Sides, John. "The Origins of Campaign Agendas." British Journal of Political Science 36, no. 3 (2006): 407-36. Simmons, Beth, and Zachary Elkins. "The Globalization of Liberalization: Policy Diffusion in the International Political Economy." American Political Science Review 98, no. 01 (2004).

356

Simon, Adam. The Winning Message: Candidate Behavior, Campaign Discourse, and Democracy. Cambridge: Cambridge University Press, 2002. Sinner, Michele. "Swiss, Luxembourg, Austria Defend Bank Secrecy." Reuters (2009), http://www.reuters.com/article/idUSL865351220090308. Siune, Karen, Palle Svensson, and Ole Tonsgaard. "The EU: The Danes Said “No” in 1992, But “Yes” in 1993: How and Why?" Electoral Studies 13, no. 2 (1994). Smith, Timothy B. France in Crisis: Welfare, Inequality and Globalization since 1980. Cambridge: Cambridge University Press, 2004. Sniderman, Paul, and Sean Theriault. "The Structure of Political Argument and the Logic of Issue Framing." In Studies in Public Opinion: Attitudes, Nonattitudes, Measurement Error, and Change, edited by Willem Saris and Paul Sniderman. Princeton: Princeton University Press, 2004. Snow, David, and Robert Benford. "Alternative Types of Cross-National Diffusion In the Social Movement Arena." In Social Movements in a Globalizing World, edited by Donatella Della Porta, Hanspeter Kriesi and Dieter Rucht. London: Macmillan, 1999. ———. "Ideology, Frame Resonance, and Participant Mobilization." International Social Movement Research 1, (1988). Snow, David, E. Burke Rochford Jr, Steven Worden, and Robert Benford. "Frame Alignment Processes, Micromobilization, and Movement Participation." American Sociological Review 51, (1986): 464-81. Soroka, Stuart. Agenda-Setting Dynamics in Canada. Vancouver: UBC Press, 2002. Soule, Sarah. "Diffusion Processes within and across Movements." In The Blackwell Companion to Social Movements, edited by David Snow, Sarah Soule and Hanspeter Kriesi. Malden: Blackwell Publishing, 2003. Spilerman, Seymour. "Structural Characteristics of Cities and the Severity of Racial Disorders." American Sociological Review 41, no. 771-793 (1976). Startin, Nick. "Maastricht, Amsterdam, and beyond: The Troubled Evolution of the French Right." In French Relations with the European Union, edited by Helen Drake. Abingdon: Routledge, 2005. STATEC National Statistical Institute of Luxembourg. "Economic and Social Portrait of Luxembourg." Information and Press Service of the Luxembourg Government and Ministry of Foreign Affairs, 2003. Stefanova, Boyka. "The 'No' Vote in the French and Dutch Referenda on the EU Constitution: A Spillover of Consequences for the Wider Europe." Political Science and Politics 39, no. 2 (2006). Stell, Gerald. Luxembourgish Standardization: Context, Ideology and Comparability with the Case of West Frisian. Leuven: Peeters Louvain- La-Neuve, 2006. Svensson, Palle. "Five Danish Referendums on the European Community and European Union: A Critical Assessment of the Franklin Thesis." European Journal of Political Research 41, no. 6 (2002).

357

Szczerbiak, Aleks, and Paul Taggart. "The Politics of European Referendum Outcomes and Turnout: Two Models." West European Politics 27, no. 4 (2004). Taggart, Paul. "Keynote Article: Questions of Europe - The Domestic Politics of the 2005 French and Dutch Referendums and their Challenge for the Study of European Integration." Journal of Common Market Studies 44, no. 1 (2006). ———. "A Touchstone of Dissent: Euroscepticism in Contemporary Western European Party Systems " European Journal of Political Research 33, no. 3 (1998). ———. "A Touchstone of Dissent: Euroscepticism in Contemporary Western European Party Systems " European Journal of Political Research 33, no. 3 (1998): 363-88. Taggart, Paul, and Aleks Szczerbiak. "Contemporary Euroscepticism in the Party Systems of the European Union Candidate States of Central and Eastern Europe." European Journal of Political Research 43, no. 1 (2004). ———. "The Party Politics of Euroscepticism in EU Member and Candidate States." Opposing Europe Research Network Working Paper 6, (2002). Takens, Janet. "A Negative Campaign? A Study of the News Coverage of the European Constitution and its Effects." Free University of Amsterdam, 2006. Tarrow, Sidney. "Bridging the Quantitative-Qualitative Divide in Political Science." The American Political Science Review 89, no. 2 (1995). ———. The New Transnational Activism. Cambridge: Cambridge University Press, 2005. ———. Power in Movement: Social Movements and Contentious Politics. Cambridge: Cambridge University Press, 1998. The National Portal for the Grand-Duchy of Luxembourg. (2010), http://www.luxembourg.public.lu/fr/societe/population/societe- multiculturelle/index.html. Thomassen, Jacques. "Nederlanders en Europa. Een Bekoelde Liefde?" In Nederlanders en Europa: Het Referendum over de Europese Grondwet, edited by Kees Aarts and Henk van der Kolk. Amsterdam: Utigeverij Bert Bakker, 2005. Tilly, Charles. Contention and Democracy in Europe, 1650-2000. Cambridge: Cambridge University Press, 2004. Torreblanca, José Ignacio. "Farewell to Funds? Keys to Understanding Spain’s Position When Negotiating the 2007-13 EU Budget." Elcano Royal Institute Working Paper 21, (2005). Torreblanca, José Ignacio, and Alicia Sorroza. "Spanish Ratification Monitor." Elcano Royal Institute Working Paper 8, (2005). Tversky, Amos, and Daniel Kahneman. "Rational Choice and the Framing of Decisions." In The Limits of Rationality, edited by Karen Cook and Margaret Levi, ix, 426 p. Chicago: University of Chicago Press, 1990. United States Department of State. "Country Reports." (2010), http://www.state.gov/p/eur/ci/lu/.

358

Vallvé, José Luis González. "España y Francia Frente al Referéndum Constitucional: Entre la Inquietud y la Esperanza." Elcano Royal Institute Working Paper 55, (2005). van Noije, Lonneke. The Democratic Deficit Closer to Home: Agenda-building Relations between Parliament and the Press, and the Impact of European Integration, in the United Kingdom, the Netherlands and France. The Netherlands 2007. van Oorschot, Wim. "The Dutch Welfare System: From Collective Solidarity towards Individual Responsiblity." In The Handbook of European Welfare Systems, edited by Klaus Schubert, Simon Hegelich and Ursula Bazant. Abingdon: Routledge, 2009. West, Darrell. Air Wars: Television Advertising in Election Campaigns, 1952- 2004. Washington D.C.: CQ Press, 2005. WRR Scientific Council for Government Policy. Rediscovering Europe in the Netherlands. Amsterdam: Amsterdam University Press, 2007. Zaller, John. The Nature and Origins of Mass Opinion. New York: Cambridge University Press, 1992.

359