THE EUROPEAN VOTERS STUDY 1989 by Cees van der Eijk ' Manfred Kuechler Herman Schmitl THEORETICAL PERSPECTIVES The European Voters Study (EVS) 1989 is a study of behavior, motivations, attitudes and perceptions of the electorates of the member states of the European Com- munity in the European Parliament Election of 1989— the third of its kind after 1979 and 1984. The objectives for designing and conducting a European Voters Study are twofold. First, a European Voters Study can be mainly looked at from the perspective of studying European elections and their place in the process of European integration. Second, more generally, it can be viewed from the perspective of comparative electoral research. The perspective of European integration . Protagonists of European integration have always showed great interest in the direct elections of the European Parliament which took place for the first time in 1979. Those who had lamented the slow pace of development of the European Community, hoped that a directly elected Parliament would provide a powerful stimulus to further integration. Unlike the other institutions of the Community, the Parliament would have its own popular mandate and would exemplify by its very existence the desire of the citizens of \he member states to live in a unified Europe. Some of these expectations reflected a certain degree of naivety with respect to the immediate political effects of these elections. Yet, the actual turnout disappointed not only the protagonists, but also startled more neutral observers. It was widely assumed that abstentions reflected a considerable degree of indifference or even opposition to the idea of European integration. No 'popular mandate' for further European integration could be inferred. In most countries the campaign was domi- nated by other, mosdy national political issues and concerns. The few exceptions to this general rule offered litde comfort from a pro-integration perspective: pre- dominandy in Denmark and to a lesser degree in Great Britain, party choice appeared to refiect a sizeable amount of anti-EC sentiment. The experience of 1979, reinforced in 1984, raised a number of questions con- cerning both turnout and party choice of European voters. Reliable answers were needed in order to prop- erly evaluate the implications for the future course of European integration. Does low turnout reflect just a widespread lack of familiarity with the European Parliament, is it just a visibility problem? Or does it reflect a more fundamental feeling that the European Parliament, and possibly the European Community at large, is irrelevant or detrimen- tal to the individual citizen's interests and concerns? Are those abstaining from the European elections decidedly critical about, ot even downright hostile towards Euro- pean integration in general and towards the European Parliament in particular? What part do the political parties play? Are they unable or unwilling to put Europe on the national agenda, to channel and represent the EC related interests of their clienteles? To which extent, then, is the voters' choice between the parties an ac- knowledgment of specific party goals with respect to European integration? Does party choice reflect different EC policy preferences or is it predominanUy determined by domestic considerations? Obviously, contingent upon the answers to these questions, very different conclusions concerning the future course of European integration can be drawn. For most, if not all of these questions, survey data representative of the electorate at large are necessary to obtain answers solidly grounded in empirical evi- dence. The perspective of comparative electoral research . The study of elections and individual voting behavior is a very well developed area of empirical political science. In virtually all western democracies large scale surveys are conducted during election times to uncover the forces which shape voting behavior and thereby election results. However, there is considerable national variation in die depth (over time), quality, and accessibility of these data. The United States, Great Britain, and West Germany have long standing traditions of scholarly election surveys which are generally available for secondary analysis. The situation in a number of otiier countries is less fortunate. Still, a number of valuable attempts have been made to utilize national election studies from various countries for cross-national comparisons (see e.g. Budge, Crewe, and Fairlie 1976; Crewe and Denver 1985; Dalton, Flanagan, and Beck 1984). On the one hand, the volumes which document these efforts exhibit the strong common strands in the design and conceptuali- zation of the various election studies. Yet, on die other hand, they clearly reveal die discrepancies between tiiem. Summer 1990 National election studies are indeed strongly national in character. To some degree this is unavoidable. The diversity reflects real differences with respect to systemic arrangements (e.g. electoral rules) and political culture. But this diversity is also due to (false) economy: ques- tions which have litde explanatory value in a strictly national study, but which are essential to establish comparability with other countries are the first ones to be cut if such questions are considered at all. Incompatibili- ties in overall research design, in choice of concepts, in manner of operationalization, in question wording and format, and — last but not least— in the demographics section are likely to continue for the noble cause of preserving national comparability over time. The situation, then, is somewhat paradoxical: while the field of electoral research is among the oldest, and certainly most developed areas of empirical social research, it has not generated the kind of large scale cross-national survey projects which have been so pivotal in the development of other areas of comparative mass political behavior (see e.g. Almond and Verba 1963; Barnes and Kaase 1979). background with the purpose of designing and organizing a truly comparative Eiu^opean voters study to be con- ducted in 1989. Subsequent meetings were held in Mannheim in May and October 1987, which resulted in the formation of a group of six scholars serving as co- principal investigators: Roland Cayrol, Cees van der Eijk, Mark Franklin, Manfred Kuechler, Renato Mannheimer, and Hermann SchmitL Though not a formal member of the group, Karlheinz Reif was essen- tial to the success of the project in providing good scholarly as well as very practical advice from the very beginning. Most members of the core group were intimately involved in earlier studies of the European election. Following precedence, cooperation was (re- )established with other research teams focusing on the campaign (coordinated by Oskar Niedermayer, at the University of Mannheim, West Germany) and on the communication process (see e.g. Blumler 1983). During the two intensive meetings in Mannheim, the group hammered out a design of the European Voters Study to be, drew up a strategy for securing funding, and decided on some division of labor. PREVIOUS WORK In the past, the Eurobarometer surveys have been utilized in various ways to generate data related to the process of European integration. Questions concerning electoral participation have been included in the surveys prior to and following the European Elections of 1979 and 1984. Questions relating to affective and evaluative orientations towards European integration, the European Community, and its various institutions and policies have been included frequently in Eurobarometer surveys and constitute an important pan of the 'trend' questions which are included in each wave. Still, in spite of the wealth of material which has been collected, a number of important lacunae remain. These originate partly from the fact that certain questions were never included (e.g. questions assessing factual knowledge), and partly from the fact that the regular Eurobarometer surveys take place too far before (March), and too late after (November) the point in time at which the European elections actually take place (June). Likewise, various slu^'eys conducted at the occasion of previous European elections do not fill this void. They have focused on media effects and on various kinds of elites including party candidates running for seats in the European Parliament (see e.g. Blumler 1983; Reif 1984, 1985; Reif and Schmitt 1980), but they did not center on the voting behavior of the electorate at large. INTERNAL ORGANIZATION AND COOPERATION During the ECPR Joint Workshops of April 1987, first contacts were established between scholars of various In terms of internal organization two factors were essential. First, most valuable support was provided by the University of Mannheim which made it possible for Hermann Schmitt to serve as the coordinator for the group. Second, ample use of electronic communication via EARN/BITNET compensated for the very limited opportunities for personal meetings of the entire group. Geographical dispersion of its member and the lack of sufficient travel funds could not have overcome other- wise. Not just with respect to travel, funding was a major problem continuously haunting the group. Funds were secured from various sources, in various amounts, at different points in time. A major portion, covering the costs of the field work for the post-election wave, was supplied by the British Economic and Social Research Council (ESRQ. Other funding sources include several national newspapers which were given priority publica- tion rights of elementary, but timely analyses of part of the data. Unfortunately, we could quite meet our funding objectives. This required several cuts and modifications in our original question program. In particular, some questions could not be replicated in all three waves as planned. Still, the core of the original plan could be carried out. A series of questions were added to the core questionnaires of the Eurobarometer surveys #30 (November 1988), #3 1 (April 1989) and #31 A (June 1989). Matter of fact, the close cooperation with the Eurobarometer proved to be an indispensable asset. Without it the study could not have been completed. It gave us — and the hopefully 10 lASSIST Quarteriy many more researchers to come— access to the standard Eurobarometer questions and with the special edition of July 1989 (#31A) it provided a base for the post-election wave. DESIGN AND CONTENTS With our theoretical focus on mass behavior, there was no alternative to a cross-national survey design. In addition, we felt that a purely cross- sectional design would be inadequate (though much more feasible) in order to study the process of cognitive, attitudinal and behavioral mobilization. The choice, then, was between a genuine panel design and a series of repeated cross- sections. Without entering the sometimes vivid debate on the advantages and disadvantages of panels in contrast to repeated cross-sections, we quickly determined that a panel design was not fundable; that only buying into an established European siu^fey like the Eurobarometer would bring cost for data collection within a feasible range. While not denying these very practical concerns, the repeated cross- sections design does match our theoreti- caJ and conceptual interests. Our emphasis was not on the dynamics of individual vote choice but on the preferences of groups and segments of voters, on the change of these group preferences over time, and on patterns of association. Below, we will briefly outline the sets of variables included in the study. In terms of our prime target, turnout and party choice in the European Elections 1989, a broader set of questions needed to be included. Previ- ous research had convincingly suggested that electoral behavior in European elections is to a large extent determined by national factors. Consequently, intended national electoral behavior was to be probed as well. Furthermore, drawing on theories on voter behavior and party competition developed in the context of the Dutch national election studies (see e.g. van der Eijk and Niemoeller 1983), a more comprehensive assessment of the electoral attractiveness of all major parties was called for— with respect to both European and national elections. Explanatory or independent variables fall into five categories. The Ftfst category consists of variables which describe the voters' social situation; in particular, their location within the cleavage structure of each country. These are necessary for explaining behavior in terms of the tradi- tional cleavage theories. These theories have come under attack in recent years, but the scholarly debate over the persistence of established cleavages is not over yet. Also, these variables are needed as controls in assessing the effects of attitudes, perceptions, experiences, and general political behavior on turnout and vote choice. These variables do not attract much attention in national studies, they are mostly part of an estabUshed demo- graphic section. However, for a cross-national study they constitute a major problem. To deal with the pervasive problems of (in)comparability which traditionally plague researchers working with these characteristics, we drew on the ongoing work of another group (Franklin, Mackie, and Valen, 1990). With a few additions, the set of demographic variables used in the Eurobarometer met our needs. A second block of independent variables deals with substantive issue concerns. Obviously, to the extent that issues play a role in voters' decision-making, they may arise from different contexts. At the least, the following kinds of issues have to be distinguished: a. Community issues (extending EC membership, common agricultural policy, payments to and subsidies received from EC, etc.), b. supra-national issues (issues pertaining to all member states but not, or only partly related to the EC like defense, unemployment, etc.), and c. country-specific issues (the most salient of these were determined in close cooperation with additional country specialists). It is desirable to tap absolute and relative saliency as well as perceptions of party competence for each one of such issues, but this would require an inordinate amount of question time. As a compromise, we constructed a list of 12 issues (4 each of the three types mentioned above). Each item was individually rated as 'very important' or 'not very important', then the respondent was asked to name the three most important ones. For these (up to) three issues, we further established which party was seen as best able to handle this problem. Funding problems restricted the full approach to the second wave, while individual salience ratings were obtained in all three waves. The third block of variables comprises European orienta- tions, which deal with the European Community, its institutions, the idea of European integration, etc. Many of the indicators which are regularly included in the Eurobarometer questionnaires capture the affective components of such attitudes. In addition, we also tapped the cognitive and evaluative aspects of European orientations. A fourth block of explanatory variables deals with specific perceptions of the political parties contesting European and national elections. One set of such perceptions deals with the parties' position on Europe, Summer 1990 another with perceptions of the parties' location on a Left-Right scale. Additional questions establish the respondents' own location or preference. A fifth and final block of questions deals with media exposure and information. Here, we closely cooperated with another project focused on the communication process in the electoral campaign (see above) and followed their lead. Most of these questions were replicated from the 1979 Communications Study (see Blumler 1983). Apart from some cuts within these five sets of questions due to funding problems, other aspects originally dis- cussed had to be shelved altogether. These include questions dealing with possible candidate effects on party choice. No attempt was made to measure the elusive concept of party identification beyond the standard item in the Eurobarometer questionnaire. However, the battery of questions in which the electoral attractiveness of all pardes is to be rated (see above) offers new options to construct possibly more valid operationalizations of this concept STRATEGIES FOR ANALYSIS AND PUBLICATIONS A number of initial analyses on the data from the first wave (EB30) have been presented and discussed in an ECPR workshop during the Joint Sessions in Paris in April 1989. Special panels at the Annual Meetings of the Midwest Political Science Association (Chicago, April 1990) and the American Political Science Association (San Francisco, August 1990) have and will provide other opporiunities to present and discuss findings from this study. A special issue of the European Journal of Political Research (planned for the second half of 1990) will contain a first set of cross-national comparative analyses by members of the core group. This will be followed by an edited volume with chapters on each of the EC member countries to which additional country specialists will contribute. It will also contain a second round of comparative analyses. To conclude this presentation, we will briefly discuss the general analytic strategy behind these pubhcation plans. At the same time, this discussion may also further productive use of this data base by other researchers in the future. As argued in more detail elsewhere (Kuechler 1987), mass survey data provide an invaluable, but also inher- ently limited base for the study of mass (political) attitudes and behavior. In general, survey data do not just speak for themselves, they require a careful interpre- tation within the context in which they are generated. This holds for any (national) survey, but it becomes even more apparent in a cross-national setting. A question may have a different meaning in a different political and cultural system, even when great care is exercised in aiming at 'functional equivalence'. A comparison of marginal distributions across nations has some heuristic value, but it does not lead to meaningful theory construc- tion. It is more useful to look for patterns of associa- tions, e.g. the impact of degree of jrolitical interest on issue evaluations, and to compare on the level of these relationship patterns. In a way, we can look at such an analysis as an instantaneous eleven-fold replication of a relational hypothesis. Our first round of analyses has produced few, if any hypotheses which can be success- fully replicated this way. Matter of fact, particular in the area of issue voting, we have come across a surprising number of sign reversals, i.e. the same two variables show a positive relationship in some countries and a negative one in others. This strongly points to the need to assess the survey data in the light of other country- specific sources of information. Detailed country- specific analyses (the second stage in our analytic strategy) then go way beyond mere idiosyncratic descrip- tion. Their prime objective is a "cross-nationally in- formed country-specific" analyses which will focus on singular and deviating patterns. In turn, these will provide the base for a second, higher level of compari- son. At this point it is premature to predict the possible returns from this three stage comparative strategy. We may find a considerable amount of higher level commu- nality, or we may conclude that idiosyncratic systemic factors tend to dominate, severely curtailing efforts of location-independent theory building. But even if our group fails, a valuable host of data will be available to the other researchers with all sorts of brilliant ideas in the very near future. The social science community is fortunate to have the services of many fine data archives available. Their supporting role is vital for the further growth of the social sciences. REFERENCES Almond, G. and S. Verba (1963). The Civic Culture. Political Attitudes and Democracy in Five Nations. Princeton, NJ: Princeton University Press. Barnes, S., M. Kaase et.al. (1979). Political Action: Mass Participation in Five Western Democra- cies. Beverly Hills, CA: Sage. Blumler, D. (ed.) (1983). Communicating to Voters. Television in the First European Parliamen- tary Elections. London: Sage. Budge, I., I. Crewe and D. Farlie (eds.) (1976) 12 lASSIST Quarteriy Party Identification and beyond: Representation of voting and Party Competition. New York: Wiley. Crewe, I. and D. Denver (eds.) (1985). Electoral Change in Western Democracies: Patterns and Sources of Electoral Volatility. London: Croom Helm. Dalton, R. J., S. C. Flanagan and P.A. Beck (eds.) (1984) Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment? Prince- ton, NJ: Princeton University Press. Eijk, C. van der and B. Niemoeller (1983). Elec- toral Change in the Netherlands. Empirical Results and Methods of Measurement. Amsterdam: CT-press. Franklin, M., T. Mackie and H. Valen (eds.) (1990, forthcoming) Electoral Change: Responses to Evolv- ing Social and Attitudinal Structures in Seventeen De- mocracies. Cambridge: Cambridge University Press. Kuechler, M. (1987). The UtiUty of Surveys for Cross-National Research. Social Science Research 16: 229-244. Reif, K. (ed.) (1984) European Elections 1979/81 and 1984. Berlin: Quorum. Reif, K. (ed.) (1985) Ten European Elections: Campaigns and Results of the 1979/1981 First Direct Election to the European Parliament. Aldershot, UK: Gower. Reif, K. and H. Schmitt (1980) Nine Second Order Elections, European Journal of Political Research 8: 3-44 'Paper prepared for lASSIST 90, Annual Conference and Workshops, May 30 - June 2, 1990, Poughkeepsie, New York. Cees van der Eijk Department of Political Science, University of Amsterdam, Netherlands Manfred Kuechler Hunter College and Graduate Center, Department of Sociology (CUNY) Herman Schmitt ZEUS, University of Mannheim, West GermanyYork. Principal co-investigators (in alphabetical order): Roland CAYROL (France), Cees van der EUK (Netherlands), Mark FRANKLIN (United Kingdom/USA), Manfred KUECHLER (West Germany/USA), Renato MANNHEIMER (Italy), Hermann SCHMITT (West Germany). Coordinator: Hermann Schmitt BITNET addresses : van der Eijk (A7150204HASARAll), Franklin (POLSlJL@UHUPVMl) Kuechler (MAKHC@CUNYVM), Schmitt (FS91@DMARUM8) Summer 1990