¹ Karlheinz Reif and Hermann Schmitt, ‘Nine Second-Order National Elections: A Conceptual Framework for the Analysis of European Election Results’, European Journal for Political Research, 8 (1980), 3–44; Cees van der Eijk and Mark Franklin, Choosing Europe? The European Electorate and National Politics in the Face of Union (Ann Arbor: The University of Michigan Press, 1996); Michael Marsh, ‘Testing the Second-Order Election Model After Four European Elections’, British Journal of Political Science, 28 (1998), 591–607; Clifford J. Carrubba and Richard J. Timpone, ‘Explaining Vote Switching Across First and Second Order Elections: Evidence from Europe’, Comparative Political Studies, 38 (2005), 260–81; Simon Hix and Michael Marsh, ‘Punishment or Protest? Understanding European Parliament Elections’, Journal of Politics, 69 (2007), 495–510; Wouter van der Brug and Cees van der Eijk, European Elections and Domestic Politics: Lessons from the Past and Scenarios for the Future (Notre Dame: University of Notre Dame Press, 2007).

² Federico Ferrara and J. Timo Weishaupt, ‘Get Your Act Together: Party Performance in European Parliament Election’, European Union Politics, 5 (2004), 283–306; Michael Marsh, ‘European Parliament Elections and Losses by Governing Parties’, in Van der Brug and Van der Eijk, European Elections and Domestic Politics, pp. 51–72; Hix and Marsh, ‘Punishment or Protest?’

³ Reif and Schmitt, ‘Nine Second-Order National Elections’. See also Karlheinz Reif, ‘National Election Cycles and European Elections, 1979 and 1984’, Electoral Studies, 3 (1984), 244–55; Van der Eijk and Franklin, Choosing Europe; Marsh, ‘Testing the Second-Order Election Model After Four European Elections’.

⁴ Angus Campbell, ‘Surge and Decline: A Study of Electoral Change’, Public Opinion Quarterly, 24 (1960), 397–418.

⁵ For the former, see Reif, ‘National Election Cycles and European Elections’; Marsh, ‘Testing the Second-Order Election Model’. For the latter, see Edward Tufte, ‘Determinants of the Outcomes of Midterm Congressional Elections’, American Political Science Review, 69 (1975), 812–26; Morris P. Fiorina, Retrospective Voting in American National Elections (New Haven, Conn.: Yale University Press, 1981); Thaddeus Kousser, ‘Retrospective Voting and Strategic Behaviour in European Parliament Elections’, Electoral Studies, 23 (2004), 1–21.

⁶ Van der Eijk and Franklin, Choosing Europe, p. 364.

⁷ Reif and Schmitt, ‘Nine Second-Order National Elections’, p. 10.

⁸ Marsh, ‘European Parliament Elections and Losses by Governing Parties’, p. 70.

⁹ Carrubba and Timpone, ‘Explaining Vote Switching Across First and Second Order Elections’.

¹⁰ Ferrara and Weishaupt, ‘Get Your Act Together’; Hix and Marsh, ‘Punishment or Protest?’.

¹¹ Hix and Marsh, ‘Punishment or Protest?’, p. 507.

¹² Ferrara and Weishaupt, ‘Get Your Act Together’, p. 301.

¹³ Mark Franklin and Christopher Wlezien, ‘The Responsive Public: Issue Salience, Policy Change, and Preferences for European Unification’, Journal of Theoretical Politics, 9 (1997), 347–63; Geoffrey Evans, ‘Europe: A New Electoral Cleavage?’ in Geoffrey Evans and Pippa Norris, eds, Critical Elections: British Parties and Voters in Long-Term Perspective (London: Sage Publications, 1999), pp. 207–22; Erik R. Tillman, ‘The European Union at the Ballot Box: European Integration and Voting Behaviour in the New Member States’, Comparative Political Studies, 27 (2004), 590–610; Geoffrey Evans and Sarah Butt, ‘Leaders or Followers? Parties and Public Opinion on the European Union’, in Alison Park, John Curtice, Katarina Thomson, Catherine Bromley, Miranda Phillips and Mark Johnson, eds, British Social Attitudes: The 22nd Report (London: Sage, 2005), pp. 197–11.

¹⁴ Anthony Downs, An Economic Theory of Democracy (New York: Harper and Row, 1957); James Enelow and Melvin J. Hinich, The Spatial Theory of Voting (Cambridge: Cambridge University Press, 1984); George Rabinowitz and Stuart Elaine Macdonald, ‘A Directional Theory of Issue Voting’, American Political Science Review, 83 (1989), 93–121.

¹⁵ Moreover in 1999, over 80 per cent of respondents said that European integration is of ‘some’ or ‘great’ importance (a similar question was not asked in 2004).

¹⁶ David RePass, ‘Issue Salience and Party Choice’, American Political Science Review, 65 (1971), 389–400; George Rabinowitz, James W. Prothro and William Jacoby, ‘Salience as a Factor in the Impact of Issues on Candidate Evaluation’, Journal of Politics, 44 (1982), 41–63; Jon A. Krosnick, ‘The Role of Attitude Importance in Social Evaluation: A Study of Policy Preferences, Presidential Candidate Evaluation, and Voting Behavior’, Journal of Personality and Social Psychology, 55 (1988), 196–210.

¹⁷ We should note that some recent research suggests that the issue of European integration can also play a role in the vote decision at some national elections, for example, see Tillman ‘The European Union at the Ballot Box’.

¹⁸ Liesbet Hooghe, Gary Marks and Carole J. Wilson, ‘Does Left/Right Structure Party Positions on European Integration?’, Comparative Political Studies, 35 (2002), 965–89.

¹⁹ Gary Marks and Marco Steenbergen, 1999 Expert Survey on National Parties on National Parties and the European Union (Center for European Studies, University of North Carolina, 2002); Kenneth Benoit and Michael Laver, Party Policy in Modern Democracies (London: Routledge, 2006).

²⁰ In the results section, we specifically discuss the relationship between Euroscepticism and defection in the three countries where governing parties are more Eurosceptic than opposition parties.

²¹ Jan Kleinnijenhuis and Jan A. de Ridder, ‘Issue News and Electoral Volatility’, European Journal of Political Research, 33 (1998), 413–37; Pippa Norris, John Curtice, David Sanders, Margaret Scammell and Holli Semetko, On Message: Communicating the Campaign (London: Sage, 1999); Claes de Vreese and Holli A. Semetko, ‘News Matters: Influences on the Vote in the Danish 2000 Euro Referendum Campaign’, European Journal of Political Research, 43 (2004), 699–722; Sara B. Hobolt, ‘How Parties Affect Vote Choice in European Integration Referendums’, Party Politics, 12 (2006), 623–47.

²² Stephen Ansolabehere, Shanto Iyengar, Adam Simon and Nicholas Valentino, ‘Does Attack Advertising Demobilize the Electorate?’, American Political Science Review, 88 (1994), 829–38.

²³ Stephen E. Finkel and John G. Geer, ‘A Spot Check: Casting Doubt on the Demobilization Effect of Attack Advertising’, American Journal of Political Science, 42 (1998), 573–95; Martin P. Wattenberg and Craig L. Brians, ‘Negative Campaign Advertising: Demobilizer or Mobilizer?’, American Political Science Review, 93 (1999), 891–900.

²⁴ These distances have been calculated by subtracting the mean self-placement of respondents voting for a governing party in the last national election on the EU and left–right dimensions (10-point scales) from the mean position of the governing party/parties, according to party expert surveys (see Appendix Table 4A).

²⁵ It should be noted that unlike much of the existing literature using the EES data (for example, see van der Eijk and Franklin, Choosing Europe) our analyses are not performed on a stacked dataset. Since we are interested in the vote choice at EP elections of people who are able to defect from the government, we restrict our sample to respondents who voted for a governing party in the last national election. This means our analyses explicitly focus on how the attributes of governing parties explain defection. Of course, this particular methodological choice may explain some of the differences in the importance of the EU issue that we find compared to past studies that have tended to examine vote choice in the wider setting of all voters and all parties.

²⁶ Our measure of previous vote choice is simply recalled vote, as the EES is not a longitudinal survey. Although there are problems with using recalled vote, since voters like to be consistent and therefore remember voting for the same party that they just voted for, this should make our tests more conservative, as rates of defection will be lower than we would expect.

²⁷ We only have 23 countries in the dataset for 2004 as Malta was not included in the survey and the questionnaire in Lithuania did not include a question on vote choice in the last national election.

²⁸ Of the 23 countries, Belgium, Cyprus, Greece, Italy and Luxembourg all have compulsory voting. The level of enforcement and degree of sanction for not voting varies greatly across the countries.

²⁹ Marco Steenbergen and Bradford S. Jones, ‘Modeling Multilevel Data Structures’, American Journal of Political Science, 46 (2002), 218–37.

³⁰ It can be argued that a random effects model is inappropriate in this case because countries cannot be regarded as a sample of a population. Rather, they should be studied as unique entities, and hence a fixed-effects model is the more appropriate choice. Since we are not interested in the individual countries per se, but rather wish to draw general inferences about the effect of individual and contextual variables on voting behaviour, a random effects model would seem more appropriate (see Tom A.B. Snijders and Roel J. Boskers, Multilevel Analysis: An Introduction to Basic and Advanced Multilevel Modelling (London: Sage, 1999), p. 43). We have also estimated the models as simple binary logistic regressions (correcting the standard errors for clustering within countries), and the results are very similar to the hierarchical linear model (HLM) estimates.

³¹ Stephen W. Raudenbush and Anthony S. Bryk, Hierarchical Linear Models: Applications and Data Analysis Methods, 2nd edn (London: Sage, 2002); Snijders and Boskers, Multilevel Analysis.

³² The models have also been estimated using a multinomial logit (and probit) link, and the results are very similar, with none of the substantive results (significance or magnitude) differing widely. For a discussion of binary versus multinomial logit models, see Michael R. Alvarez and Jonathan Nagler, ‘When Politics and Models Collide: Estimating Models of Multiparty Elections’, American Journal of Political Science, 42 (1998), 55–96.

³³ We have estimated the same model using mean voter placement of parties rather than party expert surveys and the results are almost identical. We have opted to use party expert surveys, because they provide a more objective measure of party placement with less measurement error and less missing data (see Benoit and Laver, Party Policy in Modern Democracies). Moreover, by using expert placements rather than voter placements we reduce potential problems of endogeneity. We recognize, however, that this may not completely rule out the possibility that some endogeneity exists in our measurement of the EU distance variable via the self-placement scores.

³⁴ Other individual level controls were also included in previous model estimations, such as religion, religiosity, size of town, issue salience and satisfaction with the government’s performance on the environment and immigration, but none of these variables were statistically significant when the other controls were included and, hence, they are excluded from the final models shown here.

³⁵ See van der Eijk and Franklin, Choosing Europe.

³⁶ Missing data is generally a problem when using survey data, and in our coding choices we have therefore tried to exclude as few cases as possible. In order to check that our results are not biased due to missing data, we ran a similar analysis after imputing all missing values (using multiple imputation). The results are almost identical to the ones reported here.

³⁷ Michael S. Lewis-Beck, Economics and Elections: The Major Western Democracies (Ann Arbor: University of Michigan Press, 1988); Michael S. Lewis-Beck and Mary Stegmaier, ‘Economic Determinants of Electoral Outcomes’, Annual Review of Political Science, 3 (2000), 183–219.

³⁸ Of course similar concerns could be voiced with regard to economic perceptions, see for example: Geoffrey Evans and Robert Andersen, ‘The Political Conditioning of Economic Perceptions’, Journal of Politics, 68 (2006), 194–207; James Tilley, John Garry and Tessa Bold, ‘Perceptions and Reality: Economic Voting in the European Union’, European Journal of Political Research, 47 (2008), 665–86.

³⁹ Claes de Vreese, Susan A. Banducci, Holli A. Semetko and Hajo G. Boomgaarden, ‘The News Coverage of the 2004 European Parliamentary Election Campaign in 25 Countries’, European Union Politics, 7 (2006), 477–504; Claes de Vreese, Edmund Lauf and Jochen Peter, ‘The Media and European Parliament Elections: Second-Rate Coverage of a Second-Order Event?’ in van der Brug and van der Eijk, eds, European Elections and Domestic Politics (Notre Dame: University of Notre Dame Press, 2007), pp. 116–30.

⁴⁰ De Vreese et al., ‘The Media and European Parliament Elections’. The 2004 media study focuses exclusively on EU coverage, whereas the 1999 study also includes stories of domestic and other international actors. Our estimates from 1999 include the mean of the ‘tone’ across all actors, but very similar results were found when only coverage of EU actors was included. We use the ‘all actors’ measure as data from Germany and Ireland are missing for the EU only measure.

⁴¹ The raw scores range from −1 to +1 (2004) and −100 to +100 (1999).

⁴² Reif, ‘National Election Cycles and European Elections’; Marsh, ‘Testing the Second-Order Election Model After Four European Elections’.

⁴³ These countries are the Czech Republic, Estonia, Hungary, Latvia, Poland, Slovakia and Slovenia.

⁴⁴ See, in particular, Hix and Marsh, ‘Punishment or Protest’.

⁴⁵ In 1999, 6 per cent of voters were located seven or more points away from the governing party that they previously voted for on the European integration dimension. In 2004, 4 per cent of voters were located more than 7 points away from their party.

⁴⁶ This is calculated for a 40 year old middle-class non-partisan of a mean-sized party (29 per cent of seats), who does not approve of the government, thinks the economy has got neither better nor worse, and holds the same position as his party on the left–right and EU dimensions. This voter is voting in an election taking place two years after the last general election in an established democracy.

⁴⁷ The values for the different campaign contexts have been chosen on the basis of the actual range of campaign tone values in the two elections: 0.5 was the most positive campaign tone in 2004 and the most negative score was −1.0. In 1999, the most positive campaign was scored 2 and the most negative campaign was scored −1.1.