Introduction

Abstract

The present contribution to the scholarly knowledge about European Parliament (EP) elections essentially assembles analyses of the data of the 2004 European Election Study. Due to the very nature of large‐scale cross‐national comparative survey research, it will only be published shortly after the 2009 European Parliament election. As it would certainly be inadequate to ignore the fact that another European election took place just before publication, this introduction to the analyses that follow will start out with a first inspection of the results of this most recent European election—the election of the members of the European Parliament in June 2009. The main question we will be asking is about the persistence of the ‘second‐order’ character of these elections: are the 2009 European Parliament election results in line with our expectations about European Parliament elections as second‐order national elections as laid out originally by Reif and Schmitt (1980) and restated and refined in a large number of subsequent publications? The second part of this introduction will then go on and do what every introduction does: briefly present the contributions that follow.

The 2009 Election to the European Parliament: Still Second‐Order?

European Parliament elections have been described as second‐order national elections (SOEs). This judgement was inspired by the fact that the national political arena has been so much more important than that of the European Union — at least in the early years of direct elections to the European Parliament. Under these circumstances, turnout was expected to be lower in SOEs; government parties were expected to lose votes in SOEs, and those losses were predicted to follow the national electoral cycle; small parties were expected to win and ideologically extreme or polar parties were also expected to win relative to their previous first‐order election (FOE) result (Reif and Schmitt 1980). A re‐examination of these aggregate predictions at the occasion of the 2004 EP election broadly confirmed them. The one major qualification to this conclusion was that things at the time worked out differently in the new Eastern member countries (Schmitt 2005). A tentative explanation of these deviant findings referred to the relative fluidity of the party and electoral systems in the post‐communist democracies of Eastern Europe.

We are interested here in the second‐order character of European Parliament elections five years later. Was turnout in the 2009 European Parliament elections again lower if compared to the previous national FOE in the respective country? Did government parties lose and was the magnitude of those possible losses governed by the relative position of an EP election in the national electoral cycle? And last but not least, did small parties — and ideologically more extreme parties — two qualities which go together quite regularly — do systematically better in European Parliament elections?

Turnout was Lower

Turnout is expected to be lower in second‐order elections generally, and in European Parliament elections in particular. One line of reasoning refers to the comparatively lower level of political mobilisation and stimulation (e.g. Campbell 1960). Specifically for EP elections, an alternative hypothesis has been proposed which stipulates that low turnout is a function of a lack of support for and legitimacy of the EU level of government (e.g. Blondel et al. 1998). While this second reasoning has found little empirical support so far (e.g. Van der Eijk and Schmitt 2009), it can not be ruled out as a matter of principle. A third plausible motivation for low turnout in SOE is dissatisfaction with one’s last FOE vote choice (Schmitt et al. 2009). But we are not looking at the motivations of individual turnout decisions here — this has to wait for future analyses due to the non‐availability of survey data at the time of writing. What we can do instead is to compare aggregate turnout levels for national first‐order elections with those for the EP election of 2009.

Figure 1 shows that turnout was again lower in the European Parliament election of June 2009 compared to the preceding national first‐order election. Belgium and Luxembourg are the only two exceptions among the 27 member countries. There turnout is about at the same level as in national first‐order elections. This is due to the Belgian system of compulsory voting and the fact that, in Luxembourg, European Parliament elections are held concurrently with national parliament elections.

Figure 1 Turnout is Lower in the 2009 European Parliament Elections.

In all other 25 EU member‐countries, turnout was lower than it was in the previous FOE. How much lower it was depends to a considerable degree on the turnout level that tends to be achieved in a FOE (as indicated by an R square of 0.68). This is to say that low FOE turnout rates (as found in Lithuania, Poland, Romania and Slovakia) produce record‐low EP participation rates, while high FOE turnout rates (as in Cyprus, Denmark, Italy and Malta) go hand in hand with relatively high EP turnout levels. Note however the asymmetric positioning of the regression slope relative to the (dotted) diagonal of the graph: this indicates that low turnout countries suffer somewhat more from the SOE turnout loss than high turnout countries do.

Government Parties Lost — Most of Them at Least

Logically, there are two and only two reasons why government parties can be expected to lose in SOE — one strategic, the other sincere. The strategic motivation for abandoning one’s FOE choice of a government party has to do with its performance since the FOE: voters might be dissatisfied with it and either abstain (the softer signal) or switch to another party (the harder signal) as a result of it. The sincere motivation of switching to other than the government party that a voter had chosen at the previous FOE would again be arena‐specific: a dissatisfaction with the arena‐specific — i.e. EU — policies of the previous vote choice at the time of the SOE (Schmitt et al. 2009).

Again, because of the lack of suitable survey data at the time of writing, we cannot dig into the motivations of individual vote choices. However, we can and will inspect the distributions of 27 member countries’ aggregate results. Figure 2 shows how national government parties did in the European Parliament election of June 2009. Relative to their last FOE, national governments lost. There are only two clear exceptions to this rule: the governments of Poland and Finland did better in the EP election than they did in the previous FOE.

Figure 2 Government Parties Lose the 2009 European Parliament Elections.

At the time of the 2009 EP election, the semi‐presidential system of Poland was stuck in a situation of divided government in which an outright EU‐sceptical president Lech Kaczynski (Law and Justice, PiS) was confronting a somehow more EU‐phoric prime minister Donald Tusk (Civic Platform, PO). This battle was on since Tusk won the last national legislative election, and the slight increase (in the proportion of valid votes) of Tusk in the EP election can perhaps be understood as a sincere expression of preference among the electorate for co‐operation rather than confrontation with the European Union.

The situation in Finland was similar in a way. Here as well, the strongly Euro‐positive National Coalition Party won while the senior government partner — the Centre Party — lost votes. This could be understood again as a sincere expression of support for a more Euro‐positive policy of Finland.

Examples of sizable defeats of national government parties are more numerous. We will pick just three of them, all concerning a socialist‐ or labour‐led government. The British Labour party, the Hungarian Socialist Party (MSZP), and the Bulgarian Socialist Party (BSP) all suffered a disastrous defeat which, in the case of Bulgaria, has already led to a change of the leading national government party. The political reasons for such near‐collapses are likely to be domestic rather than European. But again, micro‐level survey evidence will be needed to arrive at a firmer conclusion here.

Did Government Losses Follow the National Electoral Cycle?

National electoral cycles are referring to a particular pattern of government popularity. Shortly after a FOE, government popularity is expected to increase even beyond the level of support received in the election (‘honeymoon’ or post‐electoral euphoria), in order to more or less continuously decrease until sometime after midterm, and to rise again to a new level of support at the next FOE. This characteristic pattern of government popularity has been successful in predicting the level of support national governments manage to receive in European Parliament elections that all take place, in the various member countries, at a different point in time of the national electoral cycle (e.g. Reif 1984; Marsh 1998; Schmitt 2005).

Did government losses in the 2009 European Parliament elections again follow the national electoral cycle? The short answer is no. As Figure 3 demonstrates, this time we do not find a trace of a cyclical pattern in government losses, even if we control for possible distorting effects in the political makeup of the political systems of the new Eastern member countries of the European Union. All we can say on the basis of this joint distribution is that government losses tend to be graver the later an EP election is held in a national legislative period. Based on an R square of 0.313, however, even this generalisation is far from deterministic.

Figure 3 National Government Losses as a Function of the Timing of EP Elections in the National Electoral Cycle.

Figure 4 Small Parties did Better in the 2009 European Parliament Elections.

What does it mean, then, that there is no cyclical element left in the evolution of support for government parties in European Parliament elections? It may mean two different things. First, government support in the EU member countries might not follow a cyclical pattern anymore. The evolution of government popularity as identified in some sort of monthly barometer surveys would be useful to test this proposition. There is a second possible explanation which — should it be confirmed — would be very consequential for our understanding of the nature of European Parliament elections. It goes like this: the absence of cyclical elements in national government losses may also mean that EP electoral behaviour is less determined by national politics, and that these European elections are therefore gradually losing their second‐order character. Again, we need micro‐level survey evidence to investigate this second possible explanation.

Did Small Parties Gain Relative to their Previous FOE Result?

Because there is less at stake in European Parliament elections, there is also less of an incentive to vote strategically — i.e. to choose another than the most preferred party. Strategic voting favours larger parties over smaller ones because they can more effectively contribute to government formation. However, government formation — in the conventional sense at least –is not an issue in European Parliament elections. This is why small parties are expected to win in European Parliament elections relative to their last FOE result (Schmitt et al. 2009). We compare the effective number of electoral parties (according to Laakso and Taagepera 1979) for the EP election result and the preceding FOE result, and expect that the index is systematically higher for the EP election.

This expectation is barely confirmed by the results of the 2009 European Parliament elections. In a few countries — Belgium, Lithuania and Malta — we find hints in the opposite direction, that is: that big parties were doing better. Moreover, in 11 other countries a party’s electoral fortunes in EP elections are hardly affected by their size, and the effective number of electoral parties does not differ much between EP elections and the preceding FOE. This leaves us with 13 countries in which small parties did substantially better in EP elections, just below the simple majority of EU member countries. This is again an indication that the second order election model of European Parliament elections is somehow put into question by these 2009 election results — some of its theoretical underpinnings at least, if not the whole model.

The 2004 Study

The 2009 European Election Study ¹ is currently being conducted, and full results will be available by the end of the year. The analyses that follow are therefore all based, in total or in part, on the findings of the 2004 European Election Study. European Election Studies have been conducted from the very first direct election of the members of the European Parliament in 1979 onwards (with the one exception of 1984). All of them were ‘more’ than election studies in the narrow sense, that is: more than just a series of surveys among eligible voters. Some of these studies have also included one or more party elite surveys, be it among constituency candidates, members of the European Parliament (MEPs), or members of national parliament (MNPs); other studies have shared a particular focus on the campaign, either through participant observation or quantitative analyses of media content; last but not least, the election manifestos of political parties have been collected and content‐analysed, for all election years. The 2004 study specifically included a campaign study, a manifesto analysis, and a series of post‐election surveys in 24 of the then 25 member‐countries of the European Union.

The latter data collection exercise, the post‐election survey, was truly extraordinary on several accounts. To start with, our efforts to acquire central funding for this most expensive part of any election study failed. A research proposal was submitted to the European Commission under the 6th Framework Programme which, unfortunately, did not find the congenial reviewers that it undoubtedly would have deserved. In early 2004 we decided to try and conduct a ‘de‐central’ post‐election survey in as many member‐countries of the European Union as possible — de‐central in the sense that local study directors in each of the member‐countries would be in charge of fund raising, questionnaire translation, data collection and deposition, and reporting. Following the theoretical guidance of the unfortunate research proposal, a common questionnaire was drafted by the co‐ordination of the study at the Mannheimer Zentrum für Europäische Sozialforschung (MZES), and discussed and revised by a meeting of all local study directors convened at the MZES in Mannheim.

In early 2004, we would have considered it a big success if we were able to cover half of the member countries of the Union this way. It turned out that we were able to cover all of them — with the one exception of Malta were we could not identify somebody interested in this co‐operation. These local teams, ² numbering 24 in total, contributed in the shared understanding that if they were able to deliver their part, they would receive all other national studies in return plus, after a while, an integrated data‐file and codebook ³ as an indispensable facilitator of comparative analyses.

In addition to these surveys, the 2004 study contributed to and benefited from the research programme of CONNEX, a Network of Excellence for Research into multi‐level EU governance funded under the 6th Framework Programme of the European Commission. CONNEX facilitated data integration by paying a modest stipend to Mathew Loveless while he finished his dissertation and worked on the study at the MZES in Mannheim from mid 2004 to mid 2005. CONNEX was also extremely helpful in funding several of our authors’ meetings, not least the one in which most of the contributions that are now assembled here were presented as first drafts of their papers. ⁴

The Analyses

The objective of this study is to assess the effect of Eastern enlargement on various aspects of the input legitimacy of European Union politics. The contributions that are assembled here can be grouped in three sections: a first one dealing with East–West differences in citizens’ attitudes towards European integration and towards one another; a second with the EU party system and the election campaign after Eastern enlargement; and a third with electoral participation and vote choices in the 2004 election to the European Parliament. This last section is by far the largest of the three.

The section on European attitudes presents two analyses. Garry and Tilley deal with the question of whether EU support in the East is based on different sources than in the West. In particular, two hypotheses are derived and tested. One of them states that the economy is more important for Eastern EU support, while the other states that democracy is more important. Garry and Tilley’s analysis corroborates the first of these expectations while it refutes the second.

Scheuer and Schmitt ask whether the citizens of the enlarged European Union share a common political identity. EU enlargement is here not restricted to its latest round but conceived as a continuous process. The results of this analysis suggest that (a) yes, there is a sense of community among the citizens of the EU and (b) the farther away one gets from the core of the Union in both geographical and temporal terms, the weaker this sense of community becomes.

The section on the European party system again presents two analyses. Schmitt and Thomassen deal with the EU party system before and after enlargement. They investigate whether and how the process of Eastern enlargement has altered the EU party system. Comparing party placements on the two main dimensions of political contestation in the EU — the left–right dimension and the integration–independence dimension — it finds that Eastern enlargement did surprisingly little to the format of the EU party system and the stature of its political groups, both regarding their distinctiveness and their cohesion.

Van der Brug and Fennema in their contribution analyse whether support for radical right parties (both in the West and the East of the European Union) is caused by different factors than support for ‘other’ parties. Based on EES 2004 data, their main finding is that smallest‐distance voting according to the left–right logic is somewhat less important for supporting these parties as compared to others. As neither democracy nor EU nor government dissatisfaction plays an important role in explaining radical support for right parties, they also reject the protest‐vote hypothesis as a potential explanation of radical right party support.

The section on electoral participation and party choice presents the bulk of the contributions, five altogether. Franklin and Wessels examine the factors that cause low turnout in European Parliament elections. They propose that low turnout is a function of three deficits: a deficit of political community, a deficit of institutional effectiveness and a deficit of mobilisation. Together with the well known context characteristics of compulsory voting and the national electoral cycle, these three individual‐level predictors are found to explain turnout levels quite well. The main conclusion here is that post‐communist and Western EU citizens are following the same cues in much the same way.

Marsh, in his contribution, investigates the reasons for vote switching in the 2004 election to the European Parliament relative to the previous national election. He reviews and tests three major theories of vote switching in less important elections, and finds support for most of the predictions in most of them. What remains on the agenda, according to him, is the question of intra‐coalition switching, and the role that partisanship plays in all of this.

Clark and Rohrschneider analyse the same question with a focus on second‐order election models. They propose two oppositite hypotheses: a transfer hypothesis (voters transfer national political concerns to EP voting decisions) and a 1st‐order hypothesis (voters evaluate the EU on its own performance terms). Support for both models is found. In the narrow election context, the transfer hypothesis receives considerable support. In a broader perspective, however, voters clearly separate the two levels and evaluate each on its own terms.

Freire, Costa Lobo and Magalhaes in their contribution, finally demonstrate that survey data collected on the occasion of European Parliament elections are relevant for the study of electoral behaviour more generally. They examine whether the ideological location of citizens — in terms of left–right self‐placement — has a different impact on the vote in different types of democratic regime by controlling for three other factors hypothesized to make a difference: the permissiveness of the electoral system; the clarity of policy alternatives provided by the party system; and the particular type of party alignments. Their major finding is that the impact of left–right self‐placements is a stronger predictor of the vote the more polarized a party system is.

All in all, this collection of scholarly work that grew out of the 2004 European Election Study delineates the limited but visible impact that Eastern enlargement of the European Union had on aspects of the input legitimacy of EU politics. In terms of EU attitudes and support, the economy seems to play a different role in the East than in the West; in addition, it has been shown that mutual trust and a sense of community take time to develop and grow — a fact which will be leaving the ‘New East’ of the European Union behind for some time to come. The EU party system and EP election campaigns seem hardly altered by Eastern enlargement, and determinants of electoral participation and party choice are much the same here and there. It is just that the context factors that stimulate or facilitate the one or the other differ more or less systematically between Western Europe and the post‐communist East.

Notes

1. The 2009 EES (European Election Study) is funded under the 7th Framework Programme of the European Commission as a Design Infrastructure, and coordinated by the Robert Schuman Centre of the European University Institute in Florence.

2. The national study directors were: Günther Ogris (Austria), Marc Swyngedouw and Lieven de Winter (Belgium), James Tilley (Britain) and John Garry (Northern Ireland), Bambos Papageorgiou (Cyprus), Lukas Linek (Czech Republic), Jorgen Goul Andersen (Denmark), Alan Sikk and Vello Pettai (Estonia), Mikko Maatila and Tapio Raunio (Finland), Pascal Perrineau and Bruno Cautres (France), Hermann Schmitt and Andreas Wüst (Germany), Ilias Nikolakopoulos and Eftichia Teperoglou (Greece), Gabor Toka (Hungary), Michael Marsh (Ireland), Renato Mannheimer and Roberto Biorcio (Italy), Ilze Koroleva (Latvia), Algis Krupavicius (Lithuania), Patrick Dumont (Luxembourg), Cees van der Eijk (The Netherlands), Radoslaw Markowski (Poland), Pedro Magalhaes (Portugal), Olga Gyarfasova (Slovakia), Niko Tos (Slovenia), Juan Diez Nicolas (Spain), and Sören Holmberg (Sweden). For more information on the specifics of the 2004 surveys, see http://www.europeanelectionstudies.net.

3. In the spring of 2009, a second edition of this integrated data‐file and codebook was published which corrects a few inconsistencies present in the first edition and adds a number of constructed variables that might be useful in further analyses (Schmitt, Loveless et al. 2009 with an analysis of interview mode effects contributed by Till Weber. The data‐file and codebook is available from the EES homepage at http://www.europeanelectionstudies.net.

4. This meeting took place in late spring of 2006 at the Institute for Social Research of the University of Lisbon, Portugal. It was organised by Marina Costa Lobo which is most gratefully acknowledged.